From New Media Business Blog

Jump to: navigation, search


Salesforce Overview

Cloud computing

In its simplest form, cloud computing allows for computing services to be delivered to users over the internet. Cloud computing also provides remote access to resources. For most businesses, cloud computing is very attractive because it eliminates the costs of buying hardware, allows for instant access to resources, eliminates the need for server maintenance and can be scaled easily in accordance with business need.

Cloud services come in three forms: infrastructure as a service (IaaS), platform as a service (PaaS) and software as a service (SaaS). Infrastructure as a service refers to users renting computing infrastructure such as servers and operating systems from a provider through the cloud. Platform as a service refers to the renting of a development environment on the cloud that can be used to create custom web and mobile applications. Software as a service refers to paying a subscription fee to access and utilize software that is delivered on the cloud. [1]

What is Salesforce?

Salesforce is a cloud computing company best known for its Customer Relationship Management (CRM) platform. CRM can be defined as "practices, strategies and technologies that companies use to manage and analyze customer interactions and data throughout the customer lifecycle, with the goal of improving customer service relationships and assisting in customer retention and driving sales growth"[2]. The Salesforce platform consists of various products and services offered completely via the cloud. Initially used only as a solution to manage sales within organizations, the platform has grown to cater to all areas of Customer Relationship Management, including customer service and marketing. The platform is also widely used to build custom business applications that can be implemented and utilized to achieve better business performance.

History of Salesforce

First Office at Signal Hill source

Software as a service

In 1999, Salesforce was born out of a one-bedroom apartment in Telegraph Hill, San Fransisco. The company was founded by Marc Benioff and three other employees who included Parker Harris, Frank Dominguez, and Dave Moellenhoff. The goal was to deliver sales software through a software as a service model, a model in which businesses could access software through a website. Through doing this, businesses would eliminate the hassles and costs of owning as well as maintaining their own IT infrastructure. The idea took off and in 2003, Salesforce launched the first Dreamforce, a yearly conference in which users and creators of Salesforce come together to discuss product improvements, upgrades and share knowledge. In 2005, Salesforce launched the AppExchange, an online marketplace in which solutions developed with Salesforce could be opened up or sold to the wider Salesforce community.

Platform as a service

In 2006, Salesforce introduced its own programming language called APEX as well as VisualForce, a technology which allows developers to create custom user interfaces for applications. Apex and VisualForce together were branded, and through this, Salesforce took the first step in extending its software as a service model to a platform as a service model. Users could now create their own custom applications on the platform. In 2012, Salesforce acquired ExactTarget, a media marketing tool, and expanded its platform to support marketing applications. This was a major milestone as it was the first step in making Salesforce a more comprehensive solution that could address all areas of CRM. Salesforce also expanded its platform to mobile devices in 2013. The Salesforce1 platform was built especially for mobile devices and allowed, for the first time ever, users to access the same information from a mobile device that they could from a computer. This allowed users to access their Salesforce applications from anywhere with any device.

Artificial Intelligence

In 2015, Salesforce launched Lightning Experience, a dynamic new user interface that made building custom applications more user-friendly. Lightning Experience also encouraged a more component-orientated way of developing in which existing functional system components could be integrated and reused to create full applications, saving time as well as developer costs. Lightning Experience also provided the foundation for building smarter applications that incorporated the power of artificial intelligence (AI). AI was introduced to the platform in 2016 with the launch of Salesforce Einstein, a set of AI-powered functionalities that could be integrated into applications for better performance. Business users could now get on demand predictions, detailed analytics, and recommendations with regards to their customers that made them more competitive. Salesforce notices that AI drives the future of CRM and as such plans to greatly expand and improve its use as part of its platform within the next few years[3].

Common Misconceptions

"Salesforce is just for Sales"

While Salesforce was initially used to manage sales in organizations, it has since moved beyond this to support all aspects of CRM [4]. Custom applications can also be created that are catered towards other areas of business outside of CRM. In addition, Salesforce has recently expanded its platform to support Internet of Things (IoT) applications.

"Salesforce solves all data problems"

Salesforce can be really useful for a business if the data that's inserted into the system is of good quality. If the data that is entered into Salesforce is redundant, insufficient or inadequate, the data generated from Salesforce to aid in decision-making will not be effective. Therefore, businesses using Salesforce still need a standardized approach for entering, updating and removing data so that good data quality is upheld.

"Salesforce could be implemented without customization"

While Salesforce can be implemented without customizations, this is rarely the case. Every company has its own set of unique features and processes. No pre-built software will be able to work for all companies right out of the box. Additional customizations are required in order to adapt the software to a business. As such, there are implementation partners that specialize in making these customizations.

Salesforce CRM Market Share

CRM Market Shares source

Since the introduction of the platform to the CRM market, Salesforce has seen a steep growth in market share.[5] Currently, Salesforce has surpassed both Oracle and SAP to be the largest CRM provider in the market.The latest figures show a healthy 19.6% as compared to Oracle and SAP, who are at 7.1% and 6.5% respectively. Both Oracle and SAP have seen declining market share over the past few years with Oracle regaining some of its market share in 2017 and SAP continuing to lose more ground[6]. Other Salesforce competitors include Adobe and Microsoft, both small players in the CRM market with market shares of 3.2% and 4% respectively.

Reasons for Success

Nearly Endless Customizability

Salesforce provides a platform that is designed to support configuration and customization. This means that users are able to build onto out of the box functionality to create applications that are a near perfect fit for their business process. By using APEX code and VisualForce, users have the ability to fully customize their Salesforce applications to fit their exact business specifications. In today's competitive business environment, this makes Salesforce very attractive to many companies looking for specific solutions to unique problems that cannot be easily solved with other existing software. Salesforce can also be integrated with existing business systems, such as payment systems, for more effective business performance.

In addition to being highly customizable, Salesforce makes it easier for non-technical users to configure the platform to meet their needs. Users are able to create semi-customized applications without the use of the code. However, instead of code, one can utilize click and drag, input formulas, workflow rules, and other settings to create applications that, while not fully customized to your business process, are still relatively specific to your needs.

