Capability partner or training vendor? A decision guide

A training vendor sells delivery: a course, a trainer, a day in the room, built to a specification you write. A capability development partner is accountable for a behavior that is still visible months later: it diagnoses what is missing, designs the system that produces it, and re-measures the same behavior against the same rubric. Vendors are the right buy for a defined skill or a credential. Partners are the right buy when the result has to survive contact with Monday morning.

Most organizations do not choose badly between these two. They choose without noticing there was a choice: they write a training brief, collect three quotes, compare them on price and trainer CV, and only discover eight weeks later that they bought delivery when they needed change.

This guide is for whoever writes that brief. It is not an argument that partners are better than vendors. A vendor is often exactly right, and we say plainly below when. It is an argument that the two answer different questions, and that most wasted development budget is a correct answer to the wrong one.

TL;DR

• The difference is where accountability ends. A vendor delivers to a specification you write, and their accountability ends when the session does. A partner helps write the specification after a diagnosis, and is accountable for the behavior.
• A vendor is the right buy when the need is a defined teachable skill, when you need an externally recognized credential for a tender or audit, when one person needs to grow, or when you genuinely need a one-off.
• You need a partner when the behavior has to survive contact with Monday, when you are developing a population rather than a person, when someone will ask you to prove it worked, or when the problem is not clear yet.
• The "only 10% of training transfers" figure is folklore: Georgenson offered it rhetorically in 1982 with no evidence. Saks and Belcourt surveyed 150 organizations and found 62% applying what they learned immediately, 44% at six months, 34% after a year.
• Ask four questions in the first conversation. What will you measure, and when? What do you need from our managers? What happens between sessions? And what would make you tell us not to run this?
• Most organizations need both. The failure mode is not using vendors; it is using a vendor purchase to answer a capability question, and then concluding that training does not work.

The two products hiding behind one word

A training vendor sells execution against your specification. You define the topic, the audience and the dates; they bring content and a trainer. Their accountability ends when the session ends and the feedback forms come back. This is a real, honest product with a clear deliverable, and it is priced accordingly.

A capability development partner is accountable for the behavior, not the event. The specification is the thing they help you write, because the presenting request ("we need a communication course") is usually a symptom. They diagnose which behavior is actually missing, build the system that produces it (facilitated practice, reinforcement between sessions, the participant's own manager involved) and measure the same behavior before and after against the same rubric.

The invoices can look similar. What you can hold each of them to does not.

When is a training vendor the right buy?

Often, and we turn down work for exactly these reasons. When the need is a defined, teachable skill. When you need an externally recognized credential for a tender or an audit. When one person needs to grow. When the behavior is in place and only the knowledge is missing. Or when you genuinely need a one-off.

Not every development need calls for a partner. A vendor is often the right answer, and we turn down work for exactly the reasons below.

A vendor is the better buy when:

• The need is a defined, teachable skill. Excel, a project management method, a piece of legislation, a certification syllabus. The content exists, it is stable, and knowing it is the outcome.
• You need an externally recognized credential. When the value is the signature on the certificate (a tender requirement, an audit, a regulator, a client's supplier questionnaire), an accredited provider is the only route. No amount of internal capability work substitutes for it.
• One person needs to grow. Nobody should build a program for a single participant. Send them somewhere good.
• The behavior is already in place and only the knowledge is missing. If your managers already hold difficult conversations well and simply need to know the new policy, you have a briefing problem, not a capability problem. Buy the briefing.
• You genuinely need a one-off. A team building, a kickoff, an energizer before a hard quarter. Real needs. Just do not expect them to change how anyone works in March.

In all of these the return is bounded and immediate, and the specification is something you can write with confidence. That is the signature of a vendor purchase, and there is nothing second-class about it.

When do you need a capability partner?

When the behavior has to survive contact with Monday. When you are developing a population rather than a person. When someone will eventually ask you to prove it worked, which means measurement designed in at the start. When you have already bought training and it did not stick. Or when the problem is not clear yet.

• The behavior has to survive contact with Monday. You do not want people to know how to give feedback. You want feedback to actually happen in the one-to-one on Tuesday. That gap is not a content problem, and no amount of better content closes it.
• You are developing a population, not a person. Twelve managers, three teams, a whole layer. The unit is the group, and consistency between them is part of the result.
• Someone will ask you to prove it worked. If a CFO, a board or a sponsor will eventually ask what changed, you need a measurement designed in at the start. Retrofitting evidence onto a completed course is not possible.
• You have already bought training and it did not stick. In our experience this is the most common trigger, and the most diagnostic. If the content was good, the trainer was rated highly, and nothing changed, the missing piece was never content. It was the architecture around the learning.
• The problem is not clear yet. "Communication is bad" is a symptom. If you cannot name the specific behavior that is missing and the situation where it fails, you are not ready to write a brief; you need a diagnosis first, and a vendor is not incentivized to give you one.

What does the research actually say?

That the 10% figure is folklore. Georgenson offered it rhetorically in 1982 with no evidence, as Farrington pointed out in 2011. Saks and Belcourt found 62% of employees applied what they learned immediately, 44% at six months, 34% after a year. Training works; it stops working, and it stops after the vendor's accountability has ended.

One number gets quoted constantly in this industry: only 10% of training transfers to the job. It is worth knowing where it came from. It first appears in a 1982 article by Georgenson, who offered it rhetorically and cited no evidence for it whatsoever, a point made directly by Farrington in a 2011 rebuttal titled Training transfer: Not the 10% solution.

The real figures are both better and more useful. Surveying training professionals across 150 organizations, Saks and Belcourt (2006) found that 62% of employees applied what they had learned immediately after training, 44% at six months, and 34% after a year.

