The setup
Leadership had a legitimate complaint: reports were unreliable because reps skipped fields inconsistently. The fix, designed entirely by sales ops and IT with no field-rep involvement, was to make 22 fields required before a deal could move to the next stage — everything from referral source down to a free-text "next steps" field and a picklist for a client's internal approval process that reps often didn't have visibility into at all.
It shipped as a single release with an email announcement. No pilot group, no phased rollout, no session with reps to explain why each field mattered or to hear which ones were actually impossible to fill in accurately at that stage of a deal.
The collapse
Reps didn't complain up the chain first — they complained to each other, and adapted. Within days, the "client's internal approval process" field, which reps often genuinely didn't know the answer to at the required stage, was being filled with "N/A," "TBD," or a repeated default value just to get past the gate and keep working. The same happened with the free-text "next steps" field: reps typed a single character or copy-pasted the same boilerplate sentence across dozens of records.
The behavior spread fast because it worked — the system didn't distinguish between a thoughtful answer and a placeholder, so there was no friction stopping it and no incentive not to do it. Within six weeks, an audit of records advanced past the gate found roughly 60% contained at least one placeholder or copy-pasted value in a "required" field. The data set leadership had made mandatory specifically to improve reporting accuracy was now, in the fields that mattered most, actively lying to every report built on top of it — arguably worse than before, because the fields now looked complete instead of visibly blank.
The autopsy
Root causes on record
- Mandatory fields were designed without the people who'd fill them in. Several fields asked for information reps didn't reliably have at that stage.
- No pilot or phased rollout. 22 required fields shipped at once, all-or-nothing, to the entire org.
- Required does not mean valid. The system enforced presence of a value, not plausibility — nothing flagged "N/A" or a one-character entry as suspect.
- No feedback loop from reps back to the field design. The people best positioned to say "this field can't be answered honestly at this stage" had no channel to say so before launch.
- The metric being optimized was field completion, not data accuracy. Those look identical in a "% complete" dashboard and are not the same thing at all.
Recommendation pending
Editor's note: this slot will point to a change-management or user-adoption resource for rolling out data-quality rules with a pilot group and real field feedback instead of a single big-bang release.
What the post-mortem actually changed
Sales ops cut the mandatory list from 22 fields down to five that reps could reliably answer at each stage, added lightweight validation (an "N/A" entry now triggers a soft warning rather than being silently accepted), and ran the next round of changes through a two-week pilot with a small group of reps before rolling out org-wide. Field completion rate went down on paper. Actual usable data went up, which was the entire point in the first place.