SoftwareCrafting Logo

Selection Process

How to Choose a Software Development Partner

Most bad engagements were decided before the contract was signed.

The selection process is where the outcome is determined. Almost nobody runs one, and then everyone blames delivery.

Step 1
Write the scope before talking to anyone
2 weeks
A paid trial beats any number of calls
5 terms
The contract clauses that matter

Quick Answer

Updated August 22, 2026

How do I choose a software development partner?

Run five steps in order: write the scope down before you talk to anyone, decide which vendor tier fits (large firm, mid-size agency, boutique team, or freelancer), shortlist three inside that tier and issue all of them the same written brief, run a small paid trial with the top two, then check five contract terms — IP assignment, repository ownership from day one, notice period, handover scope, and who specifically is assigned. Skipping the written scope is the single most common cause of a failed engagement, because it makes every quote incomparable and every disagreement unresolvable.

Steps

5, in order

Shortlist

3 in one tier

Trial

2 weeks, paid, real work

Best for

  • First-time buyers of custom software
  • Teams whose last engagement went badly and want a real process
  • Founders who have collected quotes and cannot compare them

Not best for

  • Procurement with a mandated framework already in place
  • Very small engagements where the process exceeds the work
  • Situations where a trusted supplier already exists
Talk to a senior engineer

Side By Side

The five steps, and what each one protects against

How to Choose a Software Development Partner — side-by-side comparison
StepWhat you doWhat it protects against
1. Write the scopeA feature list, users, constraints, and what success looks likeIncomparable quotes and unresolvable scope arguments later
2. Choose the tierLarge firm, mid-size agency, boutique team, or freelancerPaying enterprise rates for startup work, or the reverse
3. Shortlist threeSame written brief to all three, inside one tierComparing a full-service quote against a partial one
4. Run a paid trialTwo weeks of real work with the top twoDiscovering communication and quality problems at month three
5. Check five termsIP, repository, notice, handover, named peopleA messy, expensive, or hostile exit

Each Option In Detail

What each one is actually good and bad at

Before you contact anyone

The preparation that determines whether the rest of the process can work.

A day of your time, and it repays itself many times

Strengths

  • A written feature list makes every quote comparable
  • Naming constraints — budget, deadline, existing systems — filters vendors early
  • Defining success prevents the "that was not what we meant" conversation
  • Deciding your tier stops you comparing a freelancer with an enterprise firm

Limits

  • It is genuinely work, and most buyers skip it
  • A scope written without technical input can miss the hard parts
  • It should be a starting point, not a specification set in stone

Best for: Every engagement above about two weeks of work

During evaluation

What to test while you still have leverage.

Two to four weeks of calendar time

Strengths

  • Speaking to the actual engineers reveals seniority immediately
  • Asking what went wrong recently separates honest vendors from polished ones
  • A paid trial surfaces communication problems while they are still cheap
  • Watching how they scope your brief tells you how they will handle ambiguity later

Limits

  • Trials cost money and slow you down by a couple of weeks
  • Good engineers can present badly on a sales call
  • Two weeks does not reveal everything, only the obvious things

Best for: Any engagement worth more than about $15,000

In the contract

The five terms that decide how much a bad outcome costs you.

A lawyer hour, or a careful read

Strengths

  • IP assignment signed before technical work starts
  • Repository under your organisation from the first commit, not at handover
  • A notice period you can live with — monthly rolling beats annual
  • Handover scope defined in writing rather than promised verbally
  • The specific people assigned, named in the contract

Limits

  • Contracts do not make bad suppliers good
  • Over-lawyering small engagements wastes everyone’s time
  • Some good small vendors will not accept heavy enterprise paper

Best for: Protecting the downside without pretending it removes it

Decision Guide

If this is true, choose this

Decision rules for choosing a software development partner
IfChooseWhy
You have not written the scope downDo that firstEvery later step depends on it, and quotes are meaningless without it.
Quotes vary by more than doubleRe-issue one brief to everyoneYou are almost certainly comparing different scopes rather than different prices.
The engagement is worth over about $15,000Run a paid trialTwo weeks of real work costs far less than three months of the wrong partner.
The vendor will not put the repository in your nameWalk awayIt is a small ask that reveals exactly how the exit will feel.
You have no technical person to evaluate anyoneBuy a short independent reviewA few hours of independent technical judgment is the cheapest insurance available.

The Honest Take

The part where we lose deals on purpose

This process is designed to make vendors uncomfortable, including us. It means you will ask us to name engineers, run a paid trial, accept a monthly rolling contract, and describe a project that went badly. We are fine with all of that, and we would rather you ran it properly and chose someone else than skipped it and chose us. The engagements that go wrong are almost always the ones where nobody asked hard questions early, and those cost both sides more than a lost deal ever does.

Do not hire us if any of these is true

  • Your scope is genuinely tiny and the process would cost more than the work
  • You already have a supplier you trust who has the capacity
  • Your organisation requires a formal procurement framework we do not fit
  • You want a partner who will not challenge your assumptions

Still Deciding?

Send the brief and we will tell you which option fits

Including when that option is not us. A senior engineer reads every brief and replies with the lightest sensible next step rather than the largest one.

Quick Brief

Start the conversation here

Tell us what you are weighing up. We will say which option fits and why, even when it is not us.

Your Name

Work Email

What do you need help with?

Get an honest recommendation

FAQ

Questions people ask about choosing a software development partner

What is the most important step in choosing a development partner?

Writing the scope down before you talk to anyone. It makes quotes comparable, filters unsuitable vendors early, and gives both sides something to point at when a disagreement arises. Skipping it is the most common root cause of failed engagements.

How many agencies should I shortlist?

Three, all in the same tier. More than that and the process stalls; fewer and you have no basis for comparison. Comparing across tiers — a freelancer against an enterprise firm — produces numbers that cannot be reconciled.

Is a paid trial worth it?

For anything above roughly $15,000, yes. Two weeks of real work on a real ticket reveals communication style, code quality, and honesty about problems — none of which shows up in a proposal.

What contract terms matter most?

IP assignment before work starts, repository ownership from the first commit, a notice period you can live with, a defined handover scope, and the specific people assigned named in the contract. Those five cover most of the realistic downside.

What if we have nobody technical to evaluate vendors?

Buy a few hours of independent technical review — someone with no stake in who wins, reading the proposals and joining one call. It is the cheapest insurance in this whole process, and it is the situation where mistakes are most expensive.

Ready when you are

Not sure
where to start?

Tell us your goal and we'll suggest the smallest, fastest way to get there. We reply in under 4 working hours.