The well-led question, from the inspector's chair

I inspected against the five key questions for the last decade of my time at CQC. Safe, effective, caring, responsive, well-led. By the end of any inspection I had a working view on all five, but I was the most certain about well-led the earliest in the day. Sometimes within the first hour. The reason is that well-led is the question that answers itself in how the leadership team talks to the inspector. Not in what they say. In how the sentences are constructed and what comes up unprompted in the first conversation.

This piece is about what I was actually reading for in those conversations, and in the records that followed. The standard inspection guidance describes the well-led question at the level of the framework. What I want to talk about is what inspectors do with it on the day.

The well-led key question in the official framework covers vision and culture, governance and management, engagement, learning and improvement. Six themes depending on which version of the guidance you read, with the four-point rating sitting on top.

What we were actually asking ourselves, sat in the registered manager's office on the first morning, was something narrower. Does this leadership team know what their service is doing today? When something goes wrong, do they hear about it, or do they find out at the next quarterly meeting? When they hear about it, do they act, or do they note it and move on? Are they curious about their own service, or have they stopped looking?

The well-led rating ultimately reflected the answer to those four sub-questions. Everything else (the policies, the strategy documents, the team-day slides) was either evidence for the answer or noise around it. I learned over the years to focus on the answer and not get distracted by the noise.

The opening conversation with the registered manager was where I formed my working hypothesis on well-led. What I was listening for, not what I was asking about.

Could the registered manager describe the shape of the last quarter without checking notes? Not from memory of specific incidents, but the shape: what had been busy, what had been quiet, what was on their mind, what they had been working on. Managers who could answer that within thirty seconds had a close grip on the service. Managers who said “let me check with the team” or pulled out a folder were typically running services where the team knew more about the operational reality than the leadership did. That was already a well-led signal before I had read a single document.

Did the registered manager volunteer the difficult bits, or did I have to find them? The strongest registered managers told me, unprompted in the first half hour, about the thing that was currently bothering them. The complaint pattern they were watching. The staff member they were worried about. The piece of equipment they had been waiting six weeks for. These were not pre-emptive confessions to avoid worse findings; they were the natural way a person who runs an open service talks about it. Registered managers who presented the service as uniformly fine were almost always running a service with hidden parts they did not know about.

How did the registered manager talk about their team? In the well-led services, the registered manager talked about named people, by what each person was good at and where each person was stretched. In the less well-led services, the registered manager talked about the team as a collective noun, in policy-language, in capability tiers. The latter signalled distance from the day-to-day.

After the opening conversation, the next read was the records. The well-led records sample was different from the safe-records sample. For safe, I read individual incident records and their closures. For well-led, I read the aggregate views: the meeting minutes across six months, the risk register top ten, the improvement-actions register overdue list, the complaints theme report. The well-led question is structural, so the well-led evidence is structural too.

What I was reading for, in priority order.

Did the leadership team have a regular venue where the records actually got read? Most providers have a clinical governance meeting on the calendar. The question was whether the meeting actually read the data, or whether it received a slide from someone who had read it. The latter is fine if the slide generates discussion; it is not fine if the meeting just nods at the slide and moves to the next agenda item. I could tell which by reading the minutes for the kind of follow-up questions a meeting that had actually engaged would produce.

Were the patterns visible to the leadership team visible in the data? If the registered manager had told me they were watching a complaint pattern, the complaints register should show it. If the management team had told me they had been working on staff retention, the staffing dashboard should show the cadence of that work. Where the manager's verbal account matched the records, the leadership team was reading their own data. Where it did not, something was going on: either the manager was describing the service as they wanted it to be, or the data did not surface what the manager was actually looking at, or the manager had stopped looking at the data.

Were the improvement actions tied to the meetings that opened them? When I sampled the improvement-actions register, the actions should have been traceable back to a specific meeting's minutes. The well-run services had that chain. The services with a well-led problem had improvement actions floating loose, with no clear origin, sometimes opened by individuals without the meeting structure noticing.

