Coaching: the key to lasting performance improvement

Coaching offers a greater return on a lower outlay than almost any other performance initiative an organisation can take – and it is the one they are least likely to do properly.

There is no equipment to buy and no platform to license. In sales, where performance change is comparatively easy to measure, we have found that a well-executed coaching programme can lift revenue by 25% or more while significantly improving sales profitability. It is harder to measure elsewhere in a business, but there is good reason to believe the impact can be just as real, and can arrive in a relatively short time frame.

The catch is that the one input coaching does require – a manager’s attention, spent deliberately, week after week – is the scarcest resource in most organisations. Which is why coaching remains a route seldom travelled: most initiatives are launched, publicised, and quietly abandoned, for reasons that are entirely predictable and almost entirely avoidable. This page sets out what the research now actually says about coaching – including where the evidence is thinner than the enthusiasm – and turns that into ground rules for anyone who would rather their coaching project worked.

What the evidence actually says

Coaching used to be sold on conviction. It no longer has to be – provided we are honest about what the research does and does not establish.

It works, reliably, and by a respectable margin

A 2023 meta-analysis in Frontiers in Psychology pooled eleven studies of workplace coaching and found a moderate, statistically robust positive effect (Cannon-Bowers et al., 2023). It is the third such review in a decade to reach that conclusion, following comparable analyses by Theeboom and colleagues in 2014 and Jones and colleagues in 2016. Whether coaching works is, for practical purposes, settled.

Every way you cut the data, it still works

This is the most useful finding in the analysis, and the most widely ignored. The researchers tested the obvious candidates for “the right way to coach”: non-directive facilitation against goal-setting and accountability; face-to-face against virtual; skill outcomes against attitude and wellbeing outcomes; improvement judged by the manager against improvement claimed by the individual. Every comparison pointed the same way, and none of the differences between them was statistically reliable. The largest single estimate was for skill-based outcomes, such as those which follow a learning input designed to effectively give people new skills which are required immediately.

Be clear about what that does and does not mean. The current research does not favour one methodology over another. What it does tell you, unambiguously, is that all of these approaches work. So stop waiting for the methodology argument to be settled before you start. The returns come from undertaking any coaching at all, competently and consistently.

More is not automatically better

Across all three meta-analyses, neither the number of coaching sessions nor the total hours predicted the size of the result. What we infer from that is that volume is not a strategy. A few well-aimed conversations about one specific behaviour will do more than a long series of unfocused ones.

What kills coaching is usually not the coaching

Rummler and Brache put it best back in 1990: “pit a good performer against a bad system and the system wins almost every time” (Rummler & Brache, 3rd edition, 2012). Recent UK workforce research bears them out. Across 5,000 workers, one in three said their organisation actively makes it harder, not easier, to do a good job – citing slow processes, unrealistic workloads, outdated systems and constant interruption in roughly equal measure (Ipsos Karian and Box, 2026). In those environments, motivation, pride and advocacy fall by between 37 and 54 percentage points. You cannot coach someone out of a broken system, and attempting it is how coaching acquires a reputation for being a waste of time. Hold those findings in mind. Everything that follows is an attempt to act on them.

Plan before you coach

Most coaching initiatives fail before the first conversation, because too little groundwork was done. A few organisations have a genuine coaching culture, where managers and their people both expect it to happen. Unresolved obstacles do not stay quiet: they harden into active opposition, and a project that begins mired in it rarely recovers.

Two questions are worth forcing at this stage. What are the three or four capabilities that genuinely differentiate performance in this business, given its strategy? And what in the working environment is currently stopping people from using the capability they already have? The second is the one organisations skip.

Get backing from the top

Senior support sets the expectation that coaching is a priority rather than an optional extra. Two words matter: visible and sustained.

Too many initiatives open with a blaze of publicity and senior enthusiasm that evaporates the moment attention moves elsewhere. Nothing gives coaching more impetus than senior managers who personally follow up on both its frequency and its impact. It is also worth agreeing in advance what sanction applies to a manager who simply doesn’t do it – an uncomfortable conversation to have at the outset, and a much more uncomfortable one to have later. Coaching slides down a priority list with remarkable ease, and regular senior attention is the most reliable brake on that slide.

Enable your coaches

Not every manager has the skills to coach, and some will not admit it. Fear of being exposed is one of the quieter engines of manager resistance. It also runs the other way, people resist being coached when the coaching they receive is unhelpful.

