Which three numbers separate them?

Win rate on qualified deals, activation rate of new accounts, and logo churn reason codes. Win rate has to be measured on qualified deals only, because an unqualified pipeline makes a competent sales team look broken and is itself a different problem. Activation needs a definition tied to the value the customer bought rather than to a login or a setup step. Churn reason codes need to come from conversations rather than from a dropdown in the billing system, which is where most companies' reason codes are generated and why most companies' reason codes say "price".

All three are usually available in some form within a week: win rate from the CRM once you agree the qualification filter, activation from product analytics if the events exist, and reason codes from whoever handled the last twenty cancellations. Where the numbers do not exist, that absence is itself the first finding.

What do the four combinations mean?

Two variables, four cells, and each cell points somewhere different. The value of laying it out this way is that it stops the argument being about who is at fault and makes it about which cell the data puts you in. Run the numbers before the meeting, not during it.

Swipe the table sideways to compare →

The bottom-right cell is the one most often misread. Low win rate and weak activation together usually means the segment is wrong rather than that both functions are underperforming, and the giveaway is that the accounts which do activate look nothing like the ones in the pipeline. Fixing sales execution against the wrong segment produces more of the customers who churn.

From our own category research: in our analysis of this category's search results, Google restructured them across the March–April 2026 core updates: vendor pages fell out of the top 20 while informational pages held. Buyers now research questions like this one through answers rather than vendor sites, which is a reasonable instinct here, because every vendor in either function has a structural view about which of the two the answer is.

What do founders usually get wrong?

They run the diagnosis on aggregate numbers and reach for the explanation that requires the least disruption. Aggregate win rate across segments hides the fact that one segment converts at three times the other. Aggregate activation hides that the accounts failing to activate are concentrated in a use case nobody designed for. The aggregate answer is almost always "we need more pipeline", because it is the only conclusion that requires no existing decision to be reversed.

The second mistake is asking each function to diagnose itself. Sales will report a product gap and product will report a qualification gap, both sincerely, because each is describing the part of the failure they can see. This is why the interviews matter more than the numbers: the numbers tell you which cell you are in, and only the conversations tell you why.

The aggregate answer is almost always "we need more pipeline", because it is the only conclusion that requires no existing decision to be reversed.

How do you run the two-week version?

Roughly ten to fifteen focused conversations, split deliberately: four or five recent wins, four or five closed-lost, and four or five churned accounts, weighted toward the ones that left quietly. One person runs all of them so the pattern accumulates in a single head. The script asks what they were trying to accomplish, what they compared you against, what nearly stopped the purchase, and, for the churns, what specifically had changed by the time they decided to leave.

A sample this size is usually enough to surface the dominant pattern separating a product problem from a sales problem, though saturation depends on how varied your segments are. It is not enough to build a prioritization base, and it should not be presented as one. The full-scale version of this program runs to 60–80 customer and prospect interviews and produces something you can prioritize a year against. Treat the fortnight as a triage step whose only job is deciding which function owns the next quarter.

What should you do for each diagnosis?

Sales problem: work the qualification filter, the discovery script and the competitive comparison before touching headcount, because adding sellers to a broken motion multiplies the motion. Product problem: run the full evidence program, name the two or three causes with counts attached, and stop something on the roadmap to fund the fix. Positioning problem: pick the segment where activation is strong, rewrite the qualification criteria to match it, and accept the smaller pipeline for a quarter. Neither: look at pricing and market before assuming an execution failure exists at all, and if pricing is where it lands, hiring a monetization product manager, or fixing pricing without one covers that route.

If the diagnosis lands on product and nobody internal has the standing to act on it, the options run from a scoped project at a day rate of around $1,200–$2,200 to embedded part-time leadership. Sivan Kadosh, a fractional Chief Product Officer for B2B SaaS companies between $2M and $15M ARR, works the second of those at $8,000 a month for 20–25 hours. If you have someone who can carry the finding, buy the cheaper option.

If the answer is a product problem, the pattern where output stays high while retention falls is covered in shipping constantly while churn keeps rising, and where this sits in a wider stall is in product leadership when growth stalls.

Three numbers put you in one of four cells, and roughly 10–15 focused conversations usually surface why you are in it. Run the numbers by segment rather than in aggregate, and do not let either function diagnose itself. The fortnight is triage, not research: its job is to decide which team owns the next quarter.