1. Problem
What takes time, creates errors, or stalls cases today, and for whom?
Most bad quotes are not caused by bad vendors, but by vague requests. Ask for a price on something undefined and you get a price on the vendor guess - and guesses are rarely comparable. This guide shows what your request should contain, with a template to copy straight into the email, what a good quote must answer, which red flags to watch for, and how to compare quotes that look different.
Last updated: September 26, 2026
A vendor can only price what they understand. Send three sentences about wanting AI and they must guess at scope, data, integrations, and risk - with each vendor guessing differently. The result is quotes that cannot be compared, and a project that starts in misunderstanding. A good request costs you a couple of hours, but saves weeks of clarification and rework later. It also forces internal agreement on what you actually want to achieve, before money and calendars are at stake.
Start with the problem: what takes time, creates errors, or stalls cases today, and who feels it? Then describe the job: what should the solution do, for whom, and what should be different afterwards? List the systems involved, and say something honest about the data - what exists, where it lives, and who knows what is correct. Mention user groups and access needs, sensitive data or approval requirements, and any timeframe you have. End with what you want answered: fixed scope and assumptions, testing, ownership, operations - and what the quote explicitly excludes.
A good quote mirrors the request point by point: here is the scope as we understand it, here are our assumptions, here is what we do not know yet. It describes how quality is tested - which examples, who approves, and what counts as good enough. It says who owns code, design, and data, who operates after launch, and what happens on change. It tells you what the vendor needs from you, and how much of your peoples time to budget. And it is honest about risk: what is uncertain, what can go wrong, and how is that handled? A quote containing only price and deadline answers none of this - then you are buying a surprise.
Be sceptical of fixed prices on things the vendor has barely seen - especially if your data is messy or requirements are strict. Be sceptical of quotes with no test plan: no examples, no approval, no definition of good enough. Be sceptical of unclear ownership: who owns the code if you part ways, and in what format do you take your data? Be sceptical of a missing change routine: every project changes, and without rules you end in fights or standstill. And be sceptical of vendors who need nothing from you - no access, no time, no decisions. They are building something for themselves, not for you.
Never compare the bottom line alone. Put the quotes side by side and first check they price the same scope - often they do not, and then cheapest just means smallest. Then compare assumptions: who understood the data, the integrations, and the risk best? Compare test plans, ownership, and operations: what do you take with you, who operates, and what happens on change? Ask for clarification where quotes stay silent - silence is information. Finally, weigh the people. You will work closely with this team for weeks or months; choose the ones you trust when things go wrong, not the ones with the prettiest slides.
This is the simplest way to get comparable quotes, and the most skipped. Tailor the request per vendor - more detail to the favourite, less to the rest - and you get quotes on different projects. Send the same text to all, answer follow-up questions identically, and share clarifications with every party. Then vendors compete on understanding and quality, not on who guessed your scope right. The template below exists for exactly this: fill it in once, and send it to everyone you want a quote from.
Feel free to use the template below when writing to us at post@aprex.no. The more you fill in, the more concrete our answer: does the case fit a pilot, what would we test first, and what is the next step? Where something is unknown, write unknown - that is information too. We give no fixed price before seeing the case, but we always give an honest answer on whether we think we can help, and what we need to find out.
Seven points that make a request quotable. Illustration - adapt to your case.
What takes time, creates errors, or stalls cases today, and for whom?
What should the solution do, and what should differ afterwards?
Which systems are involved, and how is data quality?
Who uses the solution daily, and who approves?
Sensitive data, privacy, security, or timeframe to respect?
What is the one measurable criterion for success?
Ask for scope, assumptions, testing, ownership, operations - and exclusions.
Read what drives AI pilot and website cost.
Read more →We would rather decline a request than price something we do not understand. A good quote - from us or others - answers scope, assumptions, testing, ownership, and operations. Demand that of everyone, and you get quotes you can actually choose between.
Use the template above, fill in what you know, and write unknown where you do not. We answer honestly.
Contact Aprex →