Acquisitions and Innovation

Salesforce has been able to expand very quickly into different areas within CRM through acquisitions. Acquisitions have allowed Salesforce to add more functionality and features to its platform in a relatively short period of time as well as helped expand into areas like the Internet of Things and artificial intelligence. In addition, through acquiring market leaders, Salesforce has been able to gain from vast amounts of industry knowledge and innovation that has given it an edge over its competitors. Most notable acquisitions have included Mulesoft, recently acquired for 6.5 billion dollars, and ExactTarget, acquired for 2.5 billion dollars in 2013. Mulesoft has allowed Salesforce to make its platform more effective in integrating with older legacy systems and ExactTarget formed the basis for new marketing offerings[7]

Salesforce has also invested heavily in research and development. The company spends around 15% of its gross profit on developing new offerings [8]and increased spending in this area has allowed for new developments such as Salesforce Einstein. These offerings have differentiated Salesforce from its competitors and allowed it to be first in expanding CRM solutions to include functionalities like artificial intelligence.

Increased Demand for Cloud-based Solutions

Salesforce has benefited from a general trend towards the adoption of cloud computing. More business are seeing the value in moving their data and systems to the cloud and this increase in demand has most definitely helped Salesforce grow. Besides the fact that moving to the cloud presents significant cost reduction for many businesses, the change in the way people work has also driven the need for cloud-based solutions. Many people now have flexible work schedules and need to access data from outside of the office from a variety of devices. In addition, having cloud-based systems has allowed businesses to outsource areas such as infrustructure maintenance that has traditionally been done in-house and has often been a significant pain point. Issues such as security were initially a major concern but with many vendors all offering cloud-based solutions and sophisticated security measures now in place, this concern has largely abated.

Salesforce Future Objectives

Expand AI Functionality

Salesforce founder Marc Benioff sees Salesforce Einstein, Salesforce's AI platform, as the future of the company. As such, Salesforce plans to continue to actively improve Einstein features and capabilities over the next few years. According to Forbes, the central vision is to create CRM applications that are able to learn from available data and be proactive in predicting customer needs[9]. Another goal is to enable development on top of Einstein infused applications. Through this, developers will be able to create custom AI solutions that are specific to their business.

Enter New Business Segments

Salesforce has committed to expanding its platform to include other business segments such as the financial services industry [10]. Salesforce recently launched its new Financial Services Cloud, allowing wealth management providers to build applications that can help manage the adviser-client relationship. Salesforce has committed itself to improving these new product offerings in the next few years and has also announced its willingness to expand its platform to other areas outside of traditional CRM.

Be 100% Sustainable

Salesforce is working towards using 100% renewable electricity resources across its global operations [11]. To work towards this, Salesforce recently announced two wind power projects to generate a combined 225, 000 megawatts of power per year in Texas and West Virginia. This power generation is likely to offset some of Salesforce's future electricity use which goes towards powering servers and large amounts of IT equipment. This move is part of a larger desire to be fully sustainable and maintain zero carbon emissions so as to decrease the company's carbon footprint.

The Salesforce Economy

With its rise as the dominant CRM provider in the market, Salesforce stands to have a major impact in new job growth, as seen below:

Salesforce Economy source[12]

User Issues with Salesforce

While Salesforce has been widely adopted, the following are three common issues identified by users:

Steep Learning Curve

While you can use Salesforce to create fully customized applications for your business, learning how to do this can be a challenge. For many users new to Salesforce, this learning curve can be quite steep. As such, businesses commonly hire administrators with knowledge of the platform or approach Salesforce implementation partners, companies that specialize in the platform, to take care of customization. This can be very costly and can greatly increase the investment needed to implement Salesforce for a business. Salesforce has launched its own training site called Trailhead that users can utilize to learn the platform, but the reality is that many business users typically do not have the time it takes to completely educate themselves on the platform. As such, having significant knowledge of the platform to make the required customizations remains a stumbling block for many.

Limited End to End Capabilities

A major shortcoming of Salesforce is that it is not an enterprise-wide solution. You cannot use Salesforce as an end to end solution and likely will have to integrate Salesforce with your existing systems [13]. In simpler terms, Salesforce can only be used to manage your CRM but you cannot use it to manage the entirety of your value chain. The major implication of this is that businesses then don't have a common data model across systems which makes it harder to retrieve and analyze data from the full scope of their operations. This also means that they could struggle to gain new insights from different systems that could make them more efficient and effective.

Hidden Costs

While on the surface Salesforce may look like a good deal, many users complain about hidden costs that end up making the investment more expensive than what they initially thought. For example, a company may decide to start its sales team with a basic Salesforce package of around $75 per user per month. However, many executives then realize that customization is necessary to make Salesforce more specific to their business process. Since the company doesn't have the knowledge necessary to make these customizations themselves, they contract a Salesforce implementation partner who can charge up to $200 per hour [14]. In addition, the company may soon also purchase upgrades to features they realize that they need. Salesforce offers such a variety of different products and services that it is relatively easy for up selling to occur. As such, the company may end up paying triple the initial amount expected to make its Salesforce application more effective and more usable for its sales staff.

Platform Architecture

Architecture Overview

Salesforce Platform source[15]

The Salesforce platform can best be described as multiple layers of technology built on top of each other. At the very heart of the platform is the multi-tenant infrastructure which allows users to have their own separate space in the cloud. Powering the platform is a strong database service known as Thunder. Thunder ensures that large amounts of data can be collected and analyzed which in turn supports Salesforce Einstein, the platform's artificial intelligence layer. On top of this layer sits various development tools, such as Lightning Experience and the AppExchange, which support the customization of the different Salesforce Clouds; standard modules and product offerings that support applications geared towards a particular business function.

Multi-tenant Infrastructure

Multi-tenancy allows a set of users to use a common cloud-based software application while maintaining their own personal space on a cloud server. [16] You can think of the infrastructure like a bus. Everyone is on the same bus, the bus is the server. You don't own the bus, likewise, you don't own the server. One person has a seat, showing the isolation of the different user instances, but you share the same seat types as everyone else. This is the same as the application of the servers. If you were to renovate the interior of the bus, there would be new features, this would be like if you were to get a new update on the servers.

Salesforce Lightning Experience

The Salesforce Lightning Experience user interface was developed in response to customer complaints that the older Salesforce Classic interface was a pain to work with. The old Classic interface was a major obstacle for many users as it was very text heavy and visually unappealing. Salesforce Lightning Experience addressed these concerns by using more visual elements to display information as well as a better use of overall screen space for easier navigation. While the older user interface was fairly static, Lightning is more dynamic and allows for a changing user interface that is more exciting to use. A visual comparison of the two interfaces can be viewed in this article [17].

