Contents

Choose a software development vendor by how they work, not by portfolio and price: how they estimate, show progress, who owns the code and accounts, and what happens when plans slip. All of this can be checked before signing — with questions, a paid pilot and calls with past clients. Below are the questions to ask and the signs of a weak vendor.

What matters most when choosing a vendor?

What matters most is process transparency and how much control you keep over your product. A strong portfolio shows the agency can deliver good projects, but it says nothing about how yours will go — with its deadlines, changing requirements and surprises.

What to look at first:

  • how they estimate — by scenario with a breakdown, or one turnkey number;
  • how they show progress — regular demos of working software, or monthly reports;
  • who owns the code and infrastructure — from day one, or "we'll hand it over at the end";
  • who exactly will work on it — named people and roles, or abstract "specialists";
  • how changes are handled — whether there is a clear procedure to estimate and approve them.

If you have not yet decided whether you need a vendor at all, start with our comparison of outsourcing and an in-house team.

What should you ask a vendor before signing?

Before signing, ask questions a weak vendor has no concrete answers to. A good vendor answers them easily because this is their everyday work, and often covers them before you ask.

About process and team

  1. Who will be on the project team and in what roles? Can we talk to the tech lead before we start?
  2. How often will we see working results rather than reports?
  3. Which task tracker do you use, and will we have access?
  4. What happens if a key developer leaves the project?

About code, quality and handover

  1. Whose accounts will hold the repository, hosting, domains and third-party services?
  2. How do testing and code review work? What counts as a finished task?
  3. What documentation will we receive at the end?
  4. If we decide to continue with another team, what will the handover look like?

Also ask for a sample estimate or report from another project — it shows how the vendor really works.

How do you compare proposals from different vendors?

Compare proposals by scope, not by the total. Two quotes that differ by a factor of two often describe different products: one includes testing, launch and a change buffer, the other only coding against a list of screens.

Put the proposals into one table and check that each covers the essentials:

What to check Good sign Warning sign
Discovery and design A separate stage with deliverables: scenarios, prototypes "We'll clarify as we go"
Development estimate Broken down by scenario or module One number, no breakdown
Testing Explicitly included with an approach Not mentioned
Launch and infrastructure Server setup, monitoring, backups Just "deploy to hosting"
Buffer for unknowns Named and explained Missing; "fixed" price on vague requirements
Post-launch support Terms and cost described "We'll discuss later"

If a proposal is missing a part, ask why. Sometimes it is a deliberate decision, but more often it is work you will need anyway — just billed later as "extra".

Four people in a meeting in a modern office Photo: Unsplash

What are the signs of an unreliable vendor?

An unreliable vendor usually shows itself during negotiations: it promises everything fast and cheap, avoids specifics and asks nothing about your business. A good team does the opposite — asks plenty of awkward questions and says honestly what can wait.

Signs that should make you cautious:

  • they agree to any deadline and scope without a single clarifying question;
  • they quote an exact price for a complex product after a five-minute call;
  • they will not set up the repository and infrastructure on your accounts;
  • they will not introduce the team until the contract is signed;
  • their portfolio is screenshots only, with no live projects or client contacts;
  • they propose starting development without scenarios "so as not to waste time".

How do you reduce risk before a large project?

Reduce risk with a small paid stage before the main work — discovery, a prototype or the first user scenario. Within a few weeks you see how the vendor communicates, keeps deadlines and writes code, and the cost of a mistake is capped at that stage.

What else helps:

  1. Talk to one or two of the vendor's past clients, ideally ones whose projects are now in support.
  2. Ask an independent developer to review the pilot's code.
  3. Put code ownership, regular demos and the handover procedure in the contract.
  4. Pay in stages for accepted results rather than upfront for the whole project.

The pilot's output — for example, scenarios written following the structure of a good specification — stays with you even if you continue with another team.

In short

  • Choose by process and transparency, not just portfolio and price.
  • Ask about the team, demos, account ownership, testing and handover before signing.
  • Compare quotes by scope: testing, launch and a buffer for unknowns belong in every one.
  • Warning signs: promises without questions, prices without breakdowns, refusing to hand over access.
  • Start with a small paid stage — the cheapest way to test a vendor.

FAQ

  • Where do you find a development vendor?

    Referrals from companies with a similar product already running are the most reliable source. Agency rankings and directories help build a shortlist, but make the final choice after talking to the team and running a pilot.

  • How many vendors should we consider?

    Three to five is usually enough. More are hard to compare meaningfully, and in-depth conversations take time better spent on a pilot with one or two finalists.

  • Should we sign an NDA before discussing the project?

    If you are sharing trade secrets or customer data, yes. A first conversation about a product idea usually does not need one: the value is in execution, and vendors rarely copy client ideas.

  • What should we do if the vendor keeps missing deadlines?

    First find the cause: waiting on client decisions, underestimation or team problems. If the cause is on the vendor's side and does not go away, start preparing a handover while the code and access are in your hands.