Last updated · Peter Baker
This worked example is illustrative, not a client engagement. The recruitment agency is composite and the figures are indicative. A real AI Opportunity Audit measures your own workflows and produces your own numbers. What is accurate here is the format: the shape, depth and honesty of the audit document you receive.
The business we're looking at
A UK recruitment agency of eighteen people, split across two desks: permanent tech hiring and contract industrial. Four consultants bill, three resource, the rest is operations, finance and leadership. They use a mainstream ATS, a separate CRM their consultants half-update, and a great deal of email.
The agency's leadership asks the question every recruitment agency arrives with: everyone says we should be using AI, so where do we actually start?
The ranked opportunities
Every opportunity is scored the same way: hours it gives back each month, what it takes to build, and what could go wrong. Ranking is by return against risk, not by how impressive it sounds.
| # | Opportunity | Hours saved / month | Build effort | Risk |
|---|---|---|---|---|
| 1 | CRM records updated from call notes Consultants dictate or paste notes; the record, next action and stage update themselves. |
~22 | Low | Low |
| 2 | Weekly client reporting packs Pipeline, submissions and placements assembled per client, ready for a consultant to check and send. |
~13 | Low | Low |
| 3 | First-pass candidate shortlisting Applications ranked against the brief with the reasoning shown, for a human to accept or override. |
~34 | Medium | Medium |
| 4 | Interview notes into structured summaries A consistent candidate summary the client can read, produced from whatever the consultant scribbled. |
~9 | Low | Low |
| 5 | Follow-up sequencing on warm candidates Nobody in the pipeline goes quiet by accident; the consultant approves each send. |
~17 | Medium | High |
Note that the biggest hours saved sits at number three, not number one. Shortlisting is the largest prize and it is not where we would start, which brings us to the more useful half of the document.
Where AI won't help you
A roadmap that recommends everything is a sales document. These are the calls we would make against spending money, and why.
Don't automate rejections
Technically trivial, commercially expensive. Rejection is the experience most candidates remember, and today's rejection is next year's placement. There is also a growing compliance burden around automated decisions affecting employment: Article 22 of the UK GDPR restricts decisions made solely by automated means that significantly affect someone, and the ICO's guidance on automated decision-making sets out what that means in practice. Keep a human on the "no".
Don't generate job adverts at volume
Every agency has this option and clients have started noticing the sameness. The advert is one of the few places your firm sounds different from the one down the road. Save the time somewhere it isn't costing you distinctiveness.
Don't put a chatbot on the careers page
At this traffic level it would answer a handful of questions a week that an email already handles. The cost is not the build, it is owning it forever. Revisit if traffic multiplies.
Don't replace the ATS
The ATS isn't the bottleneck. The manual work either side of it is. Replacing core systems is the most expensive thing on this list and it would fix the wrong problem.
The costed roadmap
Three phases, sequenced so that the cheapest and safest work builds the confidence to attempt the valuable and harder work. Build costs are indicative ranges, not quotes.
Phase one: CRM notes and client reporting
indicative £2,400–£3,800Opportunities 1 and 2. Roughly 35 hours a month returned, low risk, nothing customer-facing. Pays for itself inside a quarter on consultant time alone, and it gets the CRM into a state where anything smarter is possible. Most firms discover their data is the real blocker; this phase surfaces that cheaply.
Phase two: shortlisting and interview summaries
indicative £5,500–£9,000Opportunities 3 and 4, once phase one has proved the data is trustworthy. Shortlisting stays advisory: it ranks and explains, a consultant decides. Built with an audit trail from the start, because the day a client asks how a shortlist was produced, you need an answer.
Phase three: follow-up sequencing
indicative £3,000–£5,000Deliberately last despite good returns. It touches candidates directly, so deliverability, tone and data protection all matter, and getting it wrong is visible to the market. Worth doing. Worth doing third.
The one thing to start on Monday
CRM records updated from call notes. Not the biggest number on the page, and that is the point. It is low risk, invisible to clients and candidates, and it fixes the data quality that everything further down the roadmap depends on. Consultants feel it in week one because it removes the admin they already resent, which matters more than any business case when you need people to adopt something.
If it works, the firm has proof, cleaner data, and a team that trusts the next thing. If it doesn't, they've spent the smallest amount on the safest experiment. That's what a first project is for.
What you'd walk away with
- The ranked table above, for your workflows and your numbers
- The "don't bother" calls, argued rather than asserted
- A phased roadmap with indicative build costs and payback
- A first project you could start immediately, with us, another provider, or in-house
Fixed scope, from £750. Usually one to two weeks. No obligation to build anything afterwards, and the roadmap is yours to keep. Start a pilot within 60 days and the fee comes off it in full.
Want to see where your firm stands?
If you'd rather start with something free: the website report reads your site and tells you where it's losing enquiries, with no call needed. If you already know the problem is operational rather than on the website, a discovery call is the faster route.
Related services
- AI Opportunity Audit: a fixed-price, ranked and costed map of where AI saves your firm hours. From £750.
- Process Automation: repetitive admin turned into monitored pipelines. Fixed price.
- Roadmap & Handover: what every build ends with: documentation, running costs and a roadmap.
- AI for recruitment agencies: what to automate first, and what the ICO expects.
The weekly note. One short email a week from Akashuin, our research blog: what UK firms are actually doing with AI, the numbers behind it, and one thing you can try that week. Free, and one click to unsubscribe.