The Lightning Experience user interface also becomes a useful development tool through the Lightning App Builder. The Lightning App Builder allows users to create applications by combining components through click and drag functionality. Components can be developed in-house or downloaded from the AppExchange. This makes it easier for non-technical users to create custom applications and plays to Salesforce's vision of making Salesforce a viable solution for all types of users. For a demo of how the Lightning App Builder works, view this Lightning App Builder demo[18].

A current shortcoming of the Lightning Experience interface is that many of the features have not yet been migrated over from the old Classic interface [19]. This means that to access certain features, users need to switch back and forth from Lightning to Classic. While this is not a big issue and Salesforce is working to solve it, it needs to be considered when developing in Lightning Experience.

Salesforce AppExchange

The Salesforce AppExchange is an online marketplace in which Salesforce users can download and buy applications to customize as well as extend the functionality of their existing Salesforce applications. Users can also download completely new applications for their business. Like the Google Play store, the AppExchange has already seen millions of installs and thousands of customer reviews. The applications available on the AppExchange are not limited to CRM and users are able to download custom applications for different industries that have been built using the Salesforce platform [20]. Applications can be downloaded for free or can be bought for a fee.

The AppExchange is commonly used by consulting partners and individual developers to sell what they have developed. For Salesforce users, the AppExchange is a great way to get custom applications specific to their industry for a fraction of the cost it would take to develop it themselves. Some common applications that have been downloaded off the ApExchange include DocuSign and Conga Composer. Docusign allows users to sign contracts and other documents digitally and Conga Composer allows users to create documents as well as other media from their Salesforce data. For a visual demo of the AppExchange, have a look at this Salesforce AppExchange Demo[21].

Salesforce Clouds

Salesforce offers modularized components for its CRM to provide customers the ability to only purchase solutions that fit their needs. These components are called Clouds and they cover multiple areas of a business. This following section briefly explains what each of the Clouds does. For business applications of these Clouds, refer to the Salesforce Success Stories section.

Sales Cloud

Sales Cloud supports sales applications that help salespeople manage their relationship with new and existing clients. Sales Cloud helps salespeople qualify leads, identify contacts, manage opportunities and maintain accounts. Sales managers are also able to generate and analyze sales reports as well as dashboards that indicate performance metrics. Other features included in Sales Cloud are basic sales forecasting and process automation; the ability to automate certain functions such as email sending based on set criteria.

Service Cloud

Service Cloud supports customer service applications that help agents deal with customer problems, complaints, and requests. With applications built in Service Cloud, agents can live chat, screen share, share articles with customers on how to solve problems and manage service cases [22]. Commonly used for call centers, Service Cloud applications are also able to automate service processes and automatically route incoming cases to the appropriate agent.

Marketing Cloud

Marketing Cloud supports marketing applications that can be used to implement digital marketing campaigns through various channels [23]. With applications built in marketing cloud, users can segment as well as administer campaigns, build content, track campaign success and manage social media. Marketing Cloud also supports the Journey Builder, a feature that allows you to create a personalized buying journey for each customer.

Community Cloud

Community Cloud supports applications that increase collaboration between groups of users [24]. For example, companies can build community applications that allow customers to share information, search various topics and interact with agents from the business. Some of the features included in Community Cloud are document and image sharing, live chat, and community success dashboards.

Analytics Cloud

Analytics Cloud allows Salesforce users to build business intelligence applications that can be used to analyze data generated from applications in other Salesforce clouds eg: Sales Cloud or Service Cloud. With Analytics Cloud users can analyze as well as visualize data, create reports and see trends relevant to their business. Analytics Cloud also supports drilling into as well as across data dimensions so that different questions can be answered from the available data.

Apps Cloud

Apps Cloud was developed by Salesforce to provide an all-in-one development experience for users wanting to build their own custom applications on the platform. Apps Cloud brings together custom development tools such as (APEX and VisualForce) and Heroku. Heroku is a developer platform built into Apps Cloud that allows users to run and deploy applications that are not written in the APEX code native to the Salesforce platform [25]. In this way, users are able to integrate add-ons to applications developed in Apps cloud that are written in languages like Java or Ruby. Apps Cloud also integrates the AppExchange into the developer experience so that users can purchase or download add-ons for their applications. In addition, Apps Cloud provides a direct link to the official Trailhead learning site so that users can further increase their development knowledge.

Commerce Cloud

With Commerce Cloud, users are able to create e-commerce stores on their websites[26]. Commerce Cloud can be easily integrated with other Salesforce Clouds and is also developed to be easily accessible through all devices. With Commerce Cloud, businesses are able to engage shoppers with attention-grabbing promotions and make the shopping experience more dynamic for customers through offering product suggestions based on previous selections.

IoT Cloud

Salesforce Internet of Things (IoT) Cloud allows users to create applications that collect and store internet of things data. IoT applications can be used to analyze large amounts of data that is collected from devices, sensors, websites, customers, and partners. Applications built in IoT Cloud can also be used to direct actions in real-time through workflows. An example, as given in the referenced text, is re-booking passengers onto new planes when their connecting flight has been delayed while they are still in the air[27] . IoT Cloud Applications can also be integrated with other Salesforce Clouds so that it can be used within different business functions such as service, marketing or sales.

Salesforce Einstein

Salesforce Einstein is an artificial intelligence (AI) technology, that incorporates machine learning, and language processing, designed for the Salesforce platform. Salesforce Einstein is able to predict, recommend and provide insights based on existing data. Einstein now makes over one billion AI-driven predictions a day [28].

Thunder and Open Source Technologies

Thunder's 4 key open source technologies [29]

Powering Salesforce Einstein is a database and rules engine called Thunder. First introduced in 2016, Thunder is optimized for storing, processing, and reacting to the large volume of data that Einstein requires to operate[30]. In order to handle big data efficiently and effectively, the database employs four proven open source technologies: Apache Spark, Apache Kafka, Apache Storm, and Cassandra.

The following article[31] provides a good overview of what open source technologies are. According to Adam Bosworth, Executive Vice President at Salesforce, using open source technologies allows Thunder to take advantage of existing research and development done by the public. Three out of the four main technologies are from the Apache foundation, which is the largest non-profit organization dedicated to providing open source software to the public [32]. As such, Salesforce can reduce the development cost and time of Thunder, as well as ensure that the database is secure and extensively tested. This practice is commonly used by major technology companies, such as Google and Facebook.

Apache Spark

Apache Spark (Spark) is a distributed computing framework that specializes in processing big data in batches and real-time. In layman’s term, Spark divides the workload to multiple computers in order to compute large volumes of data efficiently. Once all computers have processed the data, all results are then combined through Spark.

