How to evaluate early-stage software as a small business

Short answer: when there is no track record to read, you judge three things instead — the people, the exit, and the smallest useful test you can run this month. Everything else is noise.

Why the usual research fails

With an established product you read reviews, ask around, and look at who else uses it. With something eighteen months old none of that exists. There are five reviews, all from people the founder knows, and a case study about a business that sounds suspiciously like the founder’s mate. The absence of evidence is not a red flag — it is simply the condition of being early — but it does mean you need a different method.

1. Work out who is actually behind it

Look for a real company, registered somewhere you can check, with a named human attached. Companies House takes two minutes. A trading business that also sells software is a different risk profile from a side project, and both are different from a venture-funded startup that needs to grow tenfold or die.

None of those is disqualifying. They just fail differently. A side project goes quiet. A funded startup pivots or gets acquired and sunsets the product. A trading business is usually the dullest and most durable of the three, because the software is not the only thing keeping the lights on.

2. Ask what is unfinished

A good answer is specific: this works, that is half-built, that other thing is not planned until next year. A bad answer is enthusiasm. If the maker cannot tell you crisply where the edges are, either they do not know their own product or they are managing you, and neither bodes well for the support conversation in month three.

3. Check the exit before the entrance

Before you put a year of quotes and invoices into something, find out how you get them out. Is there an export? What format? Does it include the attachments and the PDFs, or just a list of rows? Can you trigger it yourself without emailing anyone? If the answer is unclear, treat it as a no. This is the single most important question in the whole exercise and it gets its own article here.

4. Test support before you need it

Send a question in whatever channel they use. Time the reply. Ask something slightly awkward — how you would correct a figure the software got wrong, say. Early-stage support is often genuinely excellent, because you are talking to the person who wrote the code. It is also sometimes nonexistent. You will know within a day which you are dealing with.

5. Define the smallest useful test

Pick one job you do every week and run it through the tool for three or four weeks while continuing to do it the old way. Not your whole admin — one job. Mileage, or quotes, or receipts. Then compare: did it take less of your time, and do the numbers agree with your own?

Write down the before figure first. People consistently forget this and then cannot tell whether anything improved. We make the same point about measuring tool adoption in why tools without a plan do not stick.

6. Price it on your own numbers, not theirs

Take the monthly cost, multiply by twelve, and compare it to the hours the test actually saved multiplied by what an hour of your time is worth. If you have never worked that second figure out, here is how. The answer is often uncomfortable in both directions — sometimes a £20 tool is obviously worth it, sometimes the hours saved are a rounding error and you should not bother.

Check the vendor’s own current pricing page rather than anything written about it elsewhere, including by us, because early-stage pricing moves. GraftG’s is at graftg.co.uk/pricing; normal plans start from £14 a month.

7. Decide what would make you leave

Write it down now, while you are calm. “If it loses a quote, I am out.” “If support goes quiet for a week, I am out.” Having the line written in advance stops the sunk-cost reasoning that keeps people on bad tools for years.

What this looks like applied to us

We would rather you ran this checklist on GraftG than took our word for anything. Green & Home Ltd is a real UK trading business and you can check it. The product is in beta and we say so up front. Pricing is public. And if you want the deepest version of the test, there are 21 Founding Tradies places — twelve months free on Pro with no card, then half price for life — which makes the trial cost nothing but your attention.

Questions, or want a tool we don't have yet? Email hello@grafto.co.uk — a real person replies.