Why the Best Recruiting Tools Are Built for One Person

David Law

David Law

March 31, 2026

Why the Best Recruiting Tools Are Built for One Person

The most interesting recruitment tooling I have seen this year did not come out of a talent function with budget and engineering support. It came from individuals building for themselves, in the gaps between client calls and the school run.

That should bother anyone running a talent team. It also explains something useful about why internal build projects stall.

One user changes everything

When you build something for yourself, you carry none of the weight that makes internal tools slow. There is no enterprise customer and no small business customer to reconcile. There is no onboarding to write, no edge case to handle for the colleague who will use it once a quarter, and no support burden when it breaks on a Tuesday.

One builder I came across recently assembled a searchable knowledge base of every published recruitment benchmark report he could find, then wired an agent to refresh it weekly. Total build time was about two hours, while his children napped.

The same project inside a sixty person talent team would take months. Not because the technology is harder, but because the moment a tool has sixty users it needs explaining sixty times, and every rough edge becomes somebody else’s blocked afternoon.

That is the real constraint. It is organisational, not technical.

Taste is the part you cannot buy

Here is the argument that stuck with me. When you use someone else’s product, you can change whatever the interface allows you to change. What you lose is the logic underneath.

Consider how you actually assess tenure. You might accept an average of eighteen months, unless the person has one stint of three years or more, in which case fifteen is fine. You would never surface someone with under five years of experience for a search a client is paying serious money for. You do not want sourcers returned when you asked for recruiters, even though most systems treat both as recruitment.

That is not a filter. That is judgement, accumulated over years, and it is what separates a good shortlist from a long one. Software you build yourself can hold your judgement. Software you buy holds someone else’s.

The last twenty per cent has not got cheaper

The counterweight is worth stating plainly. Getting to eighty per cent of a working tool is close to trivial now. The final twenty per cent is where the actual work sits, and it has not compressed anything like as much.

The same problem is showing up in written work. AI generated documents have crossed the threshold from obviously poor to just about acceptable, which is worse. People now put their name to output they have not thought through, and everyone downstream reads it as though somebody did. The comprehensiveness is the tell. A document so complete that nobody wants to read it has failed at its job.

If you are building, the discipline is to keep going until the problem is solved, not until something ships. Those are different finish lines.

Start with the problem, not the build

The pattern worth copying from the strongest builders is unglamorous. Define the problem statement properly before writing anything. Work out what data you actually need. Design the experience end to end. Then build.

The failure mode inside talent teams is the reverse. Somebody runs a hackathon, pulls the team off the phones for a day, and produces something that demonstrates well and dies within a fortnight. The learning has value. The business case does not survive contact with a finance director asking what changed.

A useful test is whether the thing solves a problem your boss’s boss cares about. If a hiring manager, a recruiter and finance all hold different views of where a role stands because the data lives in three systems, fixing that is worth more than any candidate review app. It is also far less enjoyable to build, which is precisely why it stays broken.

Sometimes the honest answer is that the right solution is a workflow, or a diagram, or a product you should buy rather than build. Knowing which is the skill.

What this means if you sit inside a team

The uncomfortable conclusion is that individuals should build and organisations should mostly buy. If you are one person solving your own problem, build it. If you are equipping thirty people, the arithmetic on maintenance, training and support rarely favours something homemade.

That is not an argument against internal capability. It is an argument for being clear about which problem you are solving, and for whom.

The asset almost every team is ignoring

One more thread worth pulling. Outbound response rates are falling, and they will keep falling as generated messages get harder to distinguish from written ones. Every candidate receives more and cares less.

Which makes your company’s collective network the most valuable recruitment asset you own, and almost nobody has operationalised it. Every employee can export their own connections. Enrich that data and you can see which people are connected to ten of your staff, which of them engage with your content, and which are worth an approach from a colleague rather than a stranger.

The hard part is not the technology. It is getting everyone to spend ten minutes on the export. Set against what most companies spend on agency fees, that is not a difficult case to make.

The takeaway

Build for yourself and you move quickly because you owe nobody an explanation. Build for a team and the explanation is the job. Most talent functions get this backwards. They attempt organisational tools with individual discipline, then wonder why nothing sticks.

Start with the problem your leadership actually loses sleep over. Solve that one properly, all the way through the last twenty per cent. Everything else is a demo.

If you are working out where your talent function should build, buy, or simply stop, we should talk. Get in touch with Fulcrum.

Similar Posts