According to the Apache Software Foundation, Spark stores and processes data 100 times faster than disk-based processing frameworks used by common databases, such as Microsoft SQL Server. This is because disk-based frameworks store data in disks. Since writing and reading data in disks are slow, these frameworks are not optimized for big data purposes, which require fast read and write speed for a large amount of data. On the other hand, Spark uses in-memory technologies, which store data in the computer’s random access memory (RAM). Accessing data stored in RAM is much faster compared to accessing data on disks. As such, Spark can perform operations on this data faster, which makes it more suitable for big data processing. [33]

Apache Kafka

Kafka facilitates communication between devices [34]

Apache Kafka (Kafka) is a distributed data streaming platform that handles multiple data streams at the same time. Traditionally, data is replicated between systems in batches, which delays user response time. With data streaming, data is constantly being transferred from one device to the other without delaying replication and user response times. As such, Kafka can facilitate real-time communications between multiple devices and applications. The main functionality includes publishing and subscribing to data streams [35]. The following video [36] explains two main use cases of Kafka.

Kafka's main functionality behaves similarly to a messaging system. Kafka can take data produced by a sensor and create a constant stream of that data for other devices to consume [37]. This is similar to creating and publishing a public group chat. Messages in the chat are constantly being updated. As with any chat, multiple users can send messages at the same time. Kafka is able to process multiple data sources at the same time. Other devices can use Kafka to subscribe to published data streams and retrieve data from the stream. With a chat, users can join the chat and read previous and new messages. This function is essential for managing data streams from Internet of Things (IoT) devices and transactional systems, which generate large volumes of data in real-time. The diagram on the right depicts Kafka's main functionality as it interacts with data streams and other devices.

Apache Storm

Originally owned by Twitter and later donated to the Apache Software Foundation, Apache Storm (Storm) is a distributed computing system that processes streams of data in real time[38]. As mentioned in the section toward [Apache Spark], a distributed system divides work between multiple machines. Users can define rules on what to do with data as they come in through streams. These rules include transforming data and automating processes based on the data. An example rule is to transform data from the thermostat to Celcius, and if it exceeds 10 Celcius, send a notification to the maintenance team.

In operation, Storm uses two main components: nodes and pipelines connecting nodes together[39]. The nodes represent data streams (i.e. from sensors) and devices on a connected network. Pipelines represent data transformations that happen between each node. They act as data transformation pipelines that apply user-defined rules on the data as it comes in. The transformed data is passed on to the next node, where users can further define what to do with it.

Apache Cassandra

Originally created by Facebook and later donated to the Apache Software Foundation, Apache Cassandra (Cassandra) is a decentralized database with horizontal scalability. Cassandra is used widely by large technology companies, including Netflix, Facebook, and Google [40]. Its top use cases include Internet of things, product catalogs, and social media analytics[41]. With a decentralized database, data is stored on multiple servers called nodes. Nodes can be grouped together to create clusters of servers. Each node in the same cluster contains the same data. As such, there is no single point of failure in the architecture. If one node goes down, other nodes can take over without any downtime. Traditional databases store data on one centralized server. If this server goes down, assuming there is no backup server, the entire information system will go down as well. As for horizontal scalability, Cassandra allows for additional servers to be easily added to the system to increase processing power. This ability makes Cassandra a highly scalable database. The following sections will describe Cassandra's NoSQL and column-family features.

NoSQL database

Traditionally, relational databases use schemas to create relationships between tables. However, modern applications have database requirements that are much higher than what relational databases can produce[42]. On the other hand, NoSQL databases (Not Only SQL), part of the non-relational database category, have much faster processing speed compared to relational databases. To achieve this speed, NoSQL databases remove the automatic table joins that relational databases are known for. This means that NoSQL databases only support simple transactions and queries. Other database types are more suitable for complex transactions and queries. Overall, Cassandra embraces a simple data model to increase performance.


With traditional relational databases, row-based technologies are used to store data in rows. In doing so, the data takes up more disk memory since rows are difficult to compress[43]. This is because each row is treated as a unique combination of values. At the data compression stage, the algorithm cannot remove duplicate rows because each row is unique. Furthermore, the number of columns in each row is always the same. For example, each row in a table about fruits will always have columns called name, color, and origin.

Cassandra uses a column-family oriented approach that stores columns (attributes) related to an object in a row. To reduce data footprint, Casandra only stores unique values in each column compared to storing every row as a unique value. To illustrate this feature, if there are duplicates in the Blood Type column, those duplicates will be removed before being stored in the database. To access columns for an object, a row key is used to first identify the row. For each row, the number of columns in an object does not matter. As an example, one row can store columns related to fruits, while another row stores columns related to a tweet[44]. With such a data structure, Cassandra is able to store much more data compared to relational databases. This feature makes Cassandra a popular choice for big data tasks.

How Thunder works

How Thunder works using open source technologies[45]

Thunder combines key functionalities of the four major open source components above to process big data and support Salesforce Einstein. Firstly, Kafka connects with multiple data streams generated by various devices, including IoT sensors. Kafka then publishes these data streams for other devices to consume. Once published, Apache Spark subscribes to these data streams and replicates them into Cassandra, where the data is stored in a column-based database. At the same time, Apache Storm also subscribes to these data streams. The software applies pre-defined data rules from users to these data streams and performs automated tasks based on additional rules set by users [46].

According to Adam Bosworth, Thunder is intended to allow business users to define data logic through a user-friendly interface [47]. As such, an additional layer of code was written on top of Apache Storm to convert business logic input from users into executable codes. These logic inputs from users can be event-driven (e.g. a door is opened), time-driven (e.g. at 12 PM), profile changes driven (e.g. order status changes to processed), and various other items. This setup allows business users to automate multiple tasks within an organization with ease.

Machine Learning & Artificial Intelligence

Artificial intelligence (AI) is how machines can imitate human intelligence. Machine learning is a part of AI, because when we think about AI, we think about machine based learning. At the base, machine learning is giving a machine or model data so that it can learn for itself. In the past, people thought that robots learned everything from us, but the human brain is too complex and some actions and activities cannot be described easily.

Supervised Learning & Unsupervised Learning

Supervised Learning Source

Supervised Learning

Supervised learning is a big part of machine learning. It is called supervised learning because the algorithm will learn from a data set. In supervised learning, there are some input data X that has a matching output data Y that the algorithm will learn. Once learned they can use the information to predict any future output variable Y's. However, to learn it, it is a process for the algorithm. If they get the output answer wrong, then the data set will correct them. It will rinse and repeat until the algorithm is fully capable of producing the necessary results.[48].