Read that curve again, because the whole argument is in it. The problem was never that training does not work. It is that it stops working. Nearly half of what you bought decays within a year, and it decays after the vendor's accountability has ended.

What slows the decay is not better content. Baldwin and Ford's foundational 1988 review established that the work environment a participant returns to is one of the three inputs determining whether training transfers at all, not an afterthought to the design, but part of it. Blume et al. (2010), a meta-analysis of 89 studies in the Journal of Management then confirmed a supportive work environment among the reliable predictors of transfer.

On the practice itself, Cepeda et al. (2006) in Psychological Bulletin (839 assessments across 317 experiments) is the evidence for spacing: material practised at intervals is retained better than the same material practised in one block, and the longer you need it to last, the wider those intervals should be.

Which is a research-shaped way of saying the day in the room was never the part that decides.

What actually differs?

What you can hold each of them to. A vendor delivers to your specification and their accountability ends when the session does; a partner is accountable for the behavior, helps write the brief after a diagnosis, builds the manager's reinforcement in, and measures the same behavior before and after against the same rubric.

Four questions that tell you which one you are talking to

Ask these in the first conversation. The answers separate the two categories faster than any brochure.

1. "What will you measure, and when?" A vendor answers with feedback scores and completion. A partner answers with a behavior, a rubric, and two points in time. If the answer is "we can do an evaluation form", you are talking to a vendor.
2. "What do you need from our managers?" A vendor needs a room and a date. A partner needs the participants' direct managers for a small, specific commitment every week, and will tell you the program underperforms without it.
3. "What happens between sessions?" If the answer is "nothing" or "an optional workbook", the forgetting curve is the design. Practice between sessions is where behavior is actually built.
4. "What would make you tell us not to run this?" This is the most revealing question in the list. Anyone who cannot name a situation in which they would decline the work is selling delivery. A partner has a fit test and will use it.

The honest answer: most organizations need both

This is not a conversion story where you replace every vendor with a partner. Buy vendors for what vendors are good at: credentials, defined skills, individual development, the one-off. Buy a partner for the behaviors your strategy actually depends on, where "we ran a course" is not going to be an acceptable answer six months from now.

The failure mode is not using vendors. It is using a vendor purchase to answer a capability question, then concluding that training does not work.

Where Azimut sits

We are a capability development partner, and we are explicit about what that commits us to: we start with a diagnosis rather than a catalogue, we involve the participant's manager as part of the design rather than as an afterthought, and we measure the same behavioral clusters before and after so the change is comparable rather than impressionistic. That is the whole of Capability OS.

It also means we decline work. If you need an accredited certification for a tender, or one person developed, we will say so and point you elsewhere; that is a vendor purchase and you should make it as one.

Across 600+ programs and 20,000+ participants since 2008, the pattern has not changed: the organizations that get lasting change are not the ones who bought better content. They are the ones who bought the system around it.

FAQ

Is a training vendor ever the right choice? Often, and we turn down work for exactly these reasons. A vendor is the better buy when the need is a defined, teachable skill, when you need an externally recognized credential for a tender or an audit, when one person needs to grow, when the behavior is already in place and only the knowledge is missing, or when you genuinely need a one-off. In all of those the return is bounded and immediate, and you can write the specification yourself with confidence.

We bought training and nothing changed. What was missing? In our experience this is the most common trigger for calling a partner, and the most diagnostic. If the content was good, the trainer was rated highly, and nothing changed, the missing piece was never content. It was the architecture around the learning: a diagnosis of which behavior is actually absent, practice between sessions, a manager who reinforces, and the same measurement before and after.

Where does the "only 10% of training transfers" figure come from? It first appears in a 1982 article by Georgenson, who offered it rhetorically and cited no evidence for it. Farrington made that point directly in a 2011 rebuttal titled "Training transfer: Not the 10% solution". The better-sourced figures come from Saks and Belcourt (2006), who surveyed training professionals across 150 organizations and found that 62% of employees applied what they had learned immediately after training, 44% at six months, and 34% after a year. The problem was never that training does not work. It is that it stops working, and it stops after the vendor's accountability has ended.

What should I ask a provider in the first conversation? Four questions separate the two categories faster than any brochure. What will you measure, and when? What do you need from our managers? What happens between sessions? And what would make you tell us not to run this? A vendor answers the first with feedback scores and completion, needs only a room and a date for the second, has nothing or an optional workbook for the third, and cannot name a situation in which it would decline the work.

Do we have to replace our training vendors? No. This is not a conversion story. Buy vendors for what vendors are good at: credentials, defined skills, individual development, the one-off. Buy a partner for the behaviors your strategy actually depends on, where "we ran a course" is not going to be an acceptable answer. The failure mode is not using vendors. It is using a vendor purchase to answer a capability question, then concluding that training does not work.

Sources

• Saks, A. M., & Belcourt, M. (2006). An investigation of training activities and transfer of training in organizations, Human Resource Management. The 62% / 44% / 34% application figures across 150 organizations.
• Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L. (2010). Transfer of Training: A Meta-Analytic Review, Journal of Management. 89 studies; supportive work environment among the reliable predictors of transfer.
• Baldwin, T. T., & Ford, J. K. (1988). Transfer of Training: A Review and Directions for Future Research. The framework establishing work environment as one of three inputs determining transfer.
• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed Practice in Verbal Recall Tasks, Psychological Bulletin. 839 assessments across 317 experiments; the spacing effect.
• Georgenson, D. L. (1982). The problem of transfer calls for partnership. The origin of the "10%" figure, offered rhetorically, with no evidence cited.
• Farrington, J. (2011). Training transfer: Not the 10% solution, Performance Improvement Quarterly. The rebuttal.
• Azimut internal program data: 600+ programs delivered, 20,000+ participants, 9.27/10 average trainer score (since 2008).

By Szilárd Kacsó