Ranking Customer Evidence for Product–Market Fit Signals and Gaps

Hello. Your ICP identifies the lending institutions you believe are best suited to your collection workflow, and your role map identifies what operations and collections leaders need to see before supporting it. The next question is harder: which of those beliefs are backed by customer behavior, and which still rest on encouraging conversations?

This lesson gives you a practical way to rank product–market fit evidence in a company that already has customers. You’ll distinguish genuine adoption from interest, look for patterns within your target segment, and identify the gaps that matter most before making stronger claims about value. Allow about 40 minutes, including the video, readings, and a short review of your own evidence.

Rank what customers do before what they say

Imagine a head of collections says, “This could transform our operation.” That is worth investigating, but it is not equivalent to deploying the workflow, using it after the initial trial, and asking to extend it to another portfolio. The distinction is not whether the leader is sincere. It is whether their institution has committed effort, accepted operational change, and continued to find value.

Watch this excerpt from Y Combinator’s Peter Reinhardt on Finding Product Market Fit at Segment. It draws a useful line between requests from people who already use a product and suggestions from people who never adopt it.

For your purposes, product–market fit is not a single enthusiastic account. It is a pattern in which a recognizable type of institution adopts the workflow, continues to use it for an important job, and finds enough value to sustain or broaden that use. Commercial commitment and measurable outcomes make that pattern more convincing.

A useful starting order for evidence is:

1. Continued, meaningful production use, especially across more than one suitable account: eligible cases actually pass through the workflow, and use persists beyond launch support.
2. Commercial commitment tied to that use: renewal, expansion, or a decision to fund a wider deployment. A signed pilot matters, but a paid pilot that never becomes operational is a weaker signal.
3. Verified operational or business value: a measured change the buyer cares about, with a clear baseline and scope. Check whether other changes could explain it.
4. Specific requests from active customers: for instance, a request to cover another queue, accompanied by a named owner and an implementation discussion.
5. Stated interest alone: positive meetings, feature requests from non-users, or a willingness to “stay in touch.”

This is a guide to strength, not a rigid score. A well-documented improvement in one account may tell you more about value than a vague renewal reason; several quiet renewals may tell you more about durability than one striking case study. Keep adoption, realized value, and willingness to pay visible as separate signals rather than combining them into one flattering story.

Find the segment behind the apparent signal

A company-wide average can conceal the institutions for which the product works particularly well—or particularly badly. Suppose usage is strong among lenders with large, standardized early-stage queues but limited among lenders with highly varied case handling. The average alone would obscure an important refinement to your ICP.

First Round Review’s account of how Superhuman studied product–market fit offers a method for listening to active users, finding a segment that values the core benefit, and separating promising improvement requests from requests that would pull the product away from that benefit.

The article describes a survey that asks active users how they would feel if they could no longer use the product. Its reported 40% “very disappointed” benchmark can be a useful prompt, but it should not determine whether your enterprise product has fit. A lender may have many users and only a few people who authorize purchase; a mandatory workflow can be used frequently without being valued; and a small, selectively surveyed pilot can produce a misleading percentage. Record who answered, how recently they used the product, which accounts they represent, and how many responses you received. Never substitute a percentage for that context.

The more transferable lesson is segmentation. Ask whether your strongest customers share operational conditions from your ICP: similar case types, queue volumes, handoffs, buying triggers, or constraints. Then examine customers who value the same core benefit but use less of the workflow. Their obstacles may expose a fixable gap. A customer requesting an entirely different benefit may instead be outside your current focus.

The image suggests a quarterly rhythm. Use a cadence that matches your customers’ deployment and decision cycles; some changes will take longer than a quarter to show whether use is durable. Nor should you ignore a “not disappointed” response from a compliance or operations reviewer who controls whether deployment can continue. Low enthusiasm from an end user and an unresolved requirement from an approver are different kinds of evidence.

Check whether the pattern lasts

A snapshot of active accounts cannot show whether new deployments become durable. Stripe’s explanation of cohort analysis provides a simple corrective: compare groups that started under similar conditions at the same point in their customer lifecycle, rather than mixing mature customers with new pilots.

For a collection workflow, one useful cohort could be institutions that entered production in the same quarter, assessed after a comparable number of months. Another could compare eligible portfolios within similar institutions. Define “active” before looking at results: a login is not the same as eligible accounts being processed regularly. A practical view might track the share of eligible cases handled, the number of staff interventions or exceptions, whether the agreed workflow remains in use, and whether the institution renews or expands.

Be careful with comparisons. A change in recovery performance might reflect a different account mix, a policy change, seasonality, or a simultaneous staffing initiative. Cohort data shows where a pattern appears; conversations and a defensible comparison help explain why. If volumes are small, show the underlying accounts and dates rather than presenting a smooth percentage as though it represents a broad market.

Turn evidence into a ranked list of signals and gaps

Here is a hypothetical evidence review for an agent workflow. The entries illustrate how to reason; they are not claims about your customers.

The first two are stronger fit signals than the fourth because customers have changed their operations. The third is potentially consequential but cannot yet substantiate a recovery claim. Negative and ambiguous observations belong in the review too: a deployment that stalls at exception handling may be more important to investigate than another positive demo.

To decide which gaps matter most, use two tests:

• Decision impact: If the answer went against you, would it change your ICP, product priorities, or the claim you can responsibly make to buyers?
• Reach: Does the uncertainty affect one unusual account, or multiple institutions in the segment you intend to serve?

For the example above, establishing whether routine use produces a measurable advantage over the current approach is likely a higher-priority gap than polishing the demo. If approval repeatedly stalls because exception handling is unclear, that becomes a high-priority gap even when users like the workflow.

Spend 8–10 minutes making a compact evidence ledger for your own ICP. Use accounts rather than individual comments as the unit of comparison, so ten remarks from one institution do not masquerade as ten independent customer signals. For each of your top five observations, record:

Rank the observations by strength and relevance to your target segment. Mark the two strongest signals and the two gaps most likely to change a buying or product decision. If you lack comparable outcome data, write “outcome not yet verified”; if authority is unclear, write “buyer not confirmed.” Those are useful findings, not weaknesses to conceal.

This review will also discipline how you present the company. “Two suitable institutions continued using the workflow after launch” is a bounded statement if your records support it. “Our technology improves collections outcomes for lenders” requires substantially more evidence. Credible positioning starts with the former and makes clear what still needs validation.

What to carry forward

Strong product–market fit evidence combines meaningful use, durability, value, and commercial commitment within a recognizable customer segment. Rank observed customer actions above polite interest, examine patterns across comparable accounts and periods, and keep alternative explanations visible. The most important gap is not necessarily the most requested feature; it is the uncertainty most likely to alter your customer focus, product decision, or buyer-facing claim.

Next, you’ll turn those gaps into a structured customer interview guide. Your evidence ledger will tell you which adoption drivers, realized benefits, unmet needs, and expansion conditions the interviews need to test.