There are two groups of supervised learning: classification and regression . A classification problem happens when an output data Y already has a criteria that is required to group them by. A regression problem is when we're constantly trying to predict values.

Unsupervised Learning

Unsupervised Learning Source

Unsupervised learning is the opposite of supervised learning. Instead of having a predetermined data set, the algorithm is given an input data X and is expected to find patterns that link it to an output data Y. [49]

There are two groups of unsupervised learning: Clustering and Association . Clustering problems happen when we try to group up the data points to make it meaningful. An example would be when you're grouping customers by purchasing behaviors. Association problems happen when you're looking for a rule that fits a large portion of your data. For example, finding the rule such that people who buy X also buy Y.

Optimus Prime

Optimus Prime is the machine learning framework that Salesforce uses so that each Salesforce customer can have a model that is automatically created, without needing a data scientist behind it. "Optimus Prime is, in a sense, an AI that builds AIs—and a tool whose recursive nature is both beautiful and unsettling."[50]

Lead Scoring with Machine learning

Lead Scoring Source

Lead scoring helps us to see which leads you should focus more of your time on. This is based on how much interest a lead shows for your business. A lead is essentially a person or company you want to make a sale to. You will continue to contact the lead, and the lead will soon perform certain actions. The machine learning will rank these actions and score them. Some actions might score more because they lead to higher conversion rates. For example, if a lead opens an email that's one point and if he opens the link attached to the email, that action might have more points associated with it.

Einstein Discovery

Discovery Device Source

Einstein Discovery is known for its supervised learning. It is based on BeyondCore, which was acquired by Salesforce. Six questions (pictured right) can be asked to find out more meaning behind your data. Einstein Discovery will automatically discover any relevant patterns based on the data you have inputted.

How does it work?

Step 1: You will upload the data set. This comes from many common sources, CSV file, databases, Salesforce. Regardless of what the data set is, it's generally a spreadsheet.

Step 2: Next you will create a "Story". When you click the "Create Story" button, the model will automatically begin building a statistical model.

Step 3: The key techniques that are used to build the models are linear and logistic regressions. Einstein Discovery is able to learn quickly because of the technique that powers it.

Step 4: Finally a human is needed to make sense of the data that Einstein Discovery outputs.

Augmented Analytics with BeyondCore

Behind Salesforce's augmented analytics capabilities is BeyondCore, a recently acquired company that specializes in data analysis using artificial intelligence[51]. BeyondCore's technologies have been fully integrated into Salesforce's platform to power smart predictions and data mining. The software's main goal is to perform AI-powered analysis on big data and generate unbiased insights [52]. The following two subsections will illustrate why BeyondCore can provide better predictions than regular human analyses.

Regular Analysis Process & Biases

A normal analysis process begins with a question regarding a situation. This question is asked by the analyst who is working on the problem. After that, the analyst begins the analysis process, computing and reviewing data as they require throughout the process. As a result, the analyst uses the insight from their analysis to implement solutions to a problem.

This process is flawed since it is based on the experience, understanding, and assumptions of the analyst, which are susceptible to biases. The entire analysis process begins with a question that is influenced by the person's background and other factors. If the question is biased, the whole analysis will be biased, and thus, yield invalid insights that can potentially exacerbate the problem. There are four common biases in data analytics: confirmation bias, interpretation bias, prediction bias, and information bias [53]. Confirmation bias is when the analyst looks for data that confirms their hypothesis and rejects data that do not support their hypothesis. As for interpretation bias, it describes our tendency to interprets the data based on our background and understanding. Different people can have different interpretations of the same topic. Prediction bias refers to acting upon predictions without applying context to the data. For example, in 2017, the Chicago police used predictive analytics to predict crimes. The software attributed more risks to an African American person than a Caucasian person[54]. The police followed these predictions without thinking about the racial context of the predictions. Lastly, information bias refers to dirty data that can skew or invalidate the results. All these biases lead to flaws in the normal analysis process.

BeyondCore's Analysis Process

BeyondCore Explained

BeyondCore's process begins with the analyst providing a goal for the analysis and data. An example goal is to maximize revenue for all stores in British Columbia. After that, the software analyzes the data using AI to understand its business context and performs multiple statistical analyses to grasp the relationships between variables. As a result, BeyondCore generates an unbiased insight to be used for analysis.

Given the above process, BeyondCore can counter the common biases stated in the previous section. Firstly, since the software uses statistical models to analyze data, all data is taken into consideration to avoid confirmation bias. Secondly, by viewing data objectively through statistical analysis, BeyondCore can describe the relationships between variables objectively without any interpretation bias. Thirdly, BeyondCore suggests to the analyst which data segment to look into further to understand the results. In the same example, BeyondCore would suggest to the the police to drill-down to the African America population to understand why the crime rate is higher, instead of stopping at the initial prediction that African Americans are more likely to commit crimes. As such, the software prevents prediction biases from affecting the analysis. Lastly, BeyondCore partially prevents information biases by stating which population is underrepresented in the analysis. Overall, AI-powered analysis with BeyondCore can prevent many common biases. However, human interpretation of the results is still required.

Natural language processing

Natural Language Processing (NLP) is "a branch of AI that helps computers understand, interpret and manipulate human language"[1]. A common example of NLP would be your spam filter, which looks through your email and tries to find certain words, or phrases that are common within all spam. If it finds something like that, it will label it as spam.[2]

Under NLP, there is something known as Deep Learning . Deep learning allows a system to imitate the activities of a human brain, such as thought or decision making. Neural networks, created by reading in a lot of data by the algorithm, are used to make predictions for new data. Deep learning also allows algorithms to understand the meaning behind texts.

Einstein Sentiment & Einstein Intent

Einstein Language includes two APIs that you can use to unlock powerful insight within the text.

Einstein Sentiment Einstein Sentiment will feel your e-mails, social media and text from chat by classifying them as positive, negative or neutral. This can help you, identify the emotions of a prospect's email, whether to pursue them or not. It also allows you to help with dissatisfied customers first or to extend additional promotional offers to satisfied customers. You'll also be able to measure overall satisfaction or dissatisfaction with your products. It would also be quick for you to tell what people think about your brand across all social media channels.[3]

Einstein Intent Einstein Intent will give labels to unstructured text, such as emails, chats or web forms, so that we know what the users are trying to do . This can help you determine what product customers might be interested in and send customer inquiries to the appropriate salesperson. By analyzing an unstructured text in emails, meeting notes, or account notes, Intent will be able to identify the topics touched upon. If there are certain service cases, Intent will be able to successfully direct the cases to the right agents or departments.

