The Test and Learn approach is gaining momentum across government, and it brings with it two important questions about how everything fits together. For leaders: if teams are experimenting and adapting, how does this tie back to our wider strategy? For delivery teams: now that we test and learn, are strategy documents, OKRs and change portfolios still an appropriate framework?

Both questions assume test and learn and strategic alignment are in tension. I think the opposite is true. Maintaining a golden thread, from the digital strategy through OKRs to the portfolio and team, is the thing that stops test and learn collapsing into a collection of disconnected pilots. What changes is what flows along it.

The ‘golden thread’ is the traceable line connecting an organisation’s strategy to the daily work of its teams. The strategy names the outcomes worth pursuing. OKRs translate those outcomes into objectives a team can own and results it can measure. The portfolio allocates people and money across those objectives. And any team, asked why its work matters, can answer by walking back up the line. When a link is missing, work and intent drift apart quietly, and the drift surfaces months later as a delivery surprise.

Strategy that learns

Giles perfectly describes how strategies can be more meaningful in The strategy is enquiry. Typically, government strategies are long, inert documents full of promises that go quiet, to make them more valuable artefacts, “stop promising what you’ll deliver, start promising what you’ll discover”. Publish the strategy as a living record of what the organisation knows, doesn’t know and is learning.

I think a strategy written this way is the only kind that can sit at the top of a test and learn organisation, because it’s a great way for people inside and out to point at the learnings made over time. A “we will” strategy loses weight with each pivot. A “we don’t know yet” strategy is strengthened by it. The strategy stops being a forecast and becomes the organisation’s running account of what it’s discovering about its hardest problems, which is precisely the institutional capability that programmes like Test, Learn and Grow are trying to build.

What strategy must keep, is direction. Outcomes worth pursuing, problems worth solving, bets worth making, and the things deliberately out of scope. Teams can test and learn their way towards an outcome. Nobody can test and learn their way towards no destination at all.

OKRs as the translation layer

I’ve written before about using OKRs to point teams at outcomes, everything in that post matters more under test and learn.

An objective carries strategic intent down to a team in outcome language: the thing worth achieving, with the solution left open. Key results make progress measurable. The OKR cadence already matches the test and learn cadence; a quarterly review of evidence is exactly the decision point an experimenting team needs. The adaptation for test and learn is simple: early in an enquiry, key results measure learning. “We know whether X user will use Y service, with evidence” is a legitimate key result, and often the most valuable one a team can deliver in a quarter. As confidence grows, key results shift from learning to performance: uptake, time saved, outcomes scaled.

Run this way, OKRs become the grammar that lets a strategy of enquiry talk to a portfolio of bets. The strategy says what we need to discover. The OKR says what this team will discover this quarter, and how we’ll know they have.

Portfolio as the productivity centre

A portfolio’s primary objective is to allocate the organisation’s capacity across bets, and rebalance as the evidence develops and priorities adjust. Scale what’s working, stop what isn’t, and protect the discoveries that are too early to judge. This is a more demanding job than traditional portfolio management, which mostly tracks whether things are on schedule, and it’s the reason I keep arguing for lean governance, steering on evidence at the cadence teams work to.

Senior leaders extend trust to teams when they can trace the line from a team’s work to an outcome they care about, and see clear evidence moving along it. This golden thread is what makes autonomy and accountability auditable. A team with a clear objective, learning-based key results and a portfolio that reviews the evidence can be left alone because this connection across each exists. Remove the thread and leaders revert to the only control they have left, which is asking for plans and status reports.

The thread now runs both ways

The golden thread concept was originally designed to cascade. Strategy at the top, broken down into OKRs, allocated through a portfolio, delivered by teams. Intent flowed down. In a test and learn organisation the thread becomes a loop, because value flows upwards.

A team’s tests produce insights. The OKR review turns those learnings into evidence about an outcome. The portfolio uses that evidence for any rebalancing decisions. And, this is the step almost every organisation misses, the strategy absorbs the learning and updates what it claims to know. Giles’s living strategy recommendation is exactly the mechanism for that last step, the place where “we don’t know enough about X” becomes “here’s what we found out about X, and here’s what it changes”.

An organisation whose strategy never changes in response to its teams’ learnings is throwing away gold dust.

What this looks like in practice

For leaders wanting to check their own organisation, the thread in a test and learn world has four working parts. A strategy, ideally a living one, that states outcomes, knowns and unknowns, and is updated as insights land. OKRs that translate outcomes to teams, with learning-based key results tested early and performance-based ones measured later. A portfolio that meets at the cadence of the evidence and genuinely stops things irrespective of the sunk cost. And open communication with every team so they can see their role across the thread.

None of this requires seismic changes. You could visualise the whole thread for a mid-sized department or programme on a wall. That’s precisely what we did at GDS back in the day! What it buys is the difference between an organisation that learns and an organisation that merely runs a random set of pilots.

Questions worth asking

1. Could a team in your organisation trace the link from their current work to a strategic outcome in under a minute?
2. When did your strategy last change because of something a team learned?
3. Do your OKRs include “we found out” as a result, or do teams have to disguise learnings as delivery?

The strategy is delivery taught government that shipping things (even if they’re small) beats talking. Test and Learn teaches us that learnings beat distant promises. The golden thread is how an organisation does both at once without getting lost, and it has never been more worth holding onto.

Useful reading

1. The strategy is enquiry (Giles Turnbull)
2. Using OKRs and kick-offs to point teams at outcomes
3. Lean governance for digital service delivery
4. Test, Learn and Grow (Government Outcomes Lab)
5. The Radical How (Public Digital)