Too Big or Not Too Big?

A look into your data in Salesforce

If your company is planning to store and analyze big data, it’s natural that it should look for a big data solution. However, there is so much buzz going on around Salesforce and its AI-powered analytical components that it’s impossible to ignore it. It seems that Salesforce’s products are almighty and can work wonders! What if they are also capable of dealing with big data successfully? Our Salesforce consulting team has prepared an overview to explain how Salesforce and big data are related.

Salesforce as a source of your big data

Salesforce offers an extensive portfolio of products: Sales Cloud, Marketing Cloud, Pardot, Service Cloud, Commerce Cloud and IoT Cloud are only a couple of names. With such a diverse offering, one should expect that there is just as much data accumulated in Salesforce. Undoubtedly, this idea is right. However, what data is there — big or traditional? Well, in fact, both. As the subject of our interest is big data, let’s name some examples:

• Customer segments identified based on personal information and typical behavior;
• Marketing promotions and activities split by channels, devices, customer segments;
• Customer’s response to marketing promotions;
• Communication details: emails, requests, activities, orders, calls, online chats and call center logs;
• Recent posts in social media and industry-specific news;
• Big data generated by connected things;
• Service-related data: description of cases and questions from the community.

To form a clear idea of how this data can be presented, check this demo that shows how a customer profile in Salesforce can look like.

Salesforce as a source of insights

Salesforce offers an AI-powered analytical tool Einstein. Introduced to support data-driven decision-making, Einstein makes all four types of data analytics available. Invisibly for end users, Einstein enables smart data discovery, builds predictive models and supports end users with stats, explanations, forecasts and recommended actions. For example, a sales rep can see the opportunities won in Q3 figures that come together with Q3 vs Q2 comparison, Q4 prediction rate and the opportunities they may focus on to hit their quotas.

A special app Einstein Discovery analyzes millions of data combinations and automatically prompts possible correlations and causations (like in this example with the shrinking margin). The app enables data import from multiple formats (Salesforce objects, CSV files, Oracle, Microsoft SQL Server, Postgres, Hadoop, MySQL, etc.) and makes complicated big data analytics more available. One should not be a data scientist to get actionable insights that are based on statistical models and algorithms. Besides, Einstein Discovery even prompts on how to improve data quality, for example, by suggesting a business user delete duplicates.

In Commerce Cloud, Salesforce DMP (data management platform) with in-built Einstein allows delivering personalized recommendations and shopping experience. For instance, Adidas processes customer data to create relevant pop-up ads across different channels and remind about the product to those who surfed their website but switched to another one. Besides, if a visitor chooses to return to Adidas’ website, they will open not a general-view page, but the one tailored to the visitor’s preferences — with the products the customer might like on top.

Salesforce or a big data analytics solution?

Undoubtedly, Salesforce provides great opportunities for handling your data. The fact that Salesforce makes descriptive, diagnostic, predictive and prescriptive analytics is impressive. Besides, Salesforce’s analytical components can deal with big data.

Let’s say, you have unstructured data, for example, call center logs shown in a customer’s profile. Definitely, a sales rep will pay attention to this information and will plan their communication with the customer accordingly. However, Salesforce analytical tools will also be able to analyze all call center logs from all customers and identify top-10 topics for complaints.

When you should be careful

1. Though Einstein is impressive, you should not forget that it’s AI-based. In our example with margin shrinking, Einstein has found multiple correlations that were supposed to influence margin reduction. However, it’s up to a data analyst to question each of these correlations, select the meaningful ones and double-check them by asking additional questions.

2. Some Einstein features are reasonable only if a company has risen to a data-maturity level. For example, lead scoring powered by Einstein is not generated until a company has enough data. One of the requirements, for example, is to have at least 1,000 leads created over the last six months.

3. The algorithms that stand behind Salesforce’s AI components are built to solve a standard set of tasks. If your business tasks are untypical, you can contact Salesforce’s support with the request to retrain the model. And, hopefully, your request will be satisfied.

If the limitations listed above are critical for your company, you can consider a dedicated big data analytics solution and Salesforce as one of important big data sources. You will receive a comprehensive picture, but will also have redouble your efforts, as a big data analytics solution deserves a separate project — from design to implementation.

Afterword

The analytical components in Salesforce are quite scattered and bear different names. Only in the examples we provided, you’ve encountered Salesforce DMP, Einstein and Einstein Discovery. Not to get lost in the sea of products and names and not to miss an opportunity provided by analytical components, it may be reasonable to turn to Salesforce consultants who will carefully collect your requirements and orchestrate a solution to cover your needs.

As far as Salesforce and big data relationship is concerned, we can definitely say that Salesforce is a source of both traditional and big data. When it comes to big data analytics, you can opt for either Salesforce’s AI-powered analytical components or a dedicated big data solution — all depends on your business requirements.