While it may seem like Sentiment and Intent are able to replace the jobs of humans, it also creates more job opportunities. Data by itself is meaningless without a human putting insight into how to use that data.

Einstein in Salesforce Clouds

Since Einstein is part of the Salesforce platform, it can be integrated with other Salesforce Clouds. This section explores different features that Einstein offers for four Salesforce Clouds.

Einstein Sales Cloud

Salesforce Einstein with Sales Cloud

With Einstein integration in Sales Cloud, Salesforce offers users new capabilities, such as lead scoring, forecasting, activity intelligence, and recommended actions. The software learns from a team's activities and provide intelligent insights to users [1]. As discussed in the above section about machine learning, lead scoring allows customers to rank each of their leads based on their past interactions with a company. Forecasting allows customers to make predictions based on current data. While this function is standard for many CRM software, Salesforce uses AI to improve the accuracy of the predictions. As for Activity Intelligence, the feature uses sentiment and intent analyses to read through emails between a sales agent and a customer to track the sales progress [2]. The feature can automatically create reminders, calendar events, and update client status (such as from opportunity to lead). As such, sales agents will spend less time on administrative tasks and more time on selling. Lastly, Einstein can recommend actions for an agent to take based on an event. For example, Einstein can suggest the agent book a meeting with a company's CEO since the person sent an inquiry email earlier.

Einstein in Marking Cloud

Salesforce Einstein with Marketing Cloud

Marketing Cloud incorporates Einstein by providing automatic customer segmentation, predictive scoring, automated customer journeys, and social media post analysis. Based on customer data and activities, Einstein classifies customers into different segments that marketers can target. For each customer, a predictive score is given on how likely they are at performing a task, such as clicking on a link or opening an email [1]. Building on top of predictive scoring is the ability to create automated customer journeys. For example, marketers can create a rule to send reminder emails to customers whose email opening score is below 30%. Lastly, incorporating sentiment and intent analyses, Einstein can analyze social media posts to identify unsatisfied customers quickly. It can further understand which posts belong to which department and notify them accordingly.

Einstein in Service Cloud

Salesforce Einstein with Service Cloud

Einstein with Service Cloud enables the abilities to predict customer satisfaction (CSAT) score, recommend actions for support agents to take, identify similar support tickets, and spot high priority cases. A CSAT score is the most important KPI for call centers and customer support teams[1]. It indicates the satisfaction level of a customer. There are multiple factors affecting this score, from the total ticket processing time to the initial response time. Einstein analyzes all these factors using past ticket data from a company to generate a predictive CSAT score. Using this score as a baseline, the software can recommend the next best course of action to take to improve a customer's CSAT score. For example, it can suggest the agent offer a customer discount to increase their CSAT score by 2.71 points. Aside from CSAT scores, using natural language processing, Einstein can identify tickets with similar topics without having agents or customers tagging that ticket. Managers can use this feature to identify common issues with a product. Lastly, the software automatically identifies high priority cases since it understands the business context and sentiment. If a medium priority ticket was sent by a CEO with an angry tone, Einstein will mark the ticket as a high priority for agents to handle quickly. The video on the left demonstrates these functionalities more extensively.

Einstein in IoT Cloud

Salesforce Einstein with IoT Cloud

Since Einstein was designed specifically for big data, an integration between Einstein and IoT Cloud provides a large data lake to perform analytics on. With IoT data, users can define automation rules and execute these processes automatically. An example rule is when temperature readings from truck A's thermometer are below 10-degree Celcius, notify the maintenance team. These rules can be simple or complex based on the user's requirements. Furthermore, data is constantly being processed in real-time to ensure fast reaction time to changes in sensor conditions. Lastly, these user-defined rules can communicate between multiple Salesforce Clouds, such as Sales Cloud. Refer to the video on the right for more information.

Challenges faced by Einstein implementation

While Einstein offers many AI-powered capabilities, there are implementation challenges that are hindering its adoption. Firstly, the software is geared towards large companies that have enough data and resources. Since Einstein uses large volumes of data to create prediction models, small to medium sized businesses may not have this much data for it to analyze. As such, they are not able to take advantage of Einstein’s predictive functions. In terms of resources, implementing Salesforce alone already costs money. Einstein presents additional costs to an organization. If a business does not know how to operate Einstein, they will have to hire an implementation partner. All these costs can prevent small to medium sized businesses from implementing Einstein.

Secondly, IoT analysis requires a powerful network to support it. According to an industry study from Accenture Digital, one of the main concerns for implementing an IoT system is slow Internet speed [1]. Companies do not have sufficient bandwidth to handle the large volumes of real-time data being transferred constantly through their networks. With a weak network, Einstein is unable to process data in real-time, which can lead to delays in insights and reactions.

Lastly, human error is a concern for Einstein implementations. As with any analysis process, clean data is required; otherwise, the results will be invalid. Human errors can introduce dirty data that will then be used by Einstein to make predictions[2]. As people rely more on AI to perform analyses behind the scene, they will not be able to catch these errors. As such, there needs to be processes in place to identify human errors before they enter Einstein’s analysis.

Salesforce Success Stories

U.S. Bank

U.S. Bank Logo[3]

U.S. Bank is one the largest banks in the USA. The bank offers a variety of services including banking, investment, and payment services. U.S. Bank had been using its Sales Cloud solution since 2007 and already had around 15000 employees using the platform as part of their job duties.

Einstein Discovery

A major problem that U.S. Bank had was qualifying the millions of leads that were coming in for customers who had a potential interest in banking services. With limited wealth managers available, U.S. bank had to make sure that it was pursuing only those leads that had the most chance of becoming credible opportunities. This is when the bank decided to use Salesforce Einstein. U.S. Bank first used Einstein Discovery to analyze data from millions of leads. From this the bank was able to generate a list of top predictive factors that were most positively correlated with a lead converting into an opportunity. It was also able to discover that customers who were between 35 to 44 years of age and had a mortgage with U.S. Bank were more likely to increase their assets under management if they also opened a credit card with the bank, an insight unknown before Einstein Discovery did its analysis.[4]

Einstein Lead Scoring

With the insights from Einstein Discovery, U.S. Bank then used the Einstein Lead Scoring feature to score millions of incoming leads. The result was a list of leads ordered from highest to lowest conversion probability that was generated within the space of around 2 hours. By just focusing on those leads generated, U.S. Bank has been able to increase its lead conversion rate by a factor of 2.34. This has helped the bank not only make more money but be more efficient in its dealings with potential customers.[5]

College Forward

College Forward Logo[6]

College Forward is a small non-profit organization based in Texas. The primary aim of the organization is to help children from disadvantaged backgrounds enter college and complete their degree. Every year, College Forward works with around 12000 youth around the Austin and Houston areas to help them graduate from college. The organization does this by connecting youth with counselors who act as guardians, looking after the youth and ensuring they have what they need to be successful.