Two things are needed. Coaches need training in how to coach. They also need real credibility in the skill they are coaching, because few people take instruction seriously from someone they believe knows less than they do. If coaching matters, make the space and time for managers to make it happen.

The power of structured coaching

Learn to overcome the three biggest coaching barriers and explore how AI coaching tools can support coaching initiatives.

Watch now

Coach the middle, not the tail

The instinct is to coach the poor performers, because they are the most visible problem. Resist it.

Performance in most populations follows a familiar curve: a small group well above expectation, a small group below, and the large majority in the middle. Working with the below-expectation group is hard and slow, and at best converts poor performers into average ones – a low payoff for high effort. Your best performers are the opposite trap: coaching them is easy and pleasant, but they are already delivering and there are very few of them, so the payoff is just as low.

The return is in the middle. Moving the large average group towards the standard set by your best people is where the arithmetic works: this group improves fastest for the effort invested, and there are far more of them. Shifting the bulk of a population from average to good transforms aggregate results in a way that rescuing the tail never will.

Match the style to the person

Coaches have to adapt. A new starter needs something very different from an experienced hand, and the most common coaching error after picking the wrong people is using the same style on everyone.

Three styles cover most situations, and what selects between them is how far along the person is. Directing suits someone who does not yet know what good looks like: tell them, show them, correct them. Guiding suits someone competent and still developing – a nudge and a regular check rather than instruction. Enabling suits someone who knows their job: give them the opportunity, support planned action then stay out of the way and ask how it went.

Note that the research gives you licence to be pragmatic here. Since no coaching school has been shown to outperform another in aggregate, the question is not which approach is theoretically correct, but which one will benefit the person in front of you. Getting that judgement wrong wastes effort at best, and damages confidence at worst.

coaching style

Protect the rhythm

Weekly is ideal; fortnightly should be the floor. Not because volume buys results – the evidence is clear that on its own it doesn’t – but because rhythm is what turns coaching from an event into a habit, for both parties. A conversation that happens only when time allows is a conversation that stops happening.

How long to keep going depends on the individual, the number and complexity of the skills involved, and how often real opportunities to observe arise. Be realistic about capacity. Three or four people is a genuine coaching load for a working manager. Six-plus is a burden which leads to rushing, poor preparation and coaching quickly descends to the level of tick-box exercise.

Don’t start what you won’t finish

Experience and the available evidence point the same way: raising expectations and then abandoning coaching is worse than never starting.

It damages morale, and it damages the relationship between the person and their manager. There is also a real risk that performance falls, as someone applies half-learned skills without support, or reverts to previous practice. Coaching is an intervention, and interventions have side effects when they are done badly or stopped abruptly.

If you cannot sustain it, don’t announce it.

Work on a few things at a time

Trying to change several things at once rarely works. Pick one and stay with it until the change holds.

There is a mechanism behind this. Behaviour under pressure depends on having mental capacity available in the moment, and that capacity is easily exhausted by complexity and competing demands. Give someone six things to think about differently and they will revert to the habits they already have. Give them one, and they have a chance.

Choose the area with the largest return for the effort required, or the one that matters most to the business. Then be patient – some changes take time to embed, particularly complex ones. A single genuine improvement can move performance materially.

Lead with the positive, and never use the sandwich

Recent research has changed the way we should approach coaching

The sobering research

Start with the most sobering number in the literature. Kluger and DeNisi’s meta-analysis – 607 effect sizes drawn from more than 23,000 observations – found that feedback interventions improved performance on average, but that over a third of them made performance worse (Kluger & DeNisi, 1996). Not neutral: worse. Their explanation is the part worth holding on to. Feedback works when it directs attention to the task, and it fails as attention drifts upwards towards the self. Thirty years on, that finding has not been overturned, and it is why feedback deserves more care than almost anything else a manager does.

However, research on performance feedback is not easy to make sense of. In fact, a recent, critical systematic review of thirty years of performance feedback research found the field in some disarray: 75 distinct labels for feedback with contradictory definitions, and more than half of published studies failing even to specify whether they were studying positive or negative feedback (Heine, Stouten & Liden, 2025). Out of that mess, three findings are solid enough to act on.

