Back in December when Salesforce bought Slack for almost $30 billion, there was a pretty good Tweet floating around about no one actually knowing what Salesforce does:

Of course, if you’re reading this, you almost certainly know what Salesforce does and probably have logged in twice today. But also, none of us really know what Salesforce does because that’s kind of the idea of Salesforce. Its popularity is, at least in part, based on its flexibility. Like many large-scale enterprise systems, Salesforce is a reflection of the unique processes and business objects their customers encode into the software.

But let’s leave the philosophical questions aside and focus on the practical: what is Salesforce to most people? I think salespeople would generally say it’s a system for storing and managing contacts, accounts, opportunities, and leads. Those are the core objects that come with every Salesforce account and the things nearly every company builds around when setting up their CRM system. For our purposes, let’s leave out leads for now (that is worthy of a whole other post) and focus on the three core objects (accounts, contacts, and opportunities) and how they interact.

Low-Variance vs. High-Variance Data

Let’s start simple with accounts and contacts. An account is typically made up of one or more contact records, and, critically, a contact has just one account. The data held on these records is mostly what we’d call low-variance: the fields are made up of verifiable facts about the person or company like their address, title, employee count, or industry. It’s low-variance because no matter who was filling it out, you’d expect the answer to be the same.

On the other hand, opportunities are much higher variance: they’re asking a salesperson to make judgment calls about the state of the opportunity. From my own experience working with sales leadership and ops, there are three critical pieces of information that need to be captured on the opportunity: the stage (how far along is this deal?), the close date (when will we land it?), and the deal size (how much is it worth?). The reason leadership and ops are so focused on that data is that it’s almost impossible to forecast effectively without those three pieces of information. Of course, it doesn’t end there, and frequently, frontline sellers are asked to add lots of additional information to the opportunity to help inform that forecast. That’s why you’re required to log calls, add objections, and fill out any of the other forty fields on your company’s opportunity record.

But when you start to think about the low-variance/high-variance divide between those critical record-types, you begin to see why different technology has grown up around CRM. On the account/contact side, enrichment providers like ZoomInfo and Clearbit have emerged to make it easier to ensure that your low-variance account and contact data are clean. Instead of asking a salesperson to track down the company’s headquarters address or the title of the new contact they met, enrichment providers will drop all that data directly into Salesforce for you. Getting all the information is hard, but knowing which field to put it in is easy. Asking busy salespeople to spend time updating low-variance data is a waste of time and energy. 

Moving From Filling Out Fields to Being In The Field

Hopefully, you don’t work in a sales organization that is making you track down the phone number for HQ anymore. But there’s a good chance you do work for one that requires you to keep your opps updated. And that makes sense. Going back to our low-variance/high-variance idea, there’s a bunch of information about opportunities that require the judgment and expertise of a salesperson. As we all know, sizing a deal in an early stage is easily as much art as it is science.

But even though what goes into these fields may be high-variance, keeping them updated is not. When sales was phone calls and steak dinners, you had to transcribe your conversations and thoughts into a CRM to track the progress of your deal. But as our sales interactions have become increasingly digital, some smart companies have come along and recognized that we could use the data directly from those interactions to keep opportunities updated. This is where tools like Clari and Gong come in: they’re taking the actual interactions of sales—the emails and Zoom meetings—and using that to inform the state of the opportunity. And it works, both because there’s a lot more data to work with, and the compliance is higher than asking salespeople to manually input the data after the activity. Plus, no company has entirely eliminated the need to update CRM on occasion.

Expanding the Universe of Interactions

Where things get interesting (if you’re a nerd like me) is to think about what that interaction data is and where it belongs. Salesforce calls its contact, account, and opportunity records “objects,” which is easiest to imagine as just a database table or an Excel sheet. It’s got columns, which are fields, and rows, which represent each contact or opportunity. If you’ve ever worked with Excel, you’ll know that rule number one is to try to add rows, not columns. And when you look at Salesforce, you see the same idea represented: the core fields like opportunity owner, amount, and close date are the “columns” of the record, and the associated data like logged calls are just links to other Salesforce records.

The question, then, is how to classify all those sales activities? It doesn’t belong on the record itself—it’s not core data—instead, it’s an associated record, like a logged call or email. So what all these external systems are doing is building up the opportunity record by associating activity in the form of meetings and emails. You can generically think of these as “interactions” in that they’re interactions that occurred between a seller and prospect. The advantage to keeping them separate is that you can also link them to the contact and account. That way, you don’t only have to look at the opportunity to see all the meetings, but you can also look at the account to see all meetings across all opportunities active for that account. 

But now we’re moving into a whole new phase of selling. Not only have we shifted to having the majority of our interactions digitally, but to one where our prospects are generating a ton of data in and around our product. That data is the ultimate sales signal. After all, what someone does is always a more reliable indicator than what they say. 

So now if we want a full picture of our customer we need:

  • Account and Contact Info: the objective data about where the company or prospect is located, employee count, etc.
  • Sales/Prospect Interaction Info: the emails and meeting transcripts
  • Prospect Product/Marketing Interaction Info: the actual interactions from your product and marketing
  • Opportunity Info: the judgment calls of sellers that don’t fit into one of the other buckets

Where does all this data go?

That is the multi-hundred billion-dollar question. Obviously, Salesforce would like you to keep loading it in Salesforce, but as I outlined, it doesn’t necessarily lend itself all that well to storing this kind of interaction data. The idea that CRM can continue to be your single source of truth for everything you know about a customer is no longer realistic. 

Customer Data Platforms (CDPs) like Segment, mParticle, and Hull have a pretty good claim over much of this data as they’re able to both pipe and join interactions coming from various systems as they’re happening. But CDPs are technical tools, not something you will hand over to a team to log into directly. They feed “destinations” like Customer.io for marketing automation, Amplitude for product analytics, and Variance for sales data. 

Another obvious contender is the data warehouse. That’s where all this data lands anyway. Companies with a data warehouse running are combining their CRM, product, customer support, marketing, and finance data in one place to do analysis. The thing about a data warehouse, though, is that it’s not a primary database, and it’s still mostly updated in batch. That’s fine if you’re doing weekly or monthly analysis, but not great if you want to use it as a signal to reach out or deliver a timely message.

Catching Signals

In the end, my hunch is that we’re moving away from the idea there’s only one place to find all this stuff, and moving towards a world where data needs to be available to people when and where they need it. I would guess Salesforce agrees, and that’s part of the reason they bought Slack for nearly $30 billion.

We are moving fast towards completely automating away the low-variance parts of the sales job (🙌) and increasingly finding ways to capture every interaction around an opportunity and convert that into actionable signals for sellers. In organizations where this works it’s like a superpower for salespeople, allowing them to close deals bigger, stronger, and faster. Like William Gibson said, “The future has arrived — it’s just not evenly distributed yet.”