Einstein Discovery

A major challenge for College Forward is decreasing the drop out rate and increasing the graduation rate. With 12000 youth and only limited counselors, it is difficult to oversee all the youth within the College Forward system. When youth drop out of college it is often too late for counselors to be effective and so it is important for counselors to help an individual before he or she drops out. To be more proactive, College Forward decided to build Salesforce Einstein into their existing application on Service Cloud. Through sifting through years of student data, Einstein Discovery was able to generate a risk score for each student in the College Forward database that indicated the probability of that student dropping out of college. Variables considered by Einstein Discovery were, among others, missed school days, days late and changes in GPA.

Einstein Natural Language Processing

A big obstacle for College Forward was that most of the data entered into Salesforce on the students was qualitative or descriptive in nature. This made it difficult to deduce a score from this data because the data wasn't a concrete number or data point. To solve this, Einstein's natural language processing features were used to analyze each description and based on this analysis generate a concrete sentiment score. This sentiment score would then count towards the overall risk score for that student. For example, if a counselor entered into Salesforce that the student did not have a place to live, this would be picked up by Einstein as increasing the risk of that student dropping out and as such the risk score for the student would be increased.

Workflow Automation

With risk scores for each student, College Forward had workflows built into their application that would flag students with a risk score above a certain threshold, automatically open cases for them and send these cases to the appropriate counselors for further action. in this way, College Forward counselors have been able to more proactive and the College Forward graduation rate has increased to 85%. This is a substantial achievement since this graduation rate is higher than the national average for more privileged students.


To help other similar organizations, College Forward also made its Salesforce solution available on the AppExchange under the name CoPilot. Through this, CoPilot is now used by 40 organizations to serve more than 240 000 youth in 20 states and the African country of Rwanda.[7]


Cannon Logo [8]

Canon is a company that has been around for 80 years. Canon is a Japanese corporation that produces products such as cameras, camcorders, photocopiers, and more.[9]

Salesforce Sales Cloud

Canon not only uses Sales Cloud to track key customer information, but they also provide it directly to customers in the form of Image Tagging. Users can use it to look up a photograph they would like without them having to actually know the specific file name because of a tagging system. For example, a customer can look up "beach" and come across beach type photos because of the beach tag that is associated with it.

Salesforce also gives advice on the photographs that the customers have taken. By using the Image Analysis tool, user's can receive tips on how to and where to improve their pictures. It would take the metadata associated with each photograph, so that it knows exactly at what settings each photo was taken, to give you the best advice.[10].

Einstein Discovery

Canon has always been a big company in their industry. The company uses Salesforce to avoid duplicate information with their customer information, as well as to view complete customer information and to gain insight. When a new product is released, Salesforce is used to see how well received the product is. If a product needs tweaking, they will be able to fix it to produce a better product. [11]

App Exchange

One of the apps Canon uses is the Chatter social network for employees. This allows for teamwork across the company, ensuring that everyone can be on the same page with what the customer needs. Using the App Exchange, they are also able to build custom tools that can meet the needs of their customers. [12]

Salesforce Security

Security measures implemented

Salesforce implements six different types of security measures [13]:

  • Transparency: Salesforce's has a goal of total transparency. This is done so by giving live updates on any harmful security attempts, such as phishing, malware or intrusions. This creates a trusting relationship between Salesforce and their users, since the users always know the state of their data at all times.
  • Authentication: Providing browser encryption is important, which is why, Salesforce supports Transport Layer Security (TLS) version 1.0 or higher. They also provide a free app that allows users to access two-factor authentication (2FA).
  • Event Monitoring: When something happens, Salesforce will provide you with any information straight away. Should issues arise, it will be quickly resolved this way.
  • Multi-tenancy: Salesforce has different organization identifiers for each client. This ensures that each client only has access to their own information and data.
  • Isolation: By keeping the servers isolated, Salesforce provides a more secure server. They also use advanced technology to prevent any unwanted access..
  • Health Checks: Salesforce has a Security Health Check that helps customers identify and correct any weak security settings such as passwords, network configuration and more.

Salesforce Hacked

In October 2007, Salesforce's security was involved in phishing scams involving Automatic Data Processing (ADP) and Suntrust. The customers were sent emails that contained a link, that when clicked would download a password-stealing malicious software.

How exactly did it happen? An employee of Salesforce fell victim to a phishing scam. This scam exposed Salesforce customer lists, "including the first and last name, company names, email addresses, telephone numbers and related administrative data". [14] This event was what caused the series of highly-targeted phishing scams.

Salesforce Competitors

While Salesforce has the largest market share in the CRM market, there are other major competitors in the market. This section explores four major Salesforce competitors mentioned in the Salesforce CRM Market Share section above.


Adobe Logo [15]

Adobe’s CRM solution is called Adobe Marketing Cloud [16]. This solution includes multiple software offerings that focus on managing customers, analyzing audiences, and developing content for marketing purposes [17]. They are more geared towards the marketing aspect of CRM, especially content development and publication. In terms of AI capabilities, Adobe has developed an AI called Sensei for the content development component of its CRM platform [18]. Compared to Einstein, Adobe has chosen to focus on creating a niche AI for content creators instead of developing an AI for data analysis. However, Adobe is starting to introduce other AI features for its marketing cloud, including automatic customer targeting [19]. Since these features are still new, there has not been any case studies or customer success stories for Sensei.


Microsoft Logo [20]

Microsoft’s CRM is called Microsoft Dynamics CRM, which is part of the Microsoft Dynamic enterprise resource planning software [21]. Dynamics is a cloud solution, similar to Salesforce, that targets small to medium-sized businesses. In this way, this CRM targets a different market compared to Salesforce. However, in 2018, Salesforce announced Salesforce Essentials, which is its CRM solution for small to medium-sized businesses [22]. As for its AI functionalities, Microsoft’s AI focuses on helping sales agent nurture their relationships with customers [23]. In many cases, the AI simply reminds the agent of actions they need to take to help close a sale. For example, after a meeting, the agent is reminded to record their notes. Overall, Microsoft’s AI is not as developed as Einstein.


