How to Compare Supplier Quotes with an AI Chatbot

How to use the prompt

Paste the prompt into your AI chatbot, answer the questions it asks, then paste your quotes. It returns a normalised comparison, a ranked recommendation, the assumptions it made, and the questions to put back to your suppliers. Then you verify and decide.

1. Open a capable AI chatbot ChatGPT, Claude, Gemini or Microsoft Copilot all work well. For commercially sensitive pricing, use a business or enterprise setting with clear data handling, or a tool that does not train on your inputs.
2. Paste the prompt and answer its questions Copy the prompt below into the chatbot. It will not ask for your quotes yet: first it asks a short set of questions about what you are buying, your criteria and your currency. Answer them so the comparison reflects what actually matters to you.
3. Paste or upload your quotes When the chatbot asks, give it the quotes in any format: email text, copied PDF text, or a spreadsheet. They do not need to match each other, because normalising them is the whole point.
4. Read the comparison, assumptions and follow-ups You get a like-for-like table in one currency, total landed cost and cost per unit, a ranked recommendation against your weighted criteria, a list of everything the AI had to assume, and the questions to send back to each supplier.
5. Verify before you decide Spot-check the largest line items and anything the AI flagged as assumed against the original quotes. A single misread unit can flip the winner, so treat the AI table as a fast first pass, not the final word.

The prompt (copy and paste)

This prompt is industry-agnostic. It runs in stages: it asks about your job and your criteria first, then asks for the quotes, then normalises, compares, scores and recommends, with a hard rule not to invent numbers.

You are an experienced procurement analyst. Your job is to help me compare supplier quotes and reach a decision I can defend. Work through the stages in order and do not skip ahead.

STAGE 1: Ask me questions first.
Before I give you any quotes, ask me the questions below, then stop and wait for my answers. Keep it to a single message.
1. What am I buying? (a product or a service, with a short description)
2. What quantity and unit of measure matters to me? (for example per unit, per pallet, per hour, per licence, per 1,000)
3. What base currency should every total be shown in? If any quote is in another currency, ask me for the exchange rate to use. Do not invent exchange rates.
4. What are my evaluation criteria, and how would I weight them out of 100? (for example price, lead time, quality, warranty, support, sustainability, payment terms, supplier risk)
5. Are there any must-haves or deal-breakers? (specifications, certifications, delivery deadlines, compliance requirements)
6. What is my timeframe, and how many quotes am I comparing?
7. Is there anything unusual about this category that you should know?

STAGE 2: Ask me for the quotes.
After I answer, ask me to paste or upload the quotes. Tell me I can provide them in any format (email text, copied PDF text, a table) and that they do not need to match each other.

STAGE 3: Normalise and compare.
Once you have the quotes:
- Pull every quote into the same structure: the same line items, the same unit of measure, the same currency (my base currency) and the same quantity basis, so the numbers are genuinely comparable.
- Work out each supplier's total landed cost: goods, plus shipping or delivery, plus any fees, plus tax. Then work out the cost per unit.
- Build a like-for-like table: one row per line item and per cost component, one column per supplier, everything in my base currency.
- Do not guess. If a figure is missing, ambiguous, or you had to convert or assume anything, flag it clearly and list it. Never fill a gap with an invented number.

STAGE 4: Surface the real differences.
Highlight where the quotes differ beyond headline price: exclusions, hidden fees, minimum order quantities, lead times, payment terms, warranty and anything else that affects total cost of ownership. Say explicitly if the supplier with the lowest headline price is not the cheapest once everything is included.

STAGE 5: Score and recommend.
Rank the suppliers two ways: by total landed cost, and by fit to my weighted criteria from Stage 1. Give me a clear recommendation with the reasoning, and name the top two or three trade-offs I am making. State your confidence, and say what would change your recommendation.

STAGE 6: Give me the follow-ups and a record.
List the specific questions I should send back to each supplier to close the gaps you flagged. Then produce a short decision record I can save: what I chose, the criteria that drove it, and what I traded off.

Rules throughout: be accurate rather than fast. Do not invent prices, exchange rates or terms. If you are unsure, ask me or flag it rather than guessing. Show your working for any total. Keep the language plain.

Two small habits make it noticeably better. First, answer the Stage 1 questions properly: the weighting you give price versus lead time versus risk is what turns a number-crunch into a decision. Second, when it flags an assumption, do not wave it through: those flags are exactly where a quote is hiding something.

A word on your data

Supplier quotes are commercially sensitive. Before you paste them in, make sure you are using a tool and a setting you trust with that data.

Use a business or enterprise plan with clear data handling, or a tool that does not train on your inputs. If you are not sure, strip out supplier names and identifying details first, or keep the exercise to non-sensitive figures. The prompt keeps your data in the conversation; the responsibility for where that conversation runs is yours.

Where a chatbot stops and a colleague begins

A chatbot with a good prompt is excellent for a one-off comparison. The limit shows up next cycle, when it has forgotten everything and you re-paste your criteria, your suppliers and last quarter's prices from scratch.

That is the difference between a tool you drive and a colleague who owns the job. A general chatbot starts cold every time. An AI colleague (a Synth) keeps the context: your comparison criteria, your supplier history and your past pricing, working inside the tools your team already uses, with an audit trail of every decision. So each round gets faster instead of resetting, and the know-how does not walk out when your best buyer moves on. If that sounds useful for your procurement, book a call.

The third direction people go is a procurement or spend platform that owns sourcing, approvals and purchase orders end to end, with quote comparison as one feature among many. That is a heavier commitment than a prompt or a Synth, and it fits a different size of buyer. Our guide to Coupa alternatives and procurement software for mid-sized teams works through what those platforms do and who each one suits.

Related guide: Coupa alternatives: procurement and spend software for mid-sized teams →