How to evaluate a software agency for Internal Tools

I run a software agency, so writing a post about how to evaluate agencies is awkward. I'm writing it anyway because the number of companies pick the wrong people to help them, and their projects don’t turn out well. It makes me sad, even when they don't hire us.

Most online advice on picking an agency doesn't help. Check their portfolio. Read reviews. Make sure they communicate well. It doesn't help you differentiate between an agency that will deliver and one that's good at sales calls.

What to actually look for

The single most important thing is whether the agency understands your business problem or just your technical requirements. A bad agency will take your feature list, estimate hours, and send a quote. A good one will push back on the feature list and ask why you need each thing.

Half the time, the features a client asks for aren't the features they need. They're the features they imagined before talking to someone who's done this before.

When Blomma came to us, they had a rough idea of what they wanted. A website for their business. If we'd just built that, they'd have gotten a nice website, but with the same broken operations.

Six disconnected platforms would've stayed disconnected. Orders would still require manual inventory checks against a spreadsheet. Drivers would still get route details over the phone.

We ran workshops instead. Mapped their entire operation. Found that the real problem wasn't the website. It was the gap between their customer-facing experience and their internal operations. The website was one piece. The operations centre that connected orders, inventory, driver routing, payroll, and delivery tracking was the piece that actually mattered. An agency concerned about their Clutch ratings would've built the website and walked away.

In the discovery call, does the agency ask about your business or just your feature list? If they go straight to how many screens you need and what integrations you want, they're building to spec, not solving a problem.

Ask them what they'd cut from your project

Any agency can say yes to everything. A good one will tell you which features aren't worth building in version one, and why. When we scoped Blomma's project, we deliberately left out several things they'd asked about because those features wouldn't have solved their core problem. An agency that agrees with everything you say is either not experienced enough to have opinions or too afraid of losing the deal to share them.

Look at how they scope

A fixed quote based on a 30-minute call is a guess. An agency that runs workshops, maps your workflows, and gives you a detailed scope with clear deliverables has done the work to understand what they're building. At Blomma, we spent real time mapping every department's workflow before writing the scope document. That front-loaded investment prevented the project from becoming a six-month moving target.

Questions that tell you something

Walk me through a project that went wrong and what you did about it

Every agency has had a project go sideways. The ones who can't name one are either bluffing or haven't done enough projects to have learned anything. Listen for whether they blame the client or take ownership. Listen to what they changed in their process afterwards.

Who specifically will work on my project?

At a lot of agencies, the person on the sales call isn't the person who builds your software. The senior person sells the work, then hands it to a junior developer you've never met. Your $40K project becomes practice work for someone building their second real application. Ask who's building it, what their experience is, and whether you'll have direct access to them.

At our agency, the person on the discovery call is the person building the tool.

What happens after launch?

Software isn't done when it launches. Your team will use it for two weeks and find things that need adjusting. Some agencies build it, hand it over, and disappear. Some offer ongoing support. The first three months after launch are when you discover what actually needs to change. Blomma's project didn't end at launch. We ran their old and new systems in parallel, rolled out department by department, and adjusted based on how the team actually used it. That post-launch period is where a project goes from technically complete to actually working.

Can I talk to a recent client?

Not a testimonial on your website. An actual conversation with someone who went through the process recently. If the agency hesitates, that tells you something. Our clients have left public reviews on G2.

What pricing tells you

If an agency quotes you without asking detailed questions about your business, the price is wrong. Too high means they're padding for uncertainty. Too low means they don't understand the scope. A realistic range for internal tools at an SMB is $8K to $50K, with a timeline of 6 to 10 weeks. We break down pricing factors here. Anything significantly outside that range in either direction is a signal.

Watch how they structure payment

An agency that wants 100% upfront before showing you anything is a risk. One that ties payments to milestones (design approval, first working build, launch) gives you checkpoints to evaluate the work before committing more money.

‍

I wrote this post partly because bad agency experiences create sceptical buyers, and sceptical buyers make our sales process longer. Every company that got burned by a yes-to-everything agency is harder to sell to, even when we're the right fit. The signals above are real, regardless of whether your decision includes us.

Your tools should work for your business, not the other way around. Book a free intro call.