Measuring the return on organisation development

Organisation development is usually the first budget line to be questioned when finance tightens. Leadership programmes, culture work and structural redesign all promise better performance, but when the CFO asks what the organisation got for its money, the answer is often a satisfaction score and a few positive quotes. That is not enough, and it shouldn't be.

In this article I set out a practical way to measure the return on OD: how to frame the business problem, build a measurement chain, isolate the effect of your work, and turn the result into money finance will accept. There is a worked example you can adapt, and a short list of first steps.

Why OD struggles to show a return

The honest reason is that most OD teams measure the wrong things. They count participants, completion rates and post-session ratings, because those numbers are easy to collect. In Kirkpatrick's terms, they stop at reaction and learning and never reach behaviour or results.

There are genuine difficulties too. OD outcomes arrive late, often six to eighteen months after the intervention. They are spread across many teams. And they have many causes: a fall in attrition might come from your manager programme, a salary review, a cooling job market or a popular new director.

None of this makes measurement impossible. It means you have to design the measurement at the same time as the programme, not after it. By the time someone asks for the return, the chance to capture a baseline has usually gone.

Start with the business problem, not the programme

Every OD initiative should begin with a problem an executive already cares about, stated in business terms. "Improve leadership capability" is not a business problem. "We are losing too many experienced engineers in their second year, and each replacement delays projects" is.

Once the problem is clear, write the chain that links your intervention to it. I use four links:

1. Intervention. What we will do, for whom and when.
2. Behaviour change. What people will do differently as a result, stated so that someone could observe it.
3. Team outcome. What will change in the teams those people lead or work in.
4. Business result. What that change is worth to the organisation.

For a manager development programme, the chain might read: managers learn to hold regular career conversations; team members report clearer development paths; regretted attrition in those teams falls; the organisation saves on replacement and lost productivity.

Writing the chain does two useful things. It forces you to check that the programme is designed to change the behaviour that matters. And it tells you exactly what to measure at each stage.

Build a measurement chain

For each link, choose one or two measures, a data source and a timing. Resist the temptation to measure everything. A small set of well-chosen indicators, tracked consistently, is far more convincing than a dashboard of thirty.

Leading indicators matter because they give you an early signal. If behaviour has not shifted after six months, attrition is unlikely to follow, and you can adjust the programme before the money has all been spent.

This is also where leadership scorecards help. If managers are held accountable for a few team indicators, such as internal mobility, engagement on growth and regretted attrition, the data you need for OD measurement is collected as part of normal management rather than as a special exercise.

Isolating the effect of your work

The question a sceptical CFO will ask is simple: how do you know it was the programme? You need an answer before they ask.

The strongest practical answer is a comparison group. Most organisations cannot run a true experiment, but they can do something close:

• Staged roll-out. Run the programme for one cohort first and use the managers waiting for later cohorts as the comparison. This is usually the easiest option, because nobody is denied the programme, only scheduled later.
• Matched groups. Compare participants with non-participants who are similar in grade, function, tenure and location.
• Before and after with a trend line. If no comparison group is possible, compare results against the trend that existed before, not just a single "before" number.

Record anything else that changed during the period, such as a pay review, a restructure or a new hiring freeze. You will not be able to remove every confounding factor, but naming them shows the analysis is honest.

A modest, well-evidenced return will do more for your next budget request than an impressive number nobody believes.

Finally, agree the method with finance before launch. If the CFO's team has signed off the comparison approach and the cost assumptions in advance, they are far less likely to dismiss the result afterwards.

A worked example: a manager development programme

The numbers below are illustrative, to show the mechanics. Replace every figure with your own data and your finance team's assumptions.

Say a 3,000-person entity has 300 people managers. Regretted attrition in the teams those managers lead has become a concern, and exit interviews point to weak career conversations. The OD team designs a six-month programme and runs it first for 100 managers, with the remaining 200 scheduled for the following year as the comparison group.

The programme costs 900,000 in the entity's currency for the first cohort, including design, facilitation and managers' time away from work.

Twelve months after the start, the analysis shows:

• In the participating managers' teams (about 1,000 people), regretted attrition fell from 10% to 7%.
• In the comparison managers' teams (about 2,000 people), it fell from 10% to 9%, reflecting a general improvement across the organisation.
• The difference attributable to the programme is therefore around two percentage points, or roughly 20 fewer regretted leavers in the participating teams.

Finance agrees an average cost of 60,000 per regretted leaver, covering recruitment, onboarding and lost productivity during the vacancy. Twenty avoided departures is worth about 1.2 million, against a cost of 900,000. That gives a net benefit of around 300,000 in the first year, before counting any effect in the second year.

Present this as a range, not a single figure. If you assume only half the difference is due to the programme, the benefit falls to around 600,000 and the programme does not yet pay for itself on attrition alone. Say so. A credible range, with its assumptions visible, is what earns trust.

Getting the data in order

Most of the data you need already exists: headcount and movements in the HR system, engagement and pulse survey results, performance ratings and exit interview notes. The problem is usually quality and consistency rather than absence.

In many GCC organisations I see the same issues. Records are split across several systems after years of growth or mergers. Job titles and grades are inconsistent, which makes like-for-like comparisons hard. Leaving reasons are coded vaguely or not at all. Fixing these basics is unglamorous work, but without it no analysis will be trusted.

A few rules make a large difference:

• Define regretted attrition once, with clear criteria, and apply it consistently.
• Segment results in the ways that matter locally, such as nationals and expatriates, or grade bands, so that averages do not hide the real story.
• Protect individual privacy. Report at group level, set a minimum group size, and be open with employees about how their survey and HR data are used.

You do not need an advanced analytics platform to start. A careful analyst, a clean extract and a clear method will outperform an expensive tool used without them.

Reporting the return in business terms

When you report, lead with the business problem and the result, not the programme content. Executives want to know what changed, what it was worth, how confident you are and what you recommend.

A one-page report works well. Include the original problem, the measurement chain, the result as a range, the main assumptions, one or two short stories from managers or employees that bring the numbers to life, and a clear recommendation: extend, adjust or stop.

That final recommendation matters most. Measurement is only worth doing if it changes decisions. If a programme does not produce the behaviour change you designed it for, the most valuable thing your data can do is tell you to stop funding it. The same discipline applies to change programmes more broadly, which I discuss in leading organisational change without change fatigue.

Where to start

1. Pick one current or planned OD programme and write its business problem in a single sentence an executive would recognise.
2. Draft the four-link chain from intervention to business result, and choose one or two measures for each link.
3. Capture a baseline now, and set up a comparison group, ideally through a staged roll-out.
4. Meet finance to agree the method and the cost assumptions before results come in.
5. Schedule a review at six months on leading indicators, with an explicit decision to continue, adjust or stop.

At Humanyx we build measurement into organisation development work from the design stage, so that every programme we help deliver can show what it changed and what that was worth.