Enabling a state insurer's flagship data transformation

Workers’ compensation looks healthier than it is. Across the states covered by NCCI, the industry’s main data and rating bureau, the calendar year 2025 combined ratio came in at 91, the twelfth consecutive year below 100. The accident year figure was 102. The difference is prior year reserve releases, and that cushion is thinning: NCCI put reserve redundancy at $14 billion at the end of 2025 against $16 billion a year earlier, with medical and indemnity severity each up about 4 percent.

For a public or quasi public carrier the squeeze is sharper. A rating agency’s aggregate for competitive state funds, on 2023 data, put their combined ratio after policyholder dividends at 105.8 against 88.7 for the workers’ compensation line overall. That gap is a giveback policy rather than an underwriting failure, and it is the point: money a commercial carrier keeps as headroom, a state fund returns to policyholders. Less headroom means less room for operational error, and the error is public record.

Third party sector figures describing the market this client operates in. Neither is this client’s result, and no equivalent measurement exists for this engagement.

Medical cost, the largest single lever, is set mostly by fee schedule design in the jurisdiction. The levers that remain are internal ones: how fast the organisation decides, and whether it can tell what its own programmes achieved. Which is why programmes like this one get large, get consultancy heavy, and get slow.

The figures above come from NCCI, Washington State legislative audit reporting and trade press covering the sector. They are third party research describing the market this client operates in, not outcomes of this engagement.

A large consultancy already owned delivery of this programme. That was settled before we arrived and it was not going to change.

The insurer’s programme manager had a narrower problem, and she had diagnosed it herself. Her teams did not set goals or milestones on a quarterly basis. There was no point in the calendar where anyone asked whether the quarter’s objectives had been met. Work was planned and work was delivered, but the quarter never closed on anything.

She had also worked out why quarterly planning kept failing to stick. Part of it was capacity and part of it was that people could not yet see what it was for. Those two fail differently. Unprotected days get taken by whatever is on fire that week, and protected days do nothing for people who do not believe the exercise pays back.

Her own written assessment of the programme’s maturity on agile practice was blunt: not mature, with no formal attempt she was aware of to change that. That is rare in writing from a public body, and it made scoping short.

She found us the way most good engagements start. She took a public class in November, then came back with a real problem.

Built the agenda from her briefing, not from a catalogue. She briefed us across a few sessions before any agenda existed. We turned that into a two day Strategic Leadership Workshop for the data and analytics transformation programme, delivered to her leadership group in late April.

A public body’s leadership group has usually sat through several framework introductions. Run the catalogue version and the room evaluates the content instead of its own decisions: they rate the trainer and change nothing on their own calendar. She had also worked out, before we did, that the argument needed to come from someone other than their programme manager. Anyone making the same case internally for a year has already spent the authority it takes to make it again.

Leadership first, certification second. This is the part most often reversed. Certify the delivery teams first and you get people fluent in a cadence their leadership does not recognise and will interrupt at the first schedule conflict. Those teams conclude the method does not work here, the most expensive lesson available and the hardest to undo. Aligning leadership first settles who owns the quarterly question and whether the answer will ever be asked for.

Closed the loop rather than leaving on a high. A structured debrief in early May, roughly two weeks after the workshop. Debrief on the day and you capture sentiment. Debrief a quarter later and the decisions have merged into whatever was going to happen anyway. Two weeks is long enough to have tested the ideas against a real working week, short enough that the reasoning is still recoverable.

Certification in two waves, not one. Two separate two day classes in mid and late June rather than one large session. Splitting them kept the programme operating while part of the group was in class, which for a carrier with statutory claim handling deadlines is not a preference. Run it as one block and either operations pulls people back out of the room, or the operation absorbs a service gap in a week when the clock does not stop.

Then the layer above. By late July we were in an executive working session with four senior stakeholders, with a written follow up. In a public body, alignment that exists only in a room does not survive a reorganisation or a change of sponsor. A document circulates to the people who were not there, and it is what the next conversation starts from.

Working alongside a prime who owns delivery is a constraint most firms either refuse or quietly ignore. Refusing means no engagement at all. Ignoring it means duplicating the prime’s work, confusing the client’s teams about who owns what, and being removed inside a quarter. The ignoring version rarely looks like a land grab. It looks like helpfulness: a view on the delivery plan nobody asked for. The client’s people then get two answers to one question, and the cost of reconciling them lands on the client.

We took the bounded version deliberately. Our scope was the enablement layer: getting the leadership group aligned on why a planning cadence exists and what it asks of them personally. Programme delivery stayed where it was, and we said so in the room, so the client’s teams knew which questions to bring to whom.

That is not a modest position to hold. The enablement layer is the part a delivery prime is least able to do, because doing it well means telling the client’s leadership something they may not want to hear. Holding the line is why this programme is still running and the prime relationship is untouched. In multi vendor environments, the fastest way to lose an account is to make the client’s other supplier your problem.

If nobody asks the closing question, you do not have a quarterly cadence. You have quarterly documents. The cheapest first move costs nothing: put a date at the end of the quarter where someone senior asks whether the objectives were met, and make the answer visible internally. Do it before buying planning tooling or restructuring anything. Within one cycle it tells you whether you have a discipline problem or a capacity problem, and those need opposite responses.

Sequence leadership before teams. It is the reversal that costs the most, and it happens constantly, because delivery teams are easier to get into a room than executives are.

Do not let a platform programme become the reason nothing improves. The public record here is full of capability sequenced behind a core system replacement and then left waiting years for it. Planning cadence and review discipline are not gated on new technology. They can be improved on whatever is running today, and the improvement carries across when the new system lands.

The engagement has widened without being sold. Nine months on from that first public class, the group we work with has grown from a single programme manager to four senior stakeholders, and the conversation has moved from one workshop to a standing advisory arrangement under discussion. It opened because the people in the first room brought the people in the second.

For a public body with a formal procurement function, work that clears review without argument and comes back for more is the practical measure of whether it landed.