The $90 billion CRM industry was built for sales pipelines. You have 12 clients.
Ten minutes before a client call: searching Gmail for “board presentation,” scrolling WhatsApp for that voice note from last week, checking my calendar for what we agreed to last time. Finding most of it. Eventually.
That was my CRM for years. I’ve since built my own solution — an AI Chief of Staff named Athena who synthesizes client context before every conversation automatically. But here’s the thing: the fact that I had to build her from scratch tells you everything about the category she replaced. And if you don’t have your own Athena, that pre-call scramble is still your Monday morning.
This post isn’t about what I built. It’s about why a $90 billion industry left me no choice.
The $90 Billion Mismatch
The global CRM market is worth over $112 billion and growing at 12% annually. Salesforce alone pulls in $38 billion a year. The entire category is built around a specific set of problems: pipeline velocity, deal forecasting, lead nurturing at scale, and sales team coordination.
If you manage 200 accounts with a five-person sales team, these tools are exactly what you need. If you’re running a 10-person agency with 20 active clients — or a solo consultant with 12 — they’re an expensive filing cabinet you feel guilty about not using.
The industry knows about the adoption gap. Over 90% of companies with 10 or more employees use CRM software. Businesses under 10? Roughly half. The CRM industry reads that gap as untapped market potential — millions of small businesses that just need the right onboarding sequence or a simpler interface to get on board.
But here’s the part nobody in the CRM industry will say out loud: maybe those businesses aren’t lagging. Maybe they’re correctly sensing that the tool doesn’t match the problem.
Over 40% of businesses have abandoned a CRM because it lacked the features they actually needed. And 76% of CRM users report that less than half their data is accurate. These aren’t training failures. They’re design failures — the predictable result of forcing a pipeline-shaped tool onto a relationship-shaped problem.
Why CRM Vendors Can’t Fix This (Even With AI)
Every major CRM vendor is racing to bolt AI onto their platform right now. Salesforce just committed $4 billion to expand data center capacity for its Agentforce AI workloads. Microsoft embedded Copilot into Dynamics 365. HubSpot has AI assistants throughout the product.
The pitch is seductive: AI will finally make CRM work for everyone.
But they’re adding AI to the same pipeline architecture. An AI that helps you move deals through stages faster doesn’t help when your actual problem is “I can’t remember what my client told me last Tuesday about her board presentation.” An AI that scores leads by engagement signals is useless when you have exactly seven prospects and you already know all of them by name.
This is part of a broader pattern in enterprise software right now — vendors bundling AI into existing products to justify higher prices rather than asking whether the underlying architecture serves the user. The answer, for small service businesses, is no. And more AI on a wrong foundation is still a wrong foundation.
What You Actually Need (And What It’s Called)
Here’s what your day actually looks like. You have a call with a client at 2 PM. Before the call, you need to know: what you discussed last time, what you promised to deliver, what their current frustrations are, whether there are any outstanding invoices, and what’s happened in their email thread since your last conversation.
That information exists. It’s just scattered across your inbox, your WhatsApp messages, your calendar notes, your Google Docs, your Notion pages, and the part of your brain that’s also trying to remember to file your VAT return. And if you have a small team, multiply that by every person who’s ever talked to the client — their inbox, their notes, their memory of that one phone call three weeks ago that nobody documented.
This isn’t a CRM problem. It’s a client memory problem. And the distinction matters.
A CRM is designed to move people through stages: lead → qualified → proposal → negotiation → closed. It assumes volume, velocity, and a funnel shape to your business. A client memory layer does something entirely different. It accumulates and synthesizes context over time so you can walk into every conversation knowing everything relevant about that relationship — without spending 20 minutes doing archaeology across five different apps.
The difference between these two architectures is the difference between “move this deal to the next stage” and “remind me why this client restructured their team in November.”
I wrote last year about building an AI system that actually remembers — not preference memory like “user likes bullet points,” but real operational memory that synthesizes emails, documents, and conversations into strategic context. That project taught me something I hadn’t expected: the hardest part wasn’t building the technology. It was recognizing that the category I’d been trying to force my work into — CRM — was architecturally incapable of solving the problem.
The fact that I had to build it from scratch tells you everything about the category failure.
The Quiet Abandonment
When an enterprise abandons a CRM implementation, there are postmortems. Consultants are called. Reports are written about “change management failures” and “user adoption challenges.” Research puts the overall CRM failure rate somewhere between 30% and 70%, depending on who’s measuring and how they define failure.
When a small business abandons their CRM, it’s silent. They just stop logging in. The pipeline stays frozen at whatever state it was in four months ago. Contact records gather dust. Nobody writes a postmortem because there’s nobody to write it for.
The CRM industry doesn’t track this attrition because it looks like user error — “they weren’t committed to the process” or “they need better training.” But the pattern is too widespread to be individual failure. When your ICP research shows that small business owners consistently cite “disjointed tools” as a top pain point, and the solution the market offers them is another tool that doesn’t connect to where their actual client context lives, the diagnosis isn’t “they need more training.” It’s “the category is wrong.”
This is especially acute for entrepreneurs operating across borders — an invoice in Moneybird, a conversation in English on WhatsApp, a Dutch tax filing that changes the project scope, a client email in a completely different timezone. Your context isn’t just fragmented across tools. It’s fragmented across languages, systems, and countries. No CRM on earth was designed with that in mind.
What “Right” Would Look Like
I’m not building a product here. But I’ve spent enough time with small service businesses — solo consultants, 10-person agencies, coaches, DAFT entrepreneurs running businesses out of Utrecht — to know what the actual requirement looks like.
It’s simple, and it’s not what any CRM offers:
One place where everything about a client accumulates automatically. Not because you logged a call or updated a field, but because the system watches your email, calendar, and messaging and connects the dots. Before your 2 PM call, you see a synthesized brief: here’s what’s happened, here’s what’s outstanding, here’s what you should probably ask about. That kind of contextual synthesis — understanding not just what happened but why decisions were made — is what turns scattered information into actual business intelligence.
No pipeline. No deal stages. No lead scoring. Just: “Here’s everything you need to know about Sarah before you talk to her.”
The technology to build this exists today. The reason it hasn’t been productized for small service businesses is economic, not technical: a tool that does this well for 20 clients is a $10-20/month product. CRM vendors need to justify $50-300/month per seat. They can only do that by adding pipeline management, sales automation, reporting dashboards, and all the other features you feel guilty about not using.
The complexity isn’t a bug. It’s the business model.
Where This Gets Practical
Most small service businesses I work with have a version of this problem hiding in plain sight — client knowledge that exists but isn’t consolidated anywhere useful. It’s in the founder’s head. It’s scattered across three people’s inboxes. It’s in email threads nobody can search efficiently.
If your CRM is mostly empty and you feel vaguely guilty about it, that’s the conversation I have. Not about which CRM to pick. About whether the category serves you at all — and what to build instead.
Want more like this? I write about AI adoption for people who run things — not hype, not doom, just what’s actually working. Subscribe here.
Sources:
- Fortune Business Insights, “Customer Relationship Management Market Report” (2025)
- Wave Connect, “CRM Statistics 2026: 80+ Facts and Data” (March 2026)
- Johnny Grow, “The CRM Failure Rate is 55% in 2025” (October 2025)
- Mordor Intelligence, “Customer Relationship Management Market Size” (2026)
- Cyntexa, “CRM Statistics 2026” (December 2025)
