Guide
How to compare vendors with ChatGPT or Claude
ChatGPT or Claude can compare two to five vendors fairly if you fix the rules before it sees the prices. Ask it to propose four to seven criteria with weights that add up to 100, confirm them yourself, and treat must-haves as pass or fail, so a vendor whose key feature is only "on the roadmap" is out. Then have it score each vendor from 1 to 5 with the evidence behind each score, mark what only the vendor claims as unverified, and work out the total cost over the term at the same number of users. It ends with one recommendation, the two reasons behind it, and what to check before you sign.
Short answer
- Weights first, confirmed before scoring.
- Must-haves: pass or fail; a roadmap promise fails.
- Every score cites its evidence; vendor-only claims are unverified.
- Cost over the whole term, same users for all.
The brief to copy
Vendor comparison brief
"Compare these vendors for [need], for [number] users over [term].
First propose 4 to 7 criteria with weights adding to 100, no two the same, and stop so I can confirm them. Must-haves: [list]. A vendor that fails one is out, with the evidence; a feature that's only promised fails.
Score each remaining vendor 1 to 5 per criterion and name the source of each score: trial, reference call, contract, or vendor claim. Mark vendor-only scores (unverified). Weighted total = sum of score × weight, divided by 5.
Total cost over the term: fees, setup and migration, add-ons needed for the must-haves, and exit (data export, notice). Same user count for all; a price I didn't give is [check: price]. List 1 to 3 risks per vendor. Recommend one vendor with 2 reasons; if the top two are within 5 points, say it's close and name the tie-breaker. End with what to verify before signing.
What we have on each vendor: [paste]"
A worked example
Illustration: a made-up business. A cleaning company with 40 cleaners picks a scheduling app for 2 years. The must-haves are an Android app the cleaners can use without a laptop and a payroll export. ShiftNest charges 4 per user a month, with no setup fee; both must-haves passed in a week's trial with five cleaners. Rotaly charges 3, but its Android app is "coming in spring" according to its sales rep, so it's out. CrewClock charges 5; its payroll export worked in the trial, its Android app is described only in its docs, and a reference customer praised its support. The operations lead moved 5 points from data to support in the AI's proposed weights, then confirmed them:
| Criterion (weight) | ShiftNest | CrewClock |
|---|---|---|
| Must-have fit (30) | 5: both tested in the trial | 4: export tested; Android app per docs (unverified) |
| Cost (25) | 4: 3,840 over 2 years (4 × 40 × 24) | 3: 4,800 over 2 years (5 × 40 × 24) |
| Ease for cleaners (20) | 4: trial with 5 cleaners | 4: trial |
| Support (15) | 3: "replies within 4 hours", vendor claim (unverified) | 5: reference call |
| Data and security (10) | 3: security page (unverified) | 4: security page (unverified) |
| Weighted total | 81 | 78 |
The recommendation is ShiftNest: both must-haves proven in the trial, and 960 less over two years. At 81 against 78 it's close, so the tie-breaker is a reference call to check ShiftNest's support before signing, and its data location goes on the list to verify. The operations lead and the finance manager scored alone first, then agreed the final scores together.
What stays with you
- Confirm the weights before anyone opens a quote.
- Test the must-haves yourself; a sales promise isn't a pass.
- Call a reference before scoring support.
- Two deciders score alone first, then agree.
- Read the renewal, notice and export terms in the contract before signing.
Next step: a project that remembers your criteria
Vendor choices come back every renewal. A Claude project for operations work keeps your criteria, past comparisons and contracts in one place, so the next comparison starts from them.
Skills for this
- Operations and adminVendor comparison with weighted scoresCompares 2-5 vendors on weighted criteria you agree first: must-haves, evidence-based scores, total cost, risks, and one recommendation.
- FinanceSaaS spend reviewReviews SaaS subscriptions: owners, seats used vs paid, overlapping tools, renewal and notice deadlines, and cancel, downgrade or keep with savings.
- ManagementDecision memoWrites a short narrative decision memo: recommendation first, context, options with trade-offs, risks, and the ask with a decider and deadline.
- Legal basicsContract summary with key datesSummarizes a commercial contract into a one-page brief: fees, term, auto-renewal and notice dates, exit, liability, IP, data, and unusual clauses.
Questions
Can ChatGPT compare vendors for me?
It can do the comparison work from what you give it: quotes, pricing pages, trial notes and reference calls. Have it propose criteria and weights, confirm them before it scores, and ask for the evidence behind every score. Anything only the vendor says, such as a sales deck's response time, should be marked unverified. The decision and the contract stay with you.
How do you weight criteria in a vendor comparison?
Before anyone looks at prices or scores. Pick four to seven criteria with weights that add up to 100 and differ from each other, and treat true must-haves as pass or fail. The UK Cabinet Office's bid evaluation guidance asks buyers to test the evaluation model before finalising it, and to exclude any bidder that misses a mandatory threshold.
What is whole-life cost?
Everything the choice costs over the contract term, not just the monthly price: fees, setup and migration, add-ons you need to meet the must-haves, admin and training time, and what it costs to leave (data export, notice period). Compute it at the same number of users for every vendor, as the UK Sourcing Playbook does when it asks buyers to assess whole life cost.
AI skill
Have your AI compare vendors on what matters to you
Put this skill in AI Skill Pod and everyone on your team gets it in their Claude, Cursor or ChatGPT (Pro, Business, Enterprise): one invite link, free to start. Vendor comparison with weighted scores has your AI compare vendors on criteria you confirm before it scores: must-haves as pass or fail, a 1 to 5 score with its evidence for each criterion, the total cost over the term at the same volume, risks per vendor, and one recommendation with what to check before signing.
14 days free · No card needed
Sources
- UK Cabinet Office: Bid evaluation guidance note (May 2021) — scenario-test the evaluation model before finalising it; if a threshold is mandatory, exclude any bidder that doesn't meet it; two or more evaluators and a moderator who helps them reach a consensus (read 2026-11-05)
- UK Cabinet Office: The Sourcing Playbook (June 2023) — assess the whole life cost, not only the initial price (read 2026-11-05)