Your Best Lessons Are Costing You Fees. Make Them Somewhere Else.

By Will Bourne, Founder and CEO, Rekbot

In fifteen years of running UK recruitment agencies, I watched a lot of fees walk out of the door. Very few left because a consultant was lazy. Most left because a consultant was learning.

That is the uncomfortable truth about our industry. We learn by doing, and in recruitment “doing” means a live client, a live candidate and a live fee. Every lesson has a price tag, and someone always pays it.

I am not against mistakes. I am for making them in the right place. And we now have our own evidence of what happens when you do.

Mistakes are the best teacher. That is science, not a slogan.

In 2008, Nina Keith and Michael Frese pulled together 24 studies covering 2,183 people. Training that encouraged people to make and learn from errors beat training designed to prevent errors. The advantage grew when people had to apply the skill to new situations. That describes recruitment well: no two calls are the same.

Manu Kapur’s work on “productive failure” points the same way. A 2021 meta-analysis of 166 comparisons found that letting learners struggle before being taught beat instruction-first teaching. Robert Bjork’s research on “desirable difficulties” explains why: the conditions that produce the most errors while learning often produce the most lasting skill.

Anders Ericsson’s work on expertise adds the last piece. Experts don’t just rack up hours. They do deliberate practice: focused effort on a weakness, with feedback and repetition. Experience on its own is not practice.

So yes, mistakes make you better. The question is where you make them.

In recruitment, the classroom is a live client

Picture three moments every agency leader will recognise.

The fee negotiation. A hiring manager says, “Your competitor does 15%.” Your consultant folds from 20% in under a minute. On a £60,000 salary, that is £3,000 gone on one placement.

The counter-offer. Your candidate accepts, then goes quiet. Their employer has matched the offer. Your consultant never tested the real reasons for moving, so they have nothing to go back to. A month of work disappears.

The cold BD call. “We’ve got a PSL, we don’t use agencies.” The junior says “no problem” and hangs up. One wasted lead might not matter. Hundreds a year do, especially when Bullhorn’s GRID 2025 report found winning new clients was the top priority for a quarter of agencies.

None of these consultants are bad. They are rehearsing in front of a paying audience.

Ramp time makes it worse. Traditional graduate onboarding means 8 to 12 weeks before a new hire is confident on candidate calls, and 10 to 14 weeks to a first placement. Every week is paid for in salary, management time and burned leads.

Experienced billers leak money too

This isn’t only a junior problem. Senior consultants build habits that once worked and never get checked again. They discount by reflex, skip qualification because they “know the market”, and handle counter-offers the way they did in 2015. Years of doing the same thing are not years of getting better.

The coaching that would catch these leaks rarely happens. Gartner research found 58% of sales managers struggle to get through their tasks in the time they have, and only 40% of sellers said they work in a strong coaching culture. In most agencies I know, the manager is also the top biller. When the month gets tight, coaching is the first thing dropped.

Worse, coaching is expensive. In our own client data, onboarding alone was costing around 20 hours a week of manager and senior biller time. That is billing time the business never gets back.

Why the training day doesn’t stick, and what pilots learned

Most agencies answer all this with a training day. It feels productive. Then nothing changes.

The forgetting curve, first mapped by Hermann Ebbinghaus and replicated in 2015, shows how fast we lose what we hear once. What slows it down is retrieval: producing the skill, not just hearing about it. In Roediger and Karpicke’s 2006 experiment, students who practised recalling a passage remembered 61% a week later. Those who re-read it remembered 40%. Listening to a trainer explain objection handling is re-reading. Handling the objection out loud, under pressure, is retrieval.

Aviation and medicine solved this decades ago. You can’t let a trainee pilot learn engine failure on a full flight, so it happens in a simulator. A 2011 JAMA meta-analysis by David Cook and colleagues found medical simulation, compared with no training, produced large gains in knowledge and skills.

Safety matters too. Amy Edmondson’s Harvard research showed people learn more when they feel safe to take risks, ask questions and admit errors. A recruiter afraid of looking stupid won’t experiment on a live call. In a private simulation, they will.

The proof: four graduates, one agency, half the ramp time

When Avery Fairbank put four graduate trainees through Rekbot, their first placements landed in weeks five and six. Traditionally that takes 10 to 14 weeks. It is one agency and a small group, not a clinical trial. But every result lines up with the science above, and that is the point.

Within three days, the three remaining graduates were asking to make live calls. By week twelve, one had won a self-generated client worth over £200,000 in fees, with more than 50 roles confirmed. Charlie Fairbank, the agency’s CEO, said the efficiency felt like “a cheat code”.

How the closed loop produced those results

None of this is magic. It is the research above, built into a daily routine.

1. Volume of practice. Each graduate spent more than two hours a day in realistic candidate and client scenarios. A shadowing graduate might hear a handful of calls a day. These graduates completed the equivalent of weeks of call practice in days. That is deliberate practice at a pace no manager could deliver.

2. Mistakes with no price tag. Every fumbled objection happened against the AI, not a real candidate. That is error management training in practice: errors expected, examined and fixed, with nothing lost.

3. Feedback straight away. Adaptive feedback after every attempt meant mistakes were corrected while still fresh, not weeks later in a one-to-one. Each retry was retrieval, not re-reading.

4. Safety builds confidence. Practising privately removed the fear of looking stupid in front of a manager. By the time graduates spoke to a live candidate, they had already handled the hard moments many times. That is why they asked to start calling after three days.

5. Data, not gut feel. Because every attempt was scored on capability, confidence and behavioural fit, leaders could see learning speed within 48 hours. One of the four clearly lacked aptitude, which saved weeks of management time and salary.

6. The same standard for everyone. Training no longer depended on which biller happened to deliver it. Every graduate got the same scenarios and the same benchmark.

The result: the graduates made their expensive mistakes in week one, inside Rekbot. By the time they reached real clients, those mistakes were already behind them.

Where Rekbot fits

I built Rekbot because I spent fifteen years paying for these lessons in real fees. Rekbot is the human intelligence layer for recruitment businesses. Its AI role-play and coaching let consultants rehearse the calls that make or lose money: the fee negotiation, the counter-offer, the cold BD objection, the difficult client update.

For graduates, it halves the expensive learning curve. For experienced billers, it is a private place to test habits nobody has challenged in years. For managers, it gives back the hours that onboarding and coaching used to swallow, so they can get back to billing.

Every recruiter will make the mistake. Rekbot moves it to a place where it costs nothing.

Make your next mistake on Rekbot, not on your best client. Book a demo at rekbot.ai.