Rebuilding the Ops model around AI.
The evidence base for a build, partner or outsource decision across customer operations.
- Services
- Operating model
- Period
- 2026
- Scope
- Customer operations for a retailer serving more than a million households
A fifteen-year arrangement nobody had tested
The retailer had run a large part of its customer operations through an offshore partner for about fifteen years. The arrangement had never been taken to market. It had also drifted: over time, complex and sensitive work came back onshore, which left the offshore team with the simpler tasks while the retailer kept most of the management burden and the fixed cost. The review's description was that the model had evolved by default, not design
.
Leadership had hourly rates and not much else. Unit costs, service levels, tenure and turnover were not visible in a form anyone could argue from. Two things then forced the question. The contract had to be decided within the year. And a core platform replacement, already under way, was expected to automate a good share of exactly the work that had been sent offshore.
The review needed to be independent, because fifteen years produces alliances, and it needed to be discreet.
Build the model, then let the executives change the assumptions
The work ran in three parts: current state, a market and horizon scan, and target state recommendations. Around fifteen senior people were interviewed, from the chief executive and the group owner through the general managers running sales, transformation, risk and customer insight.
The cost model was built alongside the retailer's own commercial manager, from its own baseline. Four options were framed — continue as is, bring back sales, bring back sales and voice, bring everything back — and each was expressed as a net annual opportunity cost rather than as a cost increase, because that is what the choice actually is.
Rather than defend a set of numbers in a slide, we handed the model over. Executives set their own assumptions for automation, work complexity and how much came back onshore, and watched the answer move. The argument moved with it, from the consultant to the assumptions.
Instruments, not only documents
Three reports, an executive summary written as a board decision paper, and two tools the client can use without us in the room.
- Survey
- A target operating model survey on a platform we host ourselves, asking each leader where they expected the business to be in one to two years.
- Modeller
- An interactive scenario tool seeded with the client's own baseline, with three presets and four groups of assumptions the user controls.
- Board paper
- An executive summary and decision paper written the way a board reads: the options, the sequencing, the risks and what has to be decided when.
- Handover
- Editable versions of every deliverable, handed to the client so the work continues without us.
A decision the board can take on evidence
Modelled reduction in total customer-operations workload over the next two years, from the platform replacement and from AI. This is a modelled figure, not an achieved one. It sizes the decision; it does not report a result.
From signed statement of work to a board-ready executive summary and decision paper.
The working group settled on a common position: no scenario supports continuing the current structure unchanged.
Bringing sales work back onshore was treated as close to settled, and planning for it began before the report was final.
A sequenced recommendation — a short extension at renewal, then a progressive wind-down as the new platform lands — became the working position going into the board.
The board considers the paper at the end of the year. No decision has been taken and nothing has been implemented, so there is no outcome to report yet.
Stratum were instrumental in helping us understand where we wanted to get to, what it would require, and what we needed to build to get there.
Have a problem shaped like this one?
Fixed-scope discovery to start.
Strata: 高 长华 — Unsplash