Skip to main content

10 min readPublished November 8, 2026Talavishiraaj Groups
Implementation & Delivery

Most selection processes compare the things that are easy to compare: portfolio, day rate, team size, logos. None of those predict how the engagement goes. What predicts it is how a firm behaves when the brief is wrong, who owns the outcome when several parties are involved, and whether anyone is willing to tell you the work is not worth doing.

In short
  1. 01Day rate is the least useful comparison available. The same brief is routinely quoted ten times apart because each firm is quoting a different project.
  2. 02Ask what is excluded, in writing. Integration, testing, monitoring and post launch support are where a low number hides.
  3. 03Ask what would make them tell you not to build it. A firm with no answer has no criteria.
  4. 04Ask who maintains it in year two and whether that person has been told.
  5. 05A portfolio shows what a firm has shipped. It says nothing about what it is like when something goes wrong, which is the thing you are actually buying.

How do you choose a software development partner?

By testing the things that determine how an engagement goes, rather than the things that are easy to put side by side. Portfolio, day rate, headcount and client logos are all simple to compare and none of them predicts the outcome, which is why selection processes built on them so often end in surprise.

What does predict it: how the firm behaves when the brief turns out to be wrong, whether anyone owns the result when several parties are involved, and whether the firm is structurally able to tell you the work is not worth doing.

Those are observable during selection if you ask for them directly, and almost invisible if you do not.

Why is comparing quotes so unreliable?

Because the brief is usually underspecified, so each firm is quoting a different project. Published accounts of this market describe the same requirement being quoted at forty thousand by one firm and four hundred thousand by another, and the spread is not dishonesty. It is each party filling the gaps in the brief with different assumptions.

A low quote often assumes a narrower scope, excludes integration, or prices a demonstration rather than something your business can run on. Comparing it to a higher quote that includes those things is not comparing like with like, and the decision that follows is wrong.

The fix is not to collect more quotes. It is to define the requirement well enough that the quotes become comparable, which is work that happens before anyone is approached.

What should you ask that most buyers do not?

What is excluded, in writing. Testing, monitoring, integration and post launch support are where a low number hides its real cost, and a firm that will not list exclusions is telling you something.

What the three year total looks like, split into build, running and change. A quote that covers only the build is not a cost, it is a deposit. A firm that cannot produce the three year figure has scoped a demonstration rather than a project.

Who maintains this in year two, and has that person been told. The answer is revealing more often than it should be.

And what would make you tell me not to build this. A firm with no answer has no criteria, which means the judgement about whether your project should proceed is being made by someone who gets paid either way.

What does a portfolio actually tell you?

That the firm has shipped things, which is worth confirming and is a low bar. What it does not show is the part you are buying: what happened when a dependency changed mid build, how a scope disagreement was handled, whether the team that pitched is the team that delivered.

Case studies are selected by the firm and describe successes. That is not deceptive, it is what a case study is for, but it means the useful question is not "show me your work" but "tell me about an engagement that went badly and what you changed afterwards".

A firm that cannot answer that has either not been doing this long or is not being straight with you. Both are useful to know.

How do you judge whether they understood the problem?

By whether they changed your mind about anything. A firm that reflects your brief back to you in tidier language has told you nothing and has not done any thinking you could not have done yourself.

The signal worth watching for is a reframe: being told that the thing you described as a reporting problem is an integration problem, or that the automation you asked for would encode a process nobody has written down. You do not have to agree. You do have to notice whether anyone is capable of it.

Firms that only ever agree during selection do not become more challenging after the contract is signed.

In-house, agency, or something smaller first?

Build in house when the capability is core to your business, the requirement is well understood, and you genuinely have the people free. The cost there is hidden rather than absent: it is the roadmap those people stop working on.

Hire an agency when the requirement is settled and what you lack is capacity. The honest warning is that a build quote typically covers the build and not the integration or the years of maintenance after it.

And where the decision itself is not settled, neither of those is the right first purchase. Committing to a build in order to find out whether you should have built it is the most expensive way to answer that question.

What should the contract actually pin down?

Who owns the outcome when several vendors are involved. Where more than one party is in play, someone has to own the result across them, and if the contract does not name who, the answer in practice is nobody.

What happens when the requirement changes, because it will. What matters is whether that is treated as a costed change with an agreed rate or as an argument.

And how it ends. Documentation, runbooks and training should be scoped at the start, not offered at the end. A partner who becomes structurally necessary has failed at the job, and the measure of the work is that your team can change what was built after they leave.

Where this work sits
Implementation & Delivery →

We take responsibility for execution: running the plan, coordinating the moving parts, and staying accountable through to the result.

Weighing a decision like this one?

Describe the situation and we will tell you how we would approach it.