When pipeline numbers slip, the first instinct is almost always the same: blame the tool. The CRM is messy, the automation is old, maybe we need a migration. So a project gets scoped, an agency gets hired, fields get renamed — and six months later the pipeline looks exactly the same, only now everyone’s tired.
I’ve spent nine years inside B2B SaaS revenue operations teams, and I can count on one hand the number of times the CRM itself was the problem. The tool is almost never where the revenue goes missing. The revenue goes missing in the handoff — the stretch of the funnel where a lead stops being marketing’s and starts being sales’s.
The handoff is a system with no owner
Think about who owns each stage of your funnel. Marketing owns acquisition. Sales owns opportunities. But the middle — the moment a lead is declared “qualified”, routed, accepted or rejected, and worked or ignored — belongs to both teams, which in practice means it belongs to neither.
That ownership vacuum is where three specific leaks live. I find at least one of them in nearly every HubSpot or HubSpot + Salesforce instance I look at.
Leak one: the definition nobody re-agreed
Ask your head of marketing and your head of sales, separately, to define an MQL. If the answers differ — and they usually do — every downstream number is built on a disagreement.
The definition was probably agreed once, in a workshop, years ago. Then the ICP shifted, the scoring model was tweaked by an admin who’s since left, a new product line launched — and nobody re-opened the definition. Marketing now celebrates MQL volume that sales quietly considers junk. Sales rejects leads for reasons marketing never sees. Both teams are right, and both are working from different data.
The check takes ten minutes: pull last quarter’s MQLs and look at the sales-acceptance rate. Below roughly 60%, you don’t have a lead quality problem — you have a definition problem. (I go deeper on this in the MQL→SQL definitions piece.)
Leak two: routing rules from three admins ago
Routing logic is write-once, audit-never. Rules get layered on for a territory change here, a new rep there, an exclusion for a product launch that ended in 2024 — and the stack of rules quietly develops gaps. Leads that match no rule sit unassigned. Leads that match two rules bounce between owners. Round-robins include reps who left months ago.
The symptom is speed-to-lead measured in days rather than minutes, which matters because response time is one of the most conversion-sensitive variables in the whole funnel. The fix isn’t more rules. It’s an audit of the ones you have — here’s how I run that.
Leak three: the report everyone corrects by hand
If your board deck requires a spreadsheet “adjustment” before the CRM numbers go in, you already know about this leak — you’re just calling it a reporting quirk instead of what it is: the visible symptom of the two leaks above. When definitions drift and routing misfires, the numbers stop describing reality, people stop trusting them, and decisions move to gut feel. Reporting trust dies in a specific order, and it has to be rebuilt in the reverse order.
The version of this I see most often looks like a CRM problem and isn’t one. A team is convinced the platform is failing them — pipeline is soft, the data feels wrong, someone floats a migration. Look closer and the tool is doing exactly what it was told: an MQL definition that drifted two ICPs ago, routing rules nobody has audited since the last reorg, and a board number that gets hand-corrected every month. Nothing in the software is broken. The handoff around it is — and once the definition and routing are fixed, the same CRM starts producing numbers people trust again.
Why “fix the CRM” projects fail
A CRM rebuild resets the tool but not the system around it. The definitions still drift, the routing still rots, the trust is still broken — now in cleaner fields. That’s why teams that migrate platforms often report the same problems within a year: the leak was never in the software.
The handoff is a process problem (shared definitions, SLAs), a people problem (two teams with different incentives and no shared owner), and only then a systems problem (the rules and fields that encode the first two). Fix them in that order and the CRM largely fixes itself.
Run the three checks this week
1. Definition: MQL→sales-accepted rate, last 90 days. Below ~60% = definition drift.
1. Routing: time from MQL to first sales touch, median and 90th percentile. If the 90th percentile is measured in days, rules are misfiring.
1. Trust: does any leadership report require manual correction? If yes, trace one corrected number back to its source.
Each check takes minutes and requires nothing but the access you already have. What they usually reveal is that the pipeline isn’t leaking everywhere — it’s leaking in one or two specific joints, and those are fixable in weeks, not quarters.
—
If the checks turn something up and you want a second pair of eyes: I run a fixed-fee Pipeline Leak Audit that maps all of this with evidence, or you can start with a free 30-minute health check — you’ll leave with the two or three places I’d look first in your instance.