Did the risk register live? The risk register is the single most diagnostic well-led document if you read it for review dates rather than for content. A register where the top five risks have been reviewed in the last quarter is alive. A register where the top five risks have not been touched in twelve months is decorative. The decorative state is the well-led finding, not the risks themselves.

The four-point rating system arrived in 2014 and 2015. The bands carry a fair amount of weight in commercial life now, so it is worth being clear about what actually distinguished them on the well-led question.

Requires Improvement on well-led looked like a leadership team that had stopped looking. Meetings irregular or thinly minuted, risk register stale, improvement actions opened without a closure trail, the manager describing the service in policy-language rather than concrete terms. The services in this band were almost never badly intentioned. They were stretched, or recently transitioned, or running on the previous registered manager's system. The well-led work needed was usually fixable.

Good on well-led looked like a leadership team that knew what was happening. The cadence of governance meetings ran. Risk register reviews happened on the cadence the policy named. Improvement actions opened and closed with evidence. The manager could describe the service concretely. The team brought concerns to the manager and the manager noticed. This was the steady state of running a competent CQC-regulated service.

Outstanding on well-led was rarer and looked different. The records of a Good service looked similar to those of an Outstanding service at first glance. What separated them was whether the leadership team was actively curious about its own service. Did the meeting minutes show the team asking questions of itself? Did the improvement actions reach beyond the obvious fixes into the harder ones, like culture, training quality, or staff retention? Did the registered manager volunteer something the team was experimenting with, not because the regulator required it but because the team thought it might help? Outstanding services had that quality of self-questioning.

One thing worth saying explicitly. Outstanding is not a higher mathematical score than Good. It is a different shape of service. The four-point rating system was introduced in 2014 to 2015 and it is the framework most providers operate under now; in the pre-2014 era the verdict was compliance against essential standards, with possible compliance actions or concern grades, and the same shape distinctions applied in different vocabulary. The shape of a well-led service has been recognisable across both regulatory eras.

The honest answer is that well-led is not something you build in the four weeks before an inspection. What inspectors read for is the trailing twelve months of records and the way the leadership team talks about the service today. Both of those are cumulative. There is no short-cut.

What you can do is start running the system openly from now on. Put the governance meetings back in the diary at the cadence the policy names, even if it has slipped. Minute the decisions, not just the discussion. Open the improvement actions in the register at the meeting that decides them, with a named owner and a date. Review the risk register on its cadence. At the next meeting, read the previous meeting's actions back and confirm completion or report progress. Three to six months of doing that produces a records trail that reads like a well-led service to any sampling inspector.

What you should also do is be specific in the way you talk about the service. The next time you find yourself describing the team in policy-language or the service in capability-tiers, stop and describe it concretely instead. The named people. The thing on your mind. The pattern you are watching. The piece of equipment you are still waiting for. That is how well-led services sound when their leadership team talks about them, and the habit of describing them that way starts at the team meeting on Monday, not at the inspection.

The reason I built Verivius the way I did, with the governance meeting register and the risk register and the improvement-actions register cross-linked into the daily lifecycles, is that the trail inspectors read for on well-led should be a byproduct of running the service openly, not a separate piece of work. The minutes name decisions because the field is structured that way. The actions track back to the meeting that opened them. The risk register reviews are visible on the dashboard. The aggregate views the inspector reads for are the same views the leadership team should be reading every month.

What no platform can do for you is be curious about your own service. That part is on the people. The platform's job is to make the trail of that curiosity visible.

On the failure modes specifically: why providers fail Well-led. On Reg 17 specifically, which is the load-bearing well-led regulation: why most providers fail Reg 17, and how to evidence good governance. On the meeting cadence that makes the well-led trail readable: the governance meeting cadence that actually works. The regulation explainer: /regulations/reg-17-good-governance.