Your Coaching Programme Needs an Exit Test

Do not buy the sequel before testing the first session

You paid for coaching. People liked it. The provider recommends another programme. What, exactly, would make you say no? If you cannot answer that before the pilot starts, the evaluation is doing very little work.

Start with one job the coaching is meant to improve: agreeing a buyer-confirmed next step, handling a particular objection, or preparing a useful handoff. Attendance and satisfaction can tell you whether people took part and liked it. They do not settle whether the work improved.

What the research can—and cannot—tell you

In their workplace-coaching meta-analysis, Cannon-Bowers and colleagues report an aggregate standardized effect size of 0.44 across 11 studies. That is their reported finding, not a percentage increase in sales or an independently verified prediction for your team. [1]

This is broad workplace research, not a forecast for your SaaS team. The checked finding does not establish which sales behaviour will change, how quickly it will change, or what revenue return you should expect. The pilot below is GTMA’s proposed evaluation method, not a method validated by this finding.

Write the decision rule before the pilot

Choose one behaviour you can observe in comparable work. Describe what counts as a good example and what does not. Record a starting sample, decide when you will review another sample, and name the person who will make the decision. Use lawfully available work records and follow your organisation’s privacy rules.

Agree three possible decisions: continue if the target behaviour improves enough to justify the cost; change the approach if people understand it but cannot apply it; stop if the intervention does not address the problem. Set your own threshold before reviewing the results. There is no universal pass mark here.

Track changes in customer mix, staffing, process and workload alongside the pilot. Where practical, use a comparable group or a staggered rollout. Even then, record differences between the groups. A simple before-and-after revenue comparison cannot isolate what the coaching caused.

A fictional pilot with a decision—not a victory lap

Fictional example: A sales manager samples ten comparable discovery notes. Three contain a buyer-confirmed next step. Before a coaching pilot, the manager sets a local target of seven out of ten at the follow-up review. These are invented numbers for this example, not benchmarks.

In this fictional pilot, at follow-up, six notes meet the definition. Two sellers say they still cannot tell the difference between a suggested meeting and a buyer commitment. The manager chooses a short practice exercise and a second check rather than buying a larger programme. The improvement is encouraging, but the target was missed and it does not prove a revenue effect.

Sometimes the better alternative is not more coaching

If the obstacle is an unclear process, missing information or a decision nobody owns, fix that obstacle first. If the skill is understood but not used, examine the workflow and incentives. Coaching is one possible response. Your evaluation should leave room for another answer.

Use the worksheet

Download the three-page guide to set a coaching pilot’s behaviour check, decision rule and follow-up action.

Sources and limits

Workplace coaching: a meta-analysis and recommendations for advancing the science of coaching · Frontiers in Psychology · 2023-10-10. Cannon-Bowers, Bowers, Carlson, Doherty, Evans and Hall (2023). CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Finding paraphrased from Table 3 with original GTMA commentary. Broad workplace evidence, not a SaaS sales forecast. Qualified support; empirical applicability incomplete. The pilot, worksheet and fictional example are GTMA proposals, not methods validated by this paper.

Source support checked 2026-10-13. Practitioner accounts and proposals are identified in the article; a citation does not turn a suggestion into a proven outcome.