Article

Coaching in the Workplace: Choosing the Format That Changes Behavior

The question is not which coaching format is best. It is what behavior you are trying to change, and which format can carry that weight.

10 min read

Key takeaways

  • Coaching, mentoring and training differ in where the answer comes from: the content, the mentor’s experience, or the person being coached.
  • Most leadership development is about behavior someone already understands but is not acting on, which is why training alone so often shifts nothing.
  • One-on-one coaching changes individual behavior; team coaching changes how a group behaves together, and one cannot substitute for the other.
  • Virtual delivery solved an access problem but is not a shortcut, and a common failure is a large license purchase with low activation.
  • Coach matching matters more than platform features, because a poor match teaches people that coaching does not work for them.
  • Manager-as-coach has the widest reach of any format and is the hardest to do well, because most managers default to solving.
  • Choose the format by the unit of change, and decide what you will measure before you buy.

Coaching has become a catch-all.

Coaching in the workplace now covers a six-session executive engagement, a twenty-minute conversation on an app, a manager using coaching questions in a one-on-one, and a peer group that meets monthly. These are different interventions with different costs and different failure modes, and they get compared as though they were interchangeable.

If you are choosing between them, the question is not which is best. It is what behavior you are trying to change, and which format can carry that weight.

Coaching, mentoring, training: the distinction that matters

The textbook answer is that mentoring is advice from experience, coaching is questions that generate insight, and training transfers knowledge. Accurate, and not much help when you are writing a business case.

The practical difference is where the answer comes from.

In training, the answer is in the content. Good for knowledge and process: a new system, a new policy, a technical skill. If the person genuinely does not know something, training is the efficient choice and coaching is an expensive way to get there.

In mentoring, the answer is in the mentor's experience. Good for finding your way through a specific organization, for career decisions, for sponsorship. A mentor can tell you which meeting actually decides things, which is information no coach can generate through questions.

In coaching, the answer is in the person being coached. Good for behavior, self-awareness, and anything where someone already knows what to do but is not doing it.

That last category is where most leadership development actually sits. Which is why buying a course to fix a behavior is such a reliably expensive way to change nothing.

We have set out the difference between training and coaching in more detail elsewhere. Both have a place, and the mistake is using one where the other is needed.

One-on-one coaching

The most established format, and the most expensive per head. It works because the conversation can go wherever it needs to, and because confidentiality lets people say things they will not say in a group.

Where it fails: when the coachee has no real agenda of their own, when the objectives were set by HR rather than by the individual, and when it runs with no connection to the person's manager or role. A coaching engagement that nobody except the coach and coachee knows about will produce insight and very little visible change.

For senior leaders the case is usually clearest, because complexity and isolation rise together. The higher someone goes, the fewer people can tell them the truth, which is the premise of executive coaching for senior leaders as a distinct offer rather than a more expensive version of the same thing.

Group and team coaching

Several people, one coach, a shared session. Cheaper per head, and it does something one-on-one cannot: it changes how a group behaves together.

If your problem is a leadership team that avoids conflict, coaching them separately will not fix the dynamic between them. You will get six better-adjusted individuals and the same meeting.

It is worth separating the two variants, because they are often sold as one. Group coaching brings together people who do not work together, usually at a similar career stage, and its value is peer normalization: discovering that the thing you thought was your private failing is the standard experience of the role. Team coaching works with an intact team on the way they actually operate, and its value is the real dynamic in the room.

Where both fail: when safety is low. In a group where people are competing for the same promotion, or where a power imbalance goes unaddressed, the session produces performance rather than honesty. Putting a leader and their direct reports in the same team coaching session without preparation is a common and expensive mistake.

Virtual and digital coaching

Coaching delivered over video, usually through a platform that handles matching, scheduling and measurement.

It solved a real problem. Coaching used to be reserved for the top of the organization because delivery could not reach enough people. Virtual delivery made it realistic to offer it to middle managers and emerging talent, which is the population where a small shift in behavior reaches the most people, because each of those managers affects a team.

What the format is genuinely better at: geography stops mattering, so you can match a coach on fit rather than on ZIP code, and sessions fit around work instead of requiring travel. For people with caring responsibilities the difference between a video call and a half-day round trip is often the difference between taking up coaching and declining it.

What it is not: a shortcut. The same things still determine whether it works: coach quality, coachee motivation, a real goal, and enough sessions to get past the polite stage. Poor matching produces worse outcomes than no coaching at all, because people conclude that coaching does not work for them and stop being available to it.

