Assurance Versus Delivery: The Tension You Are Not Meant to Resolve
Every attempt to resolve the tension ends the same way: either assurance is absorbed into delivery and stops being assurance, or it hardens into an adversary and delivery quietly routes around it.
The Tension Everyone Tries to Solve
Every major programme I have observed carries a fault line running through the middle of it. On one side sits delivery: the people accountable for turning intent into outcome, under time pressure, with imperfect information, wanting momentum and the freedom to make decisions and move. On the other sits assurance: the people accountable for telling the truth about whether the programme is really as healthy as it claims, wanting distance, evidence, and the standing to say uncomfortable things. The two are not enemies. But they want different things, and the difference does not go away.
What I want to argue is simple, and slightly against the grain of how we usually talk about this. The tension between assurance and delivery is not a problem to be solved. It is a structural feature to be managed. And most of the trouble I have watched programmes get into comes from the honest but mistaken belief that with the right operating model, the right reporting lines, the right terms of reference, the tension can finally be engineered away.
It cannot. And the attempt to do so tends to make things worse.
The Two Ways Resolution Fails
When an organisation sets out to “resolve” the friction between assurance and delivery, it almost always lands in one of two ditches.
The first is capture. In the name of being helpful, collaborative, and “close to the programme,” the assurance function is drawn steadily inward. Its reviewers sit in the delivery meetings, share the delivery objectives, come to depend on the delivery leadership for their access and their standing. Relationships warm. Reports soften. The hard question that needed asking in month four gets deferred to month five, and then to the review after that. By the time an independent eye would have raised the alarm, there is no longer an independent eye — only a friendly one. Assurance has been captured, and capture is invisible precisely because it feels like good working relationships.
The second ditch is the opposite. In the name of independence, assurance is held so far at arm’s length that it becomes an adversary. Its reviews arrive as verdicts from outside, disconnected from the realities the delivery team is wrestling with. The team learns that assurance is something that happens to them, an examination to be survived rather than a resource to be used. So they manage it: they present the tidy version, they route decisions around the review points, they treat the assurance calendar as an obstacle course. The function retains its independence and loses its influence. It sees clearly and changes nothing.
“Every attempt to resolve the tension ends the same way: either assurance is absorbed into delivery and stops being assurance, or it hardens into an adversary and delivery quietly routes around it.”
What is striking is that both failures are pursued with good intentions. Capture is the failure mode of organisations that value collaboration. Adversarial distance is the failure mode of organisations that value independence. Each is a virtue pushed until it curdles — and each is what you get when you try to settle a tension that was never meant to settle.
Why It Is Structural
The reason the tension never resolves is that assurance and delivery are optimising for genuinely different, genuinely legitimate things.
Delivery is optimising for forward motion under uncertainty. Its job is to commit before all the facts are in, because waiting for certainty is itself a failure. It needs confidence, pace, and a degree of protective optimism — you cannot lead a hard programme while broadcasting every doubt.
Assurance is optimising for truth independent of motion. Its job is to be unmoved by the programme’s momentum, to notice when confidence has drifted into wishful thinking, and to be willing to say so even — especially — when it is inconvenient. It needs distance, scepticism, and enough separation that it does not catch the programme’s optimism like a cold.
These are not reconcilable into a single stance. You cannot be simultaneously committed to the plan and independent of it. The value of assurance is its difference from delivery; erase the difference and you erase the value. The tension is not a defect in the operating model. It is the operating model working. A programme with no felt tension between assurance and delivery has not achieved harmony — it has lost one of the two voices.
A programme where assurance and delivery never disagree is not a well-run programme. It is a programme where one of them has stopped doing its job.
Managing What You Cannot Solve
If the tension is structural, the practitioner’s task changes shape. The question is no longer “how do we remove the friction?” but “how do we keep the friction productive — sharp enough to surface truth, but not so sharp that it severs the relationship?” That is a matter of continuous management, not one-time design. A few things, in my experience, make the difference.
- Protect the independence of the line, not the coldness of the relationship. Independence lives in who the assurance function reports to, who controls its access, and who can edit its findings — not in how warm its people are. You can have reviewers who are trusted, even liked, provided their conclusions cannot be softened by the people they review. Keep the reporting line clean and you can afford the relationship to be humane.
- Make the disagreement visible, not buried. The dangerous programme is not the one where assurance and delivery disagree; it is the one where the disagreement never reaches the people who should weigh it. The governance forum’s job is to hear the tension — to have the delivery view and the independent view put side by side in front of a decision-maker who can hold both. When the friction is surfaced to the right level, it becomes information. When it is resolved quietly beneath that level, it becomes risk.
- Separate the timing of assurance from the rhythm of delivery. Assurance that only ever reports on the programme’s own cadence, using the programme’s own materials, is half-captured already. Independent review needs its own points, its own evidence, and the standing to look where it chooses rather than only where it is shown.
- Name the tension out loud, at the start. Much of the damage comes from each side experiencing the other’s stance as bad faith — delivery reading scepticism as obstruction, assurance reading momentum as recklessness. Saying plainly, on day one, that the two functions are meant to pull in different directions, and that this is by design, converts a personal grievance into an understood structure. People manage a tension they have been told to expect far better than one they think they are failing to resolve.
The Stance That Works
What all of this asks for is a particular temperament in the people who govern the programme — a willingness to live inside an unresolved tension without flinching. The instinct of most capable leaders is to close open questions; an open tension feels like a job half done. But some tensions are load-bearing. Remove them and the structure they were holding up comes down.
The strongest programme leadership I have seen does not resent assurance for slowing things down, and does not weaponise delivery pressure to shout it down. It treats the independent voice as a genuine second opinion on its own most important decisions — valuable precisely because it is not invested in the plan. And the strongest assurance leadership does not confuse independence with hostility, nor mistake the ability to see a problem for the achievement of fixing one. It stays close enough to be heard and separate enough to be trusted.
Hold it that way and the fault line running through the middle of the programme stops being a weakness. It becomes the thing that keeps the programme honest: two legitimate views of the same reality, kept in tension on purpose, neither allowed to silence the other. That is not a failure to resolve the problem. That is the resolution — the only one the structure actually permits.