Choosing a supplier is a comparison problem wearing a conversation problem's clothes. You run five calls; each vendor quotes a different price, a different lead time, a different payment structure, and a "but at volume we can do better" that never quite pins down. By decision day the five calls have merged into notes that don't line up, memory that leans toward whoever called last, and a nagging sense that vendor three offered something you can't now reconstruct. Then, a month after signing, vendor two "doesn't recall" the net-30 terms they agreed to out loud.
The calls were fine. The problem is that a decision needs a grid, and conversations don't produce one — unless you make them.
Key takeaways
- Vendor selection fails on incomparability: five conversations in five shapes can't be weighed against each other, so the decision drifts toward recency and charm.
- Fix the shape upfront: define the tender's fields — price, lead time, payment terms, volume discount, warranty — and let each call fill the same grid (product fact below).
- Verbal concessions stop evaporating: "you mentioned net-30" becomes a timecoded quote from the recording, not a memory to argue about (product fact).
- Honest scope: the product fills structured fields per call and answers questions from the archive — assembling the final comparison view is your step, and a quote is leverage, not a court exhibit.
Why five good calls don't make a decision
Each negotiation is optimized by the vendor to look good on the vendor's strongest axis: the cheap one leads with price, the fast one with lead time, the premium one with service. Left in conversational form, they're apples, oranges, and a service-level agreement. The buyer's job — weighing them on the same axes — is exactly what the calls don't hand you, and reconstructing it from notes after five calls is where selection quietly becomes vibes.
Two failure modes follow. Drift: without a grid, the decision tilts toward the call that's freshest or the rep who was most likeable. Leakage: the verbal sweeteners — the volume discount, the waived setup fee, the extended terms — live only in memory, and memory is contestable at signing time.
Structure the series like the decision
The fix is to decide the comparison axes before the calls, not after. Define the tender's fields in plain language — "unit price," "lead time," "payment terms," "volume discount," "warranty and returns" — and every processed call fills the same fields from what was said (product fact: custom insight fields defined in natural language, auto-filled per meeting). Five conversations become five rows with the same columns.
This is the same series-into-structure mechanism that research teams use for customer discovery interviews and founders use for investor pitches — pointed at procurement. The domain changes; the discipline is identical: decide what you're comparing, then let extraction enforce it.
The verbal concession, on the record
Procurement runs on spoken commitments that precede the paperwork — and the gap between "agreed on the call" and "in the contract" is where money leaks. Record the negotiation (with the vendor's knowledge — a routine courtesy on a business call) and the concession is preserved: the transcript holds "we can do net-30 at that volume," timecoded, one click deep (product fact). The negotiation summary also goes to your manager as a record rather than a retelling, and follow-up items — "request the spec sheet," "decision by Friday" — come out as tasks (product facts).
State the boundary plainly: a recorded quote is leverage — it settles the honest "did we say that?" the same day, and makes a vendor's "I don't recall" untenable in a negotiation. It is not, by itself, a contract or a legal exhibit; the binding terms live in the purchase agreement. The record's job is to make sure the paperwork matches what was actually agreed.
What you assemble, and what fills itself
Honest mechanics: the product fills the structured fields per call and answers questions across the archive — "which vendors offered volume discounts?" comes back with quotes (product fact). Laying those fields side by side into the final comparison — the grid you take to the decision — is your step; there's no built-in tender-scoring dashboard, and calling the field-extraction a finished comparison table would overstate it. What the pipeline guarantees is that the five calls arrive as structured, quotable rows instead of five incomparable memories.
Sources and method
Product facts (custom insight fields defined in natural language and auto-filled per meeting, timecoded transcripts, summary sharing, task extraction, archive-wide questions answered with quotes) describe MeetResult as documented in the product catalog at the time of writing. No automatic vendor-scoring or comparison-table feature is claimed — final comparison assembly is manual. A recorded quote is characterized as negotiation leverage, not legal proof. Recording-consent obligations remain the recorder's responsibility. No statistics are cited or invented.
Related: Customer discovery interviews: structure the series · Run your fundraise like research · How to turn a meeting recording into tasks · Five vendors, one comparison, zero forgotten discounts (use case)
Before your next round of vendor calls, write the five fields you'll compare on. Process the first call in @meetresultbot and watch the grid start filling itself. New accounts include free processing minutes.