Academics investigating contract cheating had to email us and wait while someone manually recorded a video. I found the gap in my own competitor research, pitched it, shipped it — and it became a billable module that helped convert a pilot into an enterprise contract.
Cadmus could reconstruct exactly how a piece of work was written — a continuous, snapshot-to-snapshot replay.
Powerful evidence in a contract-cheating case.
No surface. Every request went through a chain of people.
No ticket, no queue, no waiting on someone else's availability — and no internal staff time consumed per investigation.
1
2
3
Sandbox tenant with test students; student contact detail redacted.
An academic with a time-sensitive integrity concern was blocked on someone else's inbox.
Every investigation consumed staff time, so cost scaled with how much customers valued it — exactly the wrong shape.
Timelines became recordable for every assessment, not just where the setting was on. Without it, self-service would have silently failed on much of the work academics wanted to examine.
In the comparison I maintain — 40+ products, quarterly — submission-replay was a row where a competitor had a tick and we didn't.
A competitor already surfaced how a submission was constructed over time, as a product surface academics could reach themselves.
More detail than a text-comparison tool — but not exposed to the people who needed it, so it never counted in an evaluation.
1
2
3
4
Sandbox exam with placeholder question content; internal work identifier redacted.
Not a missing capability — a missing surface. We were losing evaluations on a feature we already had, because it lived in an internal tool. We weren't building submission replay; we were exposing it — which is what made it cheap enough to justify.
The one project in my portfolio that closes the full loop from competitive research to money on the table.
Quarterly refresh of my competitor matrix surfaced submission-replay as a row where Turnitin Clarity led and Cadmus was blank.
The commercial team recognised it as a live differentiator in deals and asked for prioritisation.
Timeline access straight from the class list, plus recording enabled for all assessments regardless of tenant setting.
Delivered with one engineer, released behind a flag, then fixed what real first use exposed.
Packaged as a billable module rather than given away in the base product.
The flag was commercial, not technical. It let the capability be packaged and priced rather than absorbed into the base tier — how a support workflow became a revenue line.
Work Playback was part of what took Massey from pilot to enterprise. A gap I found in a spreadsheet became a feature that helped close a contract — the clearest line I can draw between product research and commercial outcome.
The first release shipped without timestamps — missed in requirements, survived QA, caught by Support within days.
For an integrity investigation timestamps aren't decoration — they're how you correlate the replay against an activity report.
Fixed fast, but it shouldn't have reached users: I wrote the spec from the shape of the feature rather than the job the evidence has to do.
Treating the competitor matrix as a live instrument rather than a document. No customer request, no roadmap workshop — a quarterly refresh, spotted early enough that Sales could use it while deals were open. Research only pays off if someone reads it on a schedule.