Coaching vs Training vs Consulting: Which Do You Need?

You have a budget line, a problem you can state in one sentence, and three proposals that all claim to solve it. The training provider will run a two-day programme for twenty people. The consultancy will run a diagnostic and hand you a report with recommendations. The coach will meet four of your leaders one to one, across six months.

All three proposals are competent. Only one of them is aimed at what is actually wrong.

Describe the problem without naming a solution. What would you say?

That is the question that sorts this, and it is harder than it sounds. Most of the briefs I see arrive with the answer already in them. “We need to build influencing skills” is a training brief that has skipped the problem. “We need to fix the operating model” is a consulting brief.

Strip the solution out and what is left usually sounds more like this: our regional directors escalate everything to the MD, and it is slowing us down.

That version is diagnosable. The first two are not.

Start with the kind of gap, not the kind of vendor

Three questions separate these interventions cleanly.

Does your team lack knowledge, or a repeatable skill nobody has taught them? That is training. Someone cannot run a performance conversation, or read a P&L. There is a body of content, it can be transmitted, and afterwards they will know it. Training is very good at this, and it is the cheapest way to teach a lot of people at once.

Do you lack an answer that needs analysis you cannot do in-house? That is consulting. You need a market entry model, a restructure design, a compensation benchmark. The deliverable is the answer itself, and its quality depends on the analysis behind it. When nobody inside the organisation can produce that answer, buying the expertise is efficient.

Or is someone doing something other than what they would tell you to do? That is the coaching gap. Your director knows she should delegate. She can describe delegation accurately, has been on the course, and could probably teach it. She is still approving everything. More instruction will not close that gap, because instruction was never the constraint.

That third category is where I see the most money wasted, because from the outside it looks like a training problem and it gets a training solution. The programme runs. Feedback scores come back strong. Six weeks later the behaviour is unchanged and nobody can say why.

What each one does to your timeline, and your budget shape

Training is fixed and short. Two days and a cohort, on a date already in the calendar. You know the total cost before you start and it falls per head as the group grows. Impact on knowledge is immediate and easy to test. On behaviour it is slower, weaker, and much harder to attribute.

Consulting runs in weeks to months and produces an artefact. A report, a model, a design. The value transfers on delivery. The risk is that the artefact arrives in an organisation whose capacity to act on it has not changed, which is why so many good reports go unimplemented.

Coaching runs in months and produces no artefact at all. What changes is how one person handles a situation that keeps recurring. The cost per head does not fall as numbers grow, because the unit is one person’s attention. That makes it the most expensive per participant, and the only one of the three that produces nothing you can hand over.

Time to impact separates them further, and I rarely see it addressed in a proposal. Training shows a result on knowledge within the week. Consulting delivers on the day the work is handed over.

Coaching shows nothing useful for six to eight weeks. Then it improves in steps rather than steadily, because the leader has to meet a real situation before anything is tested. Any coach who promises a visible shift by week three is selling you a training outcome.

You are buying three different things: knowledge, an answer, and changed behaviour. Confuse them and it looks like an overpayment, when what has actually gone wrong is the diagnosis.

A decision matrix

The sixth and eighth rows are the ones I would look at hardest. Sequencing matters more than choosing, and the common mistake is buying the second intervention before the first is finished.

When coaching is the wrong answer

I have talked organisations out of coaching engagements.

Structural problems come first, and here coaching is close to cruel. If a leader carries three roles, has no deputy, and reports to two people with conflicting priorities, coaching them to cope better asks an individual to compensate for a design fault.

Fix the structure. The pattern is easy to spot: several capable people in the same role have all struggled the same way.

When you need the same thing taught to many people quickly, do not buy coaching. It does not scale, and the same money buys far more as training.

When the sponsor wants a specific outcome for a specific person, coaching cannot deliver it. Engagements commissioned to make someone more compliant, or to build a paper trail before an exit, fail on their own terms. They also damage trust in every other engagement you run. If the decision has been made, make it.

And when nobody senior has agreed what should change, coaching becomes an expensive conversation with no defined purpose. Agreement on what should change comes first. The engagement cannot produce it for you.

What only coaching does

Coaching changes how a specific person behaves in situations that recur, when the constraint is judgment, identity or pattern rather than knowledge.

The most useful study is the oldest. Olivero, Bane and Kopelman ran it in 1997, inside a single US public agency. Thirty-one managers went through training, then eight weeks of follow-up coaching. Productivity was measured after each stage.

After training, the gain was 22.4 per cent. After coaching, 88.0 per cent.

Read that carefully before you quote it, because almost everyone gets it wrong. There was no control group and no random assignment. These are two time points on the same 31 people, not two groups compared against each other, and the 88.0 is their cumulative gain rather than the amount coaching added on top. Treat it as suggestive and nothing more.