Oracle Logo [24]

Oracle’s CRM, named CX Cloud Suite, is a collection of cloud-enabled applications that cater to different areas of a business. From marketing and sales to commerce and customer service, Oracle’s offerings are modularized like Salesforce’s Clouds [25]. After acquiring Sun Microsystem in 2010, Oracle has been falling behind in the cloud department [26]. With its acquisition of Responsys in 2013, Oracle revamped their cloud offerings to compete with Salesforce. In the AI department, Oracle offers features such as personalized content for customers, predictive product recommendations, chatbots, and sell process optimization [27]. However, their AI features are not as extensive as Einstein. Regardless, Oracle is boosting its efforts in the cloud department.


SAP Logo [28]

SAP's CRM platform is called SAP C/4HANA. From a business model perspective, SAP offers modular CRM components to customers, similar to Salesforce's Clouds. As an ERP developer, SAP integrates their CRM with the rest of their ERP suite. In 2018, SAP released Customer Cockpit 360, which provides agents with a 360-degree view of a customer [29]. This means that they have access to all interactions a customer has with a company. As such, their AI is able to predict customers who are most likely to churn. Furthermore, SAP's AI can automatically explore data from customers and offer unbiased insights similar to Einstein Discovery. However, SAP's AI does not present insight in an easy to read manner like Einstein does. It still relies on the analyst to interpret the findings. Overall, SAP is moving towards a very similar cloud model as Salesforce to increase its market share in the CRM market.

Salesforce Implementation


Pricing Packages for Sales Cloud [30]

There are various costs structures for implementing Salesforce and most of the time cost is determined by the complexity of the solution that you need. However, in general, Salesforce has split up its pricing plans for basic solutions like Service and Sales Cloud (Pictured above) into four packages. The Lightning Essentials package is typically for small businesses that will require a basic Salesforce solution for no more five users. Lightning Professional is used for bigger business that needs only basic functionality and little to no additional customization. Lightning Enterprise is for large businesses that need extensive customization and need most of the platform features to support their solution. Lightning Unlimited is for businesses that need to be able to complete complex customization and need all of the platform features.

It is important to remember that if you want to add Salesforce Einstein to your package it will cost more. This cost will be determined by Salesforce and depends on what your needs are. In addition, it is important to remember that while clouds like Sales Cloud have standard pricing, most of the other clouds have pricing that is determined after an evaluation of your needs by a Salesforce sales agent. For a better look at what standard pricing is available for Salesforce, you can view the pricing structure available on their website [31].



Businesses looking to customize Salesforce can do so in-house. The following are some advantages and disadvantages of in-house development:


  • This might be a cheaper option than using an implementation partner depending on the complexity of the project
  • Businesses are able to develop their own knowledge of the Salesforce platform that can be used for future projects
  • Businesses maintain control over the whole development process and can make changes as they wish


  • The project may become more costly if the project team is inexperienced due to mistakes and lack of knowledge
  • Project team members and executives may become emotionally attached to the project and this can inhibit rational decision-making regarding its continuance
  • Project team members do not benefit from insights that can be gained from similar projects done for other organisations

Salesforce Implentation Partner

Businesses wishing to customize their Salesforce instance can also do so through a Salesforce Implementation Partner. These partners, like Traction on Demand[32] and Accenture [33], are specialists on the platform and can be utilized for customization projects. The following are some advantages and disadvantages of Salesforce Implementation Partners:


  • They have extensive experience on the platform and can help you design the best possible Salesforce solution
  • They can provide you with best practices and insights that come with implementing previous implementation projects
  • They often have the technical expertise necessary to tackle complex integrations and difficult custom development


  • They are expensive
  • Using their services increases your reliance on them for future implementation projects since they have the knowledge of how your solution was built
  • You cannot easily make changes to project requirements once you sign off on them

Salesforce Training


Trailhead [34] is Salesforce's official training site. On this site you can learn everything there is to know about the Salesforce platform. Knowledge areas are split into modules and divided further into various units. To complete a unit, you have to complete a quiz and/or a challenge, a challenge being a practical exercise performed in an actual Salesforce developer environment. You get a badge added to your profile when you finish a module. You can earn a super badge for building a certain feature or complete solution in a developer environment.

Trails are collections of modules that follow a particular learning path. You can customize your own trail with the modules you are interested in or you can follow the standard learning trails already on the site. In this way, you can customize your own learning and get the most out the training experience.[35]

Udemy and [36] and [37] are online learning platforms where expert instructors provide training courses on various topics. There are numerous Salesforce related courses on both platforms and these courses can be used to not only learn more about Salesforce but also prepare for the various Salesforce certifications.


The following are common careers related to Salesforce and their average yearly salary in Canada:

  • Salesforce Administrators typically maintain and organize an organization's Salesforce instance. Depending on their skill level, administrators are also typically responsible for some platform customization. The average salary for a Salesforce Administrator is around CA$ 94,702 per year.[38]
  • Salesforce Developers are responsible for more technical customizations, often utilizing APEX code and other languages to create custom solutions on the platform. They typically work for Salesforce implementation partners but can also work for a company as an in-house developer. The average salary for a Salesforce Developer is around CA$ 126,830 per year. [39]
  • Project Managers are in charge of managing Salesforce implementation projects and ensuring the project stays within scope. These managers can work as implementation partners but can also be employed as contractors. They have an average salary of around CA$ 125,458 per year. [40]
  • Platform Architects are responsible for designing complex solutions on the Salesforce platform to meet specific business requirements. They typically work for implementation partners or as contractors. They earn an average salary of around CA$160,858 per year. [41]
  • Data Architects are responsible for everything related to data when building a Salesforce solution. This includes data integrations with other systems, data modeling and data loads of customer data. They typically work for implementation partners or in-house for organizations. Data Architects earn an average salary of CA$ 112,899 per year.[42]


Below is a roadmap of the different certifications available on the Salesforce platform. Certifications typically cost around US$200 and can be taken remotely or at a designated testing center. For more information you can visit the Salesforce certification website[43].

Salesforce Certification Road Map source[44]


Bryan Vo Jane Gao Miguel Hof
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada


  30. source

Personal tools