The details here are an anonymised composite drawn from typical engagements. No real names, logos, or figures from a single client are used. The pattern is one we see often as buyers weigh a Microsoft 365 Copilot rollout: pressure to move fast on artificial intelligence, a per seat add on price that adds up quickly, and very little hard evidence on who will actually use it.
Situation
The organization was a roughly 3,400 employee specialty retailer with a head office, regional teams, and a large store and distribution workforce. Most knowledge workers ran on Microsoft 365, while frontline staff sat on lighter frontline licensing. A board level push to adopt artificial intelligence landed on the technology team as a simple instruction: get Copilot in. The first plan on the table was to buy the Copilot add on for every office based user from day one.
The overspend found
When we modeled that plan, the risk was clear. Microsoft 365 Copilot is a per user per month add on layered on top of an eligible base license. Buying it for the whole office population meant a large recurring commitment before anyone knew whether the assistant would change how people worked. The retailer had also assumed the entire knowledge worker base needed it, when in truth large groups, such as store operations coordinators and parts of finance, did much of their work in line of business systems Copilot would not touch.
Source: Microsoft 365 Copilot is sold as a per user per month add on, commonly listed at thirty US dollars per user per month on an annual commitment, per Microsoft pricing as of June 2026. Pricing, eligibility, and packaging change often, so confirm current terms with Microsoft.
This is a classic new add on trap, and it sits squarely in the territory of digital workplace cost optimization: an exciting capability, a per seat price, and a default instinct to buy broad. Bought that way, Copilot would have joined the long list of paid seats that go unused, the same shelfware problem covered in the true cost of SaaS shelfware.
Approach: the retailer optimizes its Microsoft 365 Copilot rollout
We reframed the rollout as a measured pilot rather than a blanket purchase. The retailer assigned a capped pool of Copilot seats across a deliberately mixed group: marketing, merchandising, parts of HR, finance analysts, and a sample of senior managers. Each participant got a clear brief on what to try, and the team set a fixed review window before any wider buying decision.
During the pilot we tracked real signals, not enthusiasm. Which roles opened Copilot daily. Which used it for drafting, analysis, or meeting summaries that genuinely saved time. Which logged in once and never returned. That usage data, pulled the same way it is in SaaS utilisation benchmarks, separated the roles where Copilot paid back from the roles where it was a novelty.
The pilot also clarified the base license question. Because the Copilot add on rides on an eligible base plan, getting the underlying Microsoft 365 estate right mattered first. We confirmed the correct base tiers as part of broader Microsoft 365 optimization work, so the company was not stacking a premium add on onto seats that were themselves the wrong size.
Outcome
After the review window, the retailer rolled Copilot out to the roles that showed real, repeated value and held back the rest. The validated population was a fraction of the original company wide plan, which cut the first year add on commitment substantially while still putting the tool in the hands of everyone who benefited. Seats that the pilot showed sitting idle were never purchased at scale, and a small group that lost access barely noticed because they had not used it.
Just as important, the company gained a repeatable method. Rather than a one time guess, it now had an adoption gate: prove value in a measured group, then expand by demand. That discipline is the governance habit at the heart of our Microsoft 365 optimization service, and it kept the rollout honest as more teams asked for access.
Lessons for buyers
The first lesson is that artificial intelligence add ons reward patience. A per seat price that looks small per user becomes a large annual line when multiplied across an entire workforce, so the cost of buying broad before proving value is real money. The second is that Copilot value is highly role dependent, which means company wide licensing almost always funds seats that will not be used. The third is to fix the base license before adding the premium layer, so the add on lands on a right sized estate rather than compounding an existing mistake.
This case study is commercial and cost advisory, not legal advice. Contract interpretation belongs with your own counsel. Our role is to give buyers the usage evidence and right sized numbers to roll out tools like Copilot at the pace the value justifies.