Positive and negative feedback are not two ends of one scale

They run on different tracks. Positive feedback consistently and predictably improves performance. Negative feedback is volatile: sometimes it improves performance, sometimes it degrades it, and it reliably raises cognitive load, defensiveness and shame. People receive far less positive feedback than they need, which is reason enough to open every coaching session by exploring what is going well.

The feedback sandwich is dead

The positive-negative-positive sequence is one of the most widely taught techniques in management, and the research has comprehensively debunked it. The reason is simple: the sandwich overloads attention, and what survives the conversation is not the useful positive content but the negative comment in the middle, with all of its unpredictable effects.

Where time is tight, a negative-positive-positive sequence performs demonstrably better, because it gives the recipient room to absorb and stabilise around the corrective point before reinforcement is added. When you provide negative or corrective feedback, leave a space for it to land before building the coachee’s belief through acknowledging why you are devoting your time to helping them succeed. It also does precisely what Kluger and DeNisi warn against: the lurch from praise to criticism and back tells the recipient that the conversation is about them, not about the work.

The relationship is the buffer

This is the one that matters most. The single strongest protection against feedback going wrong is the quality of the relationship between coach and coached. Where trust and psychological safety exist, corrective feedback is heard as development. Where they don’t, it is heard as threat. If you have not built that relationship yet, compensate deliberately: keep the coaching rhythm, but make each piece of corrective feedback highly specific, lower its frequency, and set the resulting goals collaboratively rather than issuing them.

Whatever the sequence, remain future-focused. What could you do differently? Where and when will you try it? What do you expect to get out of it? That orientation is what makes a coaching conversation feel useful rather than judicial.

The bottom line

Coaching works. Three independent reviews over a decade agree on that much. What no one has yet shown is that any particular school, format or fashion works better than the rest – so if you are waiting for that argument to resolve before you begin, you will be waiting a long time, for no good purpose. The returns are not hiding in a methodology. They are in whether the conversations happen, whether they stay focused, and whether the working environment lets people use what they have learned.

Do the groundwork. Clear the friction you can clear. Equip the coaches, start with the willing, work on the middle of the curve, take one behaviour at a time, lead with what is going well – and don’t stop. Do that, and coaching will deliver real and sustained improvement at a return far greater than the time invested. Very few approaches to developing people can match it.

Ready to improve your sales outcomes? Book a consultation with a member of our team today

Book a consultation

Ask questions and listen to the answers

Good coaches resist the urge to tell people what they did wrong and how it should have been done.

The better route is to ask people what they thought was good or poor about their own performance, and then to listen. That gives the coach something real to work with, and it lets the coach use questions to help the person think the problem through. People who arrive at their own solutions are markedly more likely to implement them.

A simple three-question structure borrowed from reflective practice is hard to beat (Rolfe, 2002). What? – what actually happened, who did what, in what order, with what result. So what? – was that effective, what could have been better, what have we learned. What next? – what will we do differently, when, and how will we make it stick. Those three questions, in that order, will carry most coaching conversations.

None of which means that you should never share your experience. As the section on style made clear, there is a proper place for a directive approach, particularly early in someone’s development. The error is making it the default or the only approach available to a diverse team with different levels of experience and capability.

Treat every contact as a coaching opportunity

Formal sessions matter, because scheduling something is how you stop it being forgotten. But the unplanned moments in a working week are often the more valuable ones.

A conversation after a difficult customer call, a debrief in the car park, a quick look at an email before it goes out – these give you live access to behaviour that formal sessions can only discuss in retrospect. Coaches who watch for these moments get far more from the same investment of time.

Measure behaviour instead of activity

Keep a record for each person of what has been coached and when, and have senior managers check those records regularly, because reviewing them is the clearest available signal that coaching matters. Measure the change in performance wherever you can and feed the data back to the people involved. Visible improvement is highly motivating, and the knowledge that the initiative is being tracked keeps everyone focused.

One warning: measure the right things. The failure mode here is the vanity metric – attendance figures, completion rates, satisfaction scores – which tells you the activity happened and nothing whatever about whether behaviour changed. Coaching sessions delivered is an input. Behaviour observed at the point of work is the outcome. Only one of them is worth reporting.

Communicate success

Build momentum by telling people what is working.

The people being coached are motivated by hearing their progress recognised more widely. The sceptics reconsider when they see that coaching produces results rather than paperwork. And the coaches themselves are motivated both by the improvement they are generating and by the knowledge that senior management has noticed.