We needed a new data warehouse tool. Four vendors on the shortlist, three demo calls a week clogging everyone's calendar, and a committee of six people who all wanted a meaningful say and exactly none of whom wanted to be the one to build the comparison spreadsheet. The traditional path was clear and deeply miserable: a fifteen-column comparison sheet that takes two weeks to assemble, then three rounds of revisions as people remember criteria they forgot, and finally a meeting where half the committee quietly admits they skimmed it in the elevator on the way in.
What I did instead of the spreadsheet
After each demo I dumped my notes into a shared doc. Not pretty notes — rough, fragmentary bullets about features, pricing hints the rep let slip, what genuinely impressed me, and what made the back of my neck prickle. After the fourth and final call, I pasted all four sets of notes into the Vendor Comparison Memo from Messy Demo Notes prompt on GPT 5.2, along with our five evaluation criteria listed in strict priority order. I chose GPT 5.2 deliberately, because this is fundamentally a reasoning task and not a formatting task — the model has to weigh partial, conflicting evidence and resist the temptation to overclaim, and GPT 5.2 handled that ambiguity with exactly the right amount of skepticism.
Twenty seconds later I had a clean comparison table with ✅, ⚠️, and ❌ scores across every criterion, a summary analysis that actually weighed our top priorities against each other, and a recommendation backed by three concrete, defensible reasons. The whole thing was three pages. Readable on a phone. The exact, beautiful opposite of the fifteen-column horror I had been dreading building.
The blank cells did the real work
Here is the moment that justified the entire exercise and changed how I think about evaluations. Two of the four vendors came back with ⬜ Unknown marks on our single most important criterion — data residency compliance — because, it turned out, neither sales rep had actually answered the question during the demo. They had both smoothly talked around it, the way good salespeople do. I had not noticed the gap at all, because my notes simply did not contain an answer, and the human eye glides right over an absence. It sees what is there, not what is missing. The prompt does the opposite. Its hard rule is to mark a missing criterion Unknown rather than guess or give the benefit of the doubt, and that one row of blank cells completely reframed the decision.
We fired off pointed follow-up emails that same afternoon and had answers back within forty-eight hours. One vendor's answer eliminated them on the spot — they could not meet our residency requirement and had been hoping we would not ask directly. We almost did not.
- Total time from the last demo to a finished, circulated memo: under two hours
- Committee read rate hit six of six, because it was three readable pages instead of fifteen intimidating columns
- Two critical data gaps surfaced that all six of us had completely missed in real time
- Our final decision landed eight full days ahead of the original timeline, which finance noticed and liked
Why I actually trust the scores
I gave it a deliberately tricky note once, as a test — 'no SSO yet, but the rep said it's on the roadmap for next quarter' — purely to see whether the model would be charmed by the magic word 'roadmap' the way humans are. It marked the security criterion ⚠️ Partial, not ✅, which is precisely the call a seasoned, slightly cynical procurement person would make. Roadmaps are promises, and promises are not features you can buy. The fact that this prompt consistently scores what exists today rather than what is sweetly promised for tomorrow is the entire reason I am comfortable letting it near a real, six-figure purchase decision.
Takeaway
I now run this for every vendor evaluation over five thousand dollars a year, and our finance team has quietly started requiring the memo as a standard purchase-approval artefact. It has also changed how I take notes during demos: knowing the prompt will mark unanswered criteria as Unknown, I now actively listen for which of my priority criteria the rep is dodging, and I jot a literal '(no answer)' next to each one so the gap is unmissable later. The tool trained me to be a better evaluator, which I did not expect from a glorified note-formatter. When the memo points to a genuinely close call between two strong options, I escalate it into my Decision Doc: Options, Risk, and a Clear Recommendation prompt to force a clean tie-break. Grab the Vendor Comparison Memo from Messy Demo Notes prompt on Prompt Dock and feed it your own pile of demo scribbles. Then watch the blank cells, because they are quietly trying to tell you the thing the sales rep would not.