Check Problem Frequency Before Prioritizing a Solution
Define and count a buyer-reported problem with a bounded event log, denominator and missing-data note before calling an anecdote a recurring business issue.
10/23/20263 min read
One difficult incident can reveal a real problem without showing how often it happens. A seller who turns a memorable example into a recurring monthly loss may overstate the need and recommend the wrong level of intervention. Establish the frequency question separately from the description of the incident.
The aim is not to demand a large study before helping. It is to use evidence proportionate to the decision and label its limits. A rare but serious failure may deserve attention, while a frequent minor inconvenience may need only a small process adjustment. Frequency alone does not establish priority.
Define the event consistently
Agree what counts as an occurrence. Reporting delayed could mean a late source file, a missed internal review or a report published after the intended date. If those events are mixed together, a count will not explain the operational problem.
Choose a relevant observation period and the population of opportunities for the event to occur. Record both, even when the data is incomplete. Three delayed handoffs among four reviewed cases means something different from three among four hundred, and neither automatically represents the whole organization.
Worked example: recurring reassignment
Fictional buyer Brook Services describes a request that was reassigned twice before reaching the correct team. The seller wants to call the issue widespread. Instead, both sides agree on an event definition: a request returned to the intake queue because the assigned team's responsibility did not match the documented category.
Using an approved, minimized process log, the buyer reviews twelve recent requests from one team. Two match that definition, one has incomplete history and nine do not. These are invented teaching numbers, not evidence about a real company or a recommended sample size.
The completed record reads: observed group, twelve requests from one team in the selected period; confirmed matching events, two; unresolved, one; excluded events, ordinary planned transfers; generalization, not established for other teams or periods. The next step is to examine the two cases for a shared cause, not multiply their count into an annual financial claim.
Include what the record cannot show
A process log may miss informal corrections or changes made outside the system. Ask how records are created and whether the selected cases systematically exclude difficult periods. Do not treat a clean database as proof that no problem exists, or an anecdotal complaint as proof that the database is wrong.
If the buyer has no usable records, a short prospective observation may be more reliable than asking people to reconstruct months of incidents from memory. Define the observation and obtain appropriate approval before collecting additional information. Use roles and event states where identities are unnecessary.
Combine frequency with consequence
Record what happens when the event occurs: extra work, delay, confusion, service disruption or another buyer-confirmed effect. Keep severity and frequency separate. Do not automatically convert minutes into cash savings if the buyer's actual costs would remain unchanged.
A decision can therefore read: observed occasionally in this limited group, low direct burden, simple clarification worth testing. Another can read: frequency unknown, potentially serious consequence, specialist review warranted. Neither requires a fabricated percentage or a universal threshold for when a buyer should purchase a service.
Choose the next evidence step
Ask which uncertainty prevents a sensible decision. It may be the event definition, the completeness of the log, the cause of the two confirmed cases or the consequences when they occur. Research that specific uncertainty rather than collecting every available field.
The final artifact is an event definition, bounded count, relevant denominator, missing-data note and proportionate next action. Provenance and privacy principles help preserve where the observations came from and minimize unnecessary collection. The worksheet itself is an editorial method, not a certified measurement standard.
A useful discovery outcome can be that the problem is less frequent than initially assumed. It can also be that the evidence is too weak to estimate frequency yet. Both conclusions support better decisions than a confident recurring-pain claim built from one memorable story.