The broader evidence survives harder tests. A 2023 meta-analysis in Academy of Management Learning & Education by de Haan and Nilsson looked only at randomised controlled trials. Across 37 trials and 2,528 people, the effect was moderate and statistically significant.

Two caveats belong with that figure, and I would rather you heard them from me. The authors report signs of publication bias, which means the true effect is probably smaller. And coaching trials cannot blind participants, so randomisation does not fully settle the objection that these studies partly measure enthusiasm. The same authors note that self-reported outcomes show larger effects than observed ones.

It is still the most rigorous evidence the field has, and it says coaching works.

Theeboom, Beersma and van Vianen found something more specific in a 2014 meta-analysis in the Journal of Positive Psychology. Coaching’s single largest effect was on goal-directed self-regulation, the technical term for setting an intention and then following through on it.

That is a real finding, and it is narrower than the industry usually claims. Coaching improves follow-through. It does not fix strategy, and it does not substitute for a decision.

How do you measure the return on executive coaching?

Measure three things: a behaviour defined in observable terms before session one, a structured baseline taken at the start and repeated at the end, and the business consequence that behaviour was supposed to affect. Do not use a return multiple.

That last point needs explaining, because you will have seen the claim that coaching returns seven times its cost. Before you put that in a board paper, know where it comes from.

It is a 2009 study commissioned by what was then the International Coach Federation, the industry’s own trade body. It asked respondents to supply the figures behind the calculation, and by the study’s own account they largely could not. The widely quoted 86 per cent recovery statistic comes from that same survey, not from a second source that corroborates it.

It is a marketing number. A sceptical CFO will find that out with one search, and you will spend the rest of the meeting defending your evidence instead of your proposal.

Start with the behaviour. “Improve executive presence” cannot be observed. “Brings a recommendation rather than three options to the exec meeting” can. Write it down, get the leader and the sponsor to agree it, and do that before session one. If you cannot write that sentence, you are not ready to commission the work.

Then the baseline. A 360 instrument at the start and again at the end gives you a defensible before and after, from the people on the receiving end of the leader’s behaviour. It also removes the argument about whether anything changed. If you are new to running one, the 360 feedback guide for L&D covers how to do it without the process eating the insight.

Last, the business consequence you expected the behaviour to affect. Be honest that it is a consequence rather than a proof. If everything escalated to one person, the measure is how many decisions now resolve below them.

Attribution will never be clean. It does not need to be clean to be persuasive, and a modest claim you can evidence beats a spectacular one you cannot.

If the board wants the evidence in more depth, what the research actually shows on coaching ROI is worth a read before the meeting.

The strongest thing you can tell a board is that independent peer-reviewed research, including randomised trials, shows a real and measurable improvement in performance, and that you have defined in advance what change you expect to see in your own organisation. That survives scrutiny. A dollar multiple from a trade survey does not.

Frequently asked questions

Is executive coaching worth the investment for companies?

It depends entirely on the problem. Coaching is worth it when a capable person already knows what to do and is not doing it, and the constraint is judgment, identity or pattern rather than knowledge. It is not worth it when the gap is a teachable skill, which is cheaper to train, or when the problem is structural, in which case coaching asks an individual to compensate for a design fault. Independent randomised trials show a moderate, statistically significant effect on performance, though the authors of that research also report signs of publication bias.

What is the difference between coaching, training and consulting?

Training closes a knowledge or skill gap and scales cheaply across a group. Consulting supplies an answer you cannot produce internally, and the deliverable is the analysis itself. Coaching changes what one person does under pressure, produces no artefact, and does not get cheaper per head. Confusing them looks like an overpayment but is usually a misdiagnosis.

When is executive coaching the wrong choice?

Four cases. When the problem is structural rather than individual, and several capable people in the same role have all struggled the same way. When many people need the same thing taught quickly. When the sponsor wants a predetermined outcome for a specific person, such as building a record before an exit. And when nobody senior has agreed what should actually change.

How do you measure the return on executive coaching?

Define the behaviour in observable terms before the first session, agreed by both the leader and the sponsor. Take a structured baseline with a 360 instrument at the start and repeat it at the end. Then track the business consequence that behaviour was expected to affect, while being honest that it is a consequence rather than a proof. Avoid return multiples such as the widely quoted seven times figure, which comes from a 2009 trade-body survey whose respondents largely could not supply the figures behind it.

Where this leaves you

Most of the organisations that come to The Clarity Practice® have already bought training that worked exactly as designed and did not produce the change they wanted. The training usually did its job, and the diagnosis stopped one question early.

Ask what kind of gap you are looking at before you ask which vendor to call. Getting that question right is most of the decision. Getting it wrong buys you a competent programme aimed at a problem you did not have, which is the most expensive way to be sensible.

Thirty minutes. A working session, not a sales call. Bring the real version of the problem, not the version written for the budget meeting, and we will work out which of the three you are actually looking at. Start here.