A common failure is a large license purchase with low activation. An organization buys several hundred seats, a fraction book a first session, fewer continue past the second, and the program gets written off. The format was not the problem. The absence of a reason to start was.

Activation is a design question rather than a marketing one. It improves when coaching is attached to a moment that already matters to the individual: a promotion, a return from leave, a new team, a first management role. It falls when coaching is announced as a benefit and left for people to discover.

That is the thinking behind our online coaching approach, and why the platform question we care about is how to scale without losing the human core rather than how many seats can be provisioned.

What about AI coaching?

AI coaching sits next to digital coaching rather than inside it, and the distinction is starting to matter commercially.

AI tools are good at a narrower set of things than the marketing suggests, and genuinely good at those: prompting reflection between sessions, rehearsing a conversation before you have it, offering a framework at the moment someone is stuck, and being available at eleven at night.

What they do not currently do is hold complexity. A tool cannot notice that the thing someone keeps not mentioning is the actual issue, and it cannot carry the accumulated context of a relationship in a way that makes a person feel known. Above a certain level of seniority, and in any situation with real emotional weight, people need a human.

Talking Talent's Global Head of Market Solutions, Lucinda Quigley, has put the limit plainly:

AI coaching can only help you so far. Above a certain level, with the complexity and the heaviness that leaders are holding, they need a human.

The useful framing is not AI versus human coaching. It is which parts of the work benefit from being always available and which parts require someone who remembers. We have written more about our own human and digital blend elsewhere.

Manager-as-coach

Not a separate product, and the version with the widest reach.

If managers hold coaching conversations in their ordinary one-on-ones, the effect compounds across the organization in a way bought coaching never can. It is also the hardest to do well, because most managers default to solving, and solving feels like helping.

There is a reason the everyday version carries so much weight. In Social (1), the social neuroscientist Matthew Lieberman argues that our need to connect with other people is more fundamental than our need for food or shelter. Which is why the routine conversation with the person you report to carries more weight than its place in the diary suggests.

The realistic ambition is not to turn managers into coaches. It is to get them to use a coaching stance in perhaps a fifth of their conversations, and to know which fifth.

Internal or external coaches?

A question most business cases have to settle, and the answer is usually both, for different work.

Internal coaches are cheaper per hour, understand the organization's politics without being briefed, and can be deployed quickly. They are the right choice for career conversations, transitions into first management roles, and anything where knowing how the organization actually works is most of the value.

Their limit is confidentiality, and it is a real one. A coachee who suspects their coach has lunch with their manager's manager will not raise the thing they most need to raise. Internal coaching programs that ignore this produce polite, useful, superficial conversations.

External coaches cost more and start with less context. What they buy is the ability to say anything. For senior leaders, for situations involving the person's own manager, and for anything touching on whether someone wants to stay, the distance is the point.

The split that tends to work is external coaching at senior level and for sensitive transitions, internal coaching for volume and for organization-specific development, with clear published rules about what is confidential in each. The rules matter more than the ratio.

How much coaching is enough?

The question every business case has to answer, and the one providers are often vague about.

Very short engagements rarely get past the polite stage. The first session is orientation, and the real agenda often surfaces in the second. Providers commonly structure individual coaching over six to twelve months, and the question worth asking any of them is not how many sessions you get but at what point they expect behavior change to become visible, and how they will know.

Longer is not automatically better. Open-ended coaching without a defined goal becomes a valued conversation rather than a development intervention, which is a pleasant outcome to pay for and a difficult one to justify twice.

Choosing between them

Match the format to the unit of change.

  • Individual behavior under pressure: one-on-one coaching.
  • How a team works together: team coaching.
  • Peer normalization for a cohort at the same career stage: group coaching.
  • Consistent capability across a whole management layer: digital coaching, with manager-as-coach development running alongside it.
  • Practice and reflection between sessions: AI tools, alongside a human coach rather than instead of one.

Most organizations get this wrong in a predictable direction. They buy the format that is easiest to procure rather than the one that fits the problem, then judge coaching on the result.

What to ask a coaching provider

Five questions separate providers quickly.

  1. How do you match coaches, and what happens if the match is wrong? A provider without a straightforward reassignment process is telling you something about how often they expect it to be needed.
  2. What are your coaches' credentials, and how many hold them? Accreditation is not a guarantee of quality, but the answer reveals whether quality is managed or assumed.
  3. What is your activation rate on comparable programs? If the answer is a license count rather than a percentage of people who completed more than two sessions, ask again.
  4. What will you report to us, and what will you not? A provider willing to say clearly what stays confidential is more trustworthy than one promising full visibility.
  5. What happens after the engagement ends? Behavior change that depends on the coach being present is not behavior change.

