Most executives will tell you that customer relationship management (CRM) systems are essential technology. Many will tell you they are the key mechanism for gaining a deeper understanding of customers, building strong relationships with them, and making data-driven decisions that maximize customer satisfaction and lifetime value.
What if I told you that this “essential” technology will fade in relevance over the next 10 years, as a new generation of event-driven systems come to the fore? And the reason for CRM’s strategic decline will be the growing realization that CRM systems cannot actually fulfill their promise of creating better interactions with customers. In fact, the opposite is true: CRM systems are a primary cause of the disconnect that most companies suffer between their sales, marketing, and customer success teams.
Simply put, CRM systems can’t do what companies need them to do. Today, companies turn to CRM systems as literal cloud-based spreadsheets of deals in the forecast. However, what companies need is a system of record that tracks activity data and provides information on what actions should be taken to maximize the chance of closing a deal as won.
CRM systems lack those capabilities, so a new approach to maximizing customer data is developing based on powerful advances in artificial intelligence, storage and computation.
To understand why CRM systems are so flawed, you need to first understand how CRM systems work and how they were developed in the first place. CRM systems are built around an object-oriented data model versus an activity streaming model. HubSpot defines a CRM object as “the different types of relationships and processes your business has,” common ones being contacts, companies, deals, tickets, or custom objects companies create. In a database, each object record is comparable to one line with many fields in an Excel file–also known as the rows and columns of structured data.
An object-oriented model is inherently limiting and problematic–and the divide between sales, marketing, and customer success teams is due to the fact that each is looking at different objects. A sales team is focused on opportunities, or “the deal.” Marketing cares less about opportunities–its concern is another CRM object, leads, for the purpose of building a pipeline. The customer success team cares less about both opportunities and leads. Its focus is accounts. And the support team only cares about tickets.
What’s really needed in a modern system of record about prospects and customers is a core focus on employee activity–no longer just objects as the lowest common denominator that is universal across all teams. Activities capture what team members are doing, day-in and day-out, in their interactions with customers and prospects. When we track and analyze employees’ activities at scale, we can readily identify patterns, glean insights, and operationalize a focus on productive activities that help the company reach its goals.
So why are CRM systems constructed around objects instead of activities (or more generally, “events”)? When CRMs were first built decades ago, they lacked the computing and storage capabilities to handle the sheer volume of activities–i.e., any given deal might have 10,000 associated activities, and a company might have 10,000 deals in progress. The limitations of SQL databases–the main technology when CRM systems were developed and still used by every CRM system today–forced a focus on objects rather than activities because SQL simply couldn’t scale.
In addition, there was no context-aware technology to match activities to the right objects so teams could glean useful information. AI-driven matching capabilities that could figure out that, say, this phone call is related to this specific deal (out of 20 active ones) with this prospect (out of 10 other accounts with the same domain and company name), were decades in the future. With the advent of cloud storage, cloud compute, and cloud distributed AI, and with legal systems getting closer to allowing cross-customer AI learning (similarly to how Google trains their AI across all users’ searches), it is now possible to build an activity-centric system that learns from the behavior of all customers and sellers.
The data pipelines and streaming systems that are necessary to scale for action-based systems of record, like Kafka, have only emerged in the past 10 years. Confluent, which provides auxiliary tools that make Kafka easier to use, jumped to a $16 billion market valuation after its IPO last year–a strong indicator that systems of record that process events are supplanting systems of record that process objects.
The current state of CRMs, as object-oriented systems of record, is not fixable. Companies will still need to store objects. However, the way we store objects today is not extensible to human actions or events. As such, an events layer, spanning instances and data silos, will be built on top of the legacy CRM systems.
Within 10 years, every object-oriented system of record will have an AI-enabled events data pipeline on top of it, developed by a new generation of third-party vendors with expertise in these new technologies. (An object-oriented vendor that does SQL is not going to be good at building an events data pipeline on Kafka.)
My advice for business and IT leaders today? Think through your systems of record strategy. First, identify where you have only object-oriented systems bet, like CRM, ATS, ERP, etc. Then think about the event-level data and insight you are missing. Are you missing your sales activity? When you have a marketing activity, are you missing the related BDR activities? When you are hiring, are you missing recruiting activities that are happening with those job candidates? In a nutshell, are you collecting all the granular activity data that precisely describes what your best people do and what makes them so good vs the rest?
Identify those data collection and storage gaps because they indicate what you need to address in your future systems of record strategy. Today’s CRM systems are poised to move over time to an “invisible” storage and integration layer. The companies that are planning today for the shift to AI-enabled data pipelines will be in a much stronger position in the future to leverage data to maximize customer satisfaction and lifetime value.