SetpointHQ · Marketing engineering
100 job adverts a week, rewritten by hand by 2 people. That’s what I found when I joined a national healthcare provider as head of marketing. By the time I left, an automated system was writing and publishing more than 300 a week.
I’ll come back to how. First, the bit that matters if you’re about to commission some software.
Building tech is the easy part now
Software has never been cheaper or quicker to build. AI coding assistants, no-code platforms and off-the-shelf APIs mean a small team can have a working tool in weeks. Sometimes days. Recruitment agencies are doing it, and so is pretty much every business that markets itself or hires people.
So speed is sorted. What hasn’t moved on much is whether the thing you build does the job you actually needed doing.
Where the brief goes missing
You’ve probably been here. You write the brief. Someone turns it into a spec. A developer reads the spec. By the time anyone writes a line of code, the reason you asked for it in the first place has worn a bit thin.
I don’t blame developers for this. They’re judged on shipping, stability and closing the ticket, so they take the cleanest route to something that works. Fair enough. The cleanest route just isn’t always the right one.
Same healthcare provider, different project. During a CRM migration, the front office wrote the spec and a dev team built it. To the letter. Trouble is, the front office weren’t thinking about what a new system could do. They wanted one that worked like the one they knew, which is a fair thing to want when you use it all day. And nobody between them and the code asked what else was possible. What came out was a replica of the old system sitting inside a new tool. Less flexible, harder to use, slower. Nobody enjoyed using it.
And because rework costs time and money, most businesses take what they’re given and move on.
Nearly right, 5 times over
One tool that’s nearly right is irritating. 5 of them and the whole business starts to drag.
People build workarounds. A spreadsheet turns up to bridge 2 systems that were supposed to talk to each other. Someone loses Friday afternoon copying data from one platform into another. Adoption drops, because the tool never quite fits how the work gets done. The bottleneck you paid to remove is still there. It’s just moved somewhere you can’t see it.
The bit nobody put in the spec
A recruitment business I worked with wanted contractors to upload their own timesheets into a new system, instead of their consultants doing it for them as before. Sensible on paper.
Hardly anyone used it. So the business made it mandatory.
What nobody had clocked was that the weekly timesheet conversation was never really about the timesheet. It was a check-in. Consultant and contractor caught up, the relationship got a bit stronger each week, and the consultant picked up intel on new jobs, on what was going on in the team, on where vacancies were about to pop up.
A good chunk of a desk’s pipeline, hiding inside an admin task.
The mandate took that conversation away. And some consultants ended up keying in timesheets on the quiet for their contractors anyway, doing admin when they should have been growing their desk.
Madness, frankly.
Back to the adverts
Briefed as “automate the adverts”, a lot of developers would have built a template generator. Fields in, advert out. Done, technically.
But I’d spent years learning what a job advert has to do for the person reading it and for the team hiring. So the system (AI, automations and job data agents) was built around that, not around what was quickest to code. 100-odd adverts a week became more than 300.
The volume isn’t the bit I care most about, though. That team was maxed out. Taking the repetitive work off them gave them room to think, and what they did with that room was down to them, not the software.
What changes when a marketer builds it
I’ve been in marketing since 1998, 20 of those years in recruitment marketing. At SetpointHQ we build the systems ourselves, with AI doing a lot of the heavy lifting. Marketers first, builders second. The order matters.
In practice we start from the number the build has to move (applications, cost per hire, conversion, hours saved) and choose features to serve it. There’s no handover, because the people who understood the brief are the people building. And when there’s an easier way to build something, we ask whether the easier version still moves the number. Sometimes it does, and we take the shortcut. Often it doesn’t.
Small and sharp
What we build: recruitment software, process tools that take manual work off busy teams, automations that join up systems never designed to talk to each other, and marketing tech that ties spend to results. Studio, our self-serve recruitment marketing toolkit, is one of them. Most of it is bespoke, shaped around one business’s workflow.
A small tool that does one job brilliantly will beat a platform that does 10 jobs adequately. It’s cheaper to run and quicker to change. And people actually use it.
Before you brief your next build
Who on the project owns the outcome? Are they the one building it?
If those are 2 different people, expect some of your brief to go missing on the way. Recruitment is where we go deepest, but the same thing happens anywhere a process or a marketing channel underperforms because the tech behind it was built to the wrong brief.
Tell us what it has to move.
Tell us what you need built. A senior member of the team replies within 1 working day. If it’s recruitment marketing tools you’re after right now, Studio is the self-serve route.
Talk to usExplore StudioRelated reading
Written by the SetpointHQ team, from our founder’s time in-house. We are recruitment marketers and marketing engineers: we build the tech behind the strategy.