What to measure

Whichever format you choose, decide in advance what you will measure.

Satisfaction and self-report data has its place in a business case. On its own, though, it tells you whether people valued the experience rather than whether behavior changed.

Better measures exist and are not exotic. Behavior observed by others at ninety days. Retention in the coached population against a comparable group. Internal promotion and lateral move rates. Progress against the goal the coachee set at the start, rated by the coachee and their manager separately, because the gap between those two ratings is often the most interesting number you will get.

The International Coaching Federation's summary of coaching ROI research (2) reports that 72% of respondents in its 2023 Defining New Coaching Cultures work saw a link between coaching and higher employee engagement, and cites an older PwC and Association Resource Center survey putting average return at around seven times the cost of the coach. Figures like these are useful for a business case and useless as a substitute for measuring your own.

Talking Talent's published figures show what this looks like when it is tracked rather than surveyed: an 83% retention rate among program participants at one client, retention rising from 84% to 97% with maternity coaching at another, and a 144% rise in post-maternity retention within two years at a third.

Coaching is not one product you pick a format of. It is a set of distinct interventions, and choosing between them is a design decision rather than a procurement one.

The question is not which coaching format is best. It is what you are trying to change, and whether the format you have bought can carry it.

Working out which coaching format fits the problem you actually have? Get in touch with us and we will help you think it through.

References

  1. Matthew D. Lieberman, Social: Why Our Brains Are Wired to Connect (Crown, 2013)
  2. International Coaching Federation, Coaching Statistics: The ROI of Coaching in 2024 (2024)

Questions about coaching in the workplace

What is coaching in the workplace?

Workplace coaching is a structured conversation in which a coach helps someone reach their own answer rather than supplying one. It is used for behavior change, self-awareness, decision-making and transitions. It differs from training, where the answer sits in the content, and from mentoring, where it sits in the mentor’s experience.

What is the difference between coaching and mentoring?

A mentor shares their own experience and offers advice; a coach asks questions so the person reaches their own conclusion. Mentoring suits navigating a specific organization, career decisions and sponsorship. Coaching suits behavior, self-awareness and situations where someone knows what to do but is not doing it.

What types of coaching are used in organizations?

Employee coaching in organizations usually takes four forms: one-on-one coaching, group and team coaching, virtual or digital coaching delivered through a platform, and manager-as-coach, where managers use coaching skills in their own conversations. Each suits a different unit of change, and each has a distinct way of failing.

Does virtual coaching work as well as face-to-face coaching?

For most purposes, yes. Virtual delivery removes the geography constraint, which usually improves coach matching rather than worsening it. What determines effectiveness is coach quality, the coachee’s motivation and a real goal, not the medium. The risk specific to virtual programs is low activation rather than low quality.

What is digital coaching?

Digital coaching is coaching delivered through a platform that handles matching, scheduling, content between sessions and measurement, usually with video conversations at its center. The term is sometimes used for AI-only tools, which is a different thing: useful for prompts and practice, limited where complexity and emotion are involved.

How many coaching sessions does someone need?

The first session is orientation and the real agenda often surfaces in the second, so very short engagements rarely get past the polite stage. Providers commonly structure individual coaching over six to twelve months. The more useful question is at what point a provider expects behavior change to become visible, and how they will measure it.

How do you measure whether workplace coaching is working?

Measure behavior rather than satisfaction. Useful indicators include behavior change observed by others at around ninety days, progress against the goal the coachee set themselves, retention in the coached group against a comparable group, and internal promotion rates. Satisfaction scores tell you whether people enjoyed it, not whether anything changed.

Connection Audit

How connected is your organization?

Take the three-minute Connection Audit for an instant snapshot across four areas of connection.

Take the Connection Audit

Sample snapshot

Overall 6.9/10

  • Connected Life

    9.1/10

    strong
  • Connected Talent

    7.4/10

    moderate
  • Connected Leaders

    6.2/10

    moderate
  • Connected Culture

    4.8/10

    low
16 statements · 3 minutes to completeYour results, explained in a complimentary 30-minute conversation to turn insight into action.

Newsletter

The Talent Brief

An occasional email for HR and people leaders: what our coaches and leadership team are writing about, news of our work with clients, and events where you can hear from us.

We’ll send you The Talent Brief by email. You can unsubscribe at any time. See our privacy policy.