Most RFPs get worse responses than they deserve, and the cause is almost always the same: ambiguity. A supplier who cannot tell what you mean does not ring you up to ask — they price the uncertainty and move on. That risk buffer is real money you pay for nothing. This is written from the side that reads these documents, including the parts suppliers do not usually say out loud.
We receive these, we do not issue them. That makes this a biased source, and also a more useful one than another template — because the thing that most improves an RFP is knowing what happens to it after you press send.
What actually happens when your RFP lands
Someone senior enough to decide reads the first two pages. Within about ten minutes they are answering one question: do we understand what they want, and can we win this?
If the answer to either is unclear, one of three things follows. The strongest suppliers decline quietly — you never learn they considered it. Others respond with something largely generic, because effort is a bet and the odds look poor. And some respond properly but price defensively, because they have to cover the interpretation that turns out to be true.
All three outcomes look the same from your side: a thin shortlist and prices that vary more than they should.
Ambiguity is priced, not queried
This is the part worth internalising. When a requirement can be read two ways, a supplier does not usually ask — the Q&A window is short, questions are often published to all bidders, and asking reveals what you have noticed. So they estimate the interpretation they think is likely, then add a buffer for the one they think is possible.
This also explains something buyers often misread. Wide variance between quotes is usually diagnostic, not competitive. If three suppliers come back at very different numbers for the same document, the most likely explanation is that your requirements permitted three different readings. Picking the lowest means picking whoever interpreted the scope most narrowly — which you discover during delivery.
The sections that change response quality
Most RFP length is boilerplate. Four sections do nearly all the work:
- The current situation. What exists now, what is wrong with it, what you have already tried. This is the most skipped section and the most valuable one, because it tells a supplier which solutions are already ruled out.
- Who uses it, and how many. Five internal users and fifty thousand customers are different projects wearing the same feature list. Include volumes, peaks and the environment people work in.
- Constraints you cannot move. Systems that must be integrated, a regulatory deadline, a platform decision already made, data that cannot leave a jurisdiction. Suppliers can design around a stated constraint; they cannot design around one they meet in week three.
- What success looks like. Not a feature list — the outcome. "Cut call-centre volume" or "pass the audit in March" tells a supplier what to optimise when a trade-off appears, and trade-offs always appear.
Include a budget range
The usual objection is that suppliers will simply quote up to the number. Some will. The cost of withholding it is higher.
Without a range you receive proposals spread across an order of magnitude, built to different shapes, that cannot be compared on any axis. With one, a supplier proposes the best version of the thing that fits it — and if the number is unrealistic, a good supplier tells you so before either side spends weeks. That is genuinely useful information, and you only get it by asking.
If you truly cannot name a figure, say what the problem is worth: the cost of the current process, the revenue at stake, the fine you are avoiding. That is enough to anchor a sensible response.
Questions that produce useless answers
Some standard RFP questions are answered identically by every supplier, which makes them worthless for discriminating between them.
- "Describe your quality assurance process." Everyone describes a reasonable process. Ask instead: tell us about a project that went wrong and what you changed afterwards. The answer, and the willingness to give one, is far more revealing.
- "List your relevant experience." Produces a curated list. Ask instead: who specifically would work on this, and what have they personally shipped? It is remarkable how often the team that wins the work is not the team that does it.
- "Confirm you can deliver by [date]." Everyone confirms. Ask instead: what would have to be true for that date to hold, and what would you cut first if it slipped?
- "Are you GDPR compliant?" Everyone says yes. Ask what data they would hold, where, and under what agreement — we went into why that distinction matters in GDPR when you outsource development.
The pattern: replace questions that invite a statement of intent with questions that require a specific past fact or a concrete trade-off.
What makes strong suppliers walk away
Worth knowing, because the suppliers you most want are the ones most able to be selective:
- A fixed price against requirements you have said are still evolving. That does not transfer risk, it prices it — and the suppliers willing to absorb it are often the ones planning to recover it through change requests.
- Unlimited liability or uncapped indemnities. Many firms simply cannot sign these, so the clause silently filters your shortlist before anyone reads the requirements.
- Ten or more invited bidders. At those odds the expected value of a serious response is low, so you get volume responses from firms that mass-produce them.
- A one-week window over a holiday period. You find out who was not busy.
- Price-only scoring. It tells a supplier exactly what kind of engagement this will be, and the ones who optimise for quality withdraw.
Scoring, and the shortlist
Agree the weighting before responses arrive, not after — retrofitting criteria to a preferred answer is common and corrosive, and experienced suppliers can usually tell.
Invite three to five. Fewer gives no comparison; more degrades the quality of what comes back for everyone, including you.
And consider replacing some of the document with a small paid pilot for your top two. A week of real work against a real problem tells you more about how a supplier thinks, communicates and handles an obstacle than fifty pages of prose ever will. It costs a fraction of the mistake it prevents.
Questions we get asked
Should an RFP include a budget range?
Yes — withholding it is the most common self-inflicted wound in software procurement. A range lets a supplier propose a solution shaped to fit it. Without one you get proposals spread across an order of magnitude that cannot be compared. If you genuinely do not know, say what solving the problem is worth instead.
How long should a software RFP be?
Long enough to describe the problem, short enough that a good supplier reads all of it. Most length in a weak RFP is boilerplate and feature lists; most value in a strong one is three or four pages on the current situation, the users, the constraints and what success looks like.
Why do suppliers quote such different prices?
Usually because the document allows more than one reasonable reading and each priced a different one. Wide variance is diagnostic: it says the requirements are ambiguous, not that one supplier is cheap. Find the question they answered differently rather than taking the lowest number.
What makes a good supplier decline to bid?
A fixed price against admittedly evolving requirements. Uncapped liability. A response window too short for real thinking. Ten or more invited bidders. And price-only scoring, which tells a supplier what kind of engagement this will be.
Fixed price or time and materials?
It depends how well defined the work is — asking for a fixed price on undefined work prices risk rather than transferring it. A common middle path is a fixed-price discovery producing a specification, then a fixed price for the build against it.
How many suppliers should we invite?
Three to five. Fewer gives no comparison; more reduces each supplier's odds enough that the strongest put in less effort or decline. Inviting ten optimises for the appearance of diligence rather than the quality of responses.
If you are drafting one now, the most useful hour is spent writing the current-situation section properly — it is the part suppliers read most carefully and the part most often missing. We are happy to read a draft and tell you where it is ambiguous, whether or not we end up bidding on it; that is how most custom software engagements start.