ARTICLE

How to Write a Project Brief That Gets Accurate Dev Quotes

How to Write a Project Brief That Gets Accurate Dev Quotes

Most software quotes are wrong because most project briefs are vague. An agency receives three paragraphs describing a “customer portal,” and produces a number that is really a guess wrapped in a PDF. Then the project starts, the real requirements surface, and the number moves.

A good software project brief fixes this. It is two to four pages, takes a few hours to write, and is the single highest-leverage document in the procurement process — because it is what turns three wildly different quotes into three comparable ones.

Why vague briefs produce bad quotes

When an agency cannot see the scope clearly, they do one of two things. Honest ones pad the estimate to cover the uncertainty, so you pay for risk that may never materialise. Less honest ones quote low to win the work, knowing the gap will be recovered through change requests later.

Either way you lose. And you cannot compare vendors, because each one is quoting for a different imagined project.

What a good brief contains

1. The business problem, not the solution

Start with what is wrong today and what it costs you. “Our operations team manually re-keys 200 orders a day from email into the ERP, taking roughly 15 hours a week and producing around 30 errors a month.”

That is useful. “We need an order management system” is not, because it has already decided the solution — and a good partner might tell you an integration costs a tenth of a new system.

2. Who uses it, and what they do

List each user type, roughly how many of them there are, and the two or three things each needs to accomplish. Keep it to one line each:

  • Operations staff (6 people) — review incoming orders, correct errors, push to ERP.
  • Account managers (20) — check order status for a customer without calling ops.
  • Customers (≈1,200 active) — submit an order and see its status.

This single section eliminates more estimate variance than any other, because user types drive permissions, interfaces and testing effort.

3. Must-have vs nice-to-have

Split your feature list explicitly. The must-have list is what launch means. Everything else goes in a second list marked for later.

Be ruthless here. If you cannot cut anything, you have not prioritised — you have just written a wish list, and the quote will reflect that. Our guide to MVP scoping covers how to make these cuts.

4. The systems it must talk to

Name every system, its version, and whether it has an API. This is where estimates swing by six figures.

“Integrates with our ERP” could mean a documented REST API (two weeks) or a 1990s system with a nightly CSV drop and no support contract (four months). If you do not know, say so — “SAP, version unknown, we will need to check what integration options exist” is far more useful than silence.

5. Constraints

Everything that limits the solution space, even when it feels obvious to you:

  • Regulatory requirements — GDPR, HIPAA, PCI DSS, data residency
  • Technology mandates from your IT department
  • An immovable date and what it is tied to
  • Existing vendor contracts that restrict options
  • Internal capacity — who on your side is available, and how much

6. Your budget range

This is the section clients most often omit, usually because they believe withholding it prevents vendors from quoting to the number.

In practice it wastes everyone’s time. Without a range, agencies guess at scale and half the responses are irrelevant. Give a range — “$60,000–$100,000 for phase one” — and good agencies will tell you honestly whether your scope fits it, and what they would cut if it does not. That conversation is worth more than the quote.

7. How you will decide

State your criteria and their weighting, your timeline for deciding, and who makes the call. Vendors invest proportionally to how real a process looks.

What to leave out

Technology choices you have not reasoned about. Specifying React because you heard it is modern narrows the field for no benefit. State the constraint behind it if there is one.

Detailed screen designs. At brief stage these over-specify the solution and reduce the chance a vendor proposes something better.

Boilerplate. Three pages of company history buries the information that actually drives the estimate.

A structure you can copy

SectionLengthWhat it answers
Context & problem½ pageWhy this, why now, what it costs today
Users & jobs½ pageWho uses it and what they need to do
Scope: must / later1 pageWhat launch means
Systems & data½ pageWhat it connects to
Constraints¼ pageWhat limits the solution
Budget & timeline¼ pageRange and any fixed date
Process¼ pageHow and when you decide

Three pages. If yours runs past five, you are writing a specification rather than a brief — which is a later-stage document, produced during discovery, usually with the partner you chose.

Frequently asked questions

How long should a software project brief be?

Two to four pages for most projects. Long enough to remove ambiguity about scope, users and integrations; short enough that every vendor reads all of it.

Should I really share my budget?

Yes, as a range. It filters out mismatched vendors immediately and turns the conversation into what is achievable at that level, which is the useful discussion.

How many agencies should I send it to?

Three to five. Fewer gives you no comparison; more means none of them invest serious effort, because the odds look poor.

What if I do not know the technical details?

Write what you know and flag what you do not. Explicit unknowns are far more useful than confident guesses, and they let vendors price the investigation properly.

Want a second opinion on your brief before it goes out? Send it over — we will tell you what a vendor will still have to guess at.

SHARE THIS ARTICLE
Facebook
Twitter
LinkedIn
WhatsApp

Related Articles

Ready to build yours?

Tell us what you’re building. We’ll come back with scope, timeline and a realistic budget.

Scroll to Top