The 18-month playbook: from AI assessment to measurable ROI
How three TMT enterprises moved from pilot purgatory to production — and what their first 90 days actually looked like.
FIG. 01 Hudson Group architectural analysis and deployment trajectory.
Most enterprise AI programs do not fail because the technology is wrong. They fail because the organization never defined what success would look like, never instrumented for it, and never resourced the unglamorous middle — the twelve months between a promising pilot and a system the business actually depends on. We call that gap pilot purgatory, and over the past two years we have watched three TMT enterprises climb out of it on roughly the same timeline.
This is the playbook those three deployments share. It is not a maturity model or a vendor checklist. It is a sequence — assessment, architecture, deployment, and compounding — mapped to an eighteen-month horizon, with the milestones that actually predicted whether a program would reach payback.
PHASE 01 — MONTHS 0-3
Assessment is a measurement exercise, not a workshop
The instinct on day one is to inventory use cases. That is the wrong first move. The enterprises that reached ROI started by instrumenting their current state — the cost-per-task, cycle time, and error rate of the workflows they intended to change — before a single model was selected. Without that baseline, every downstream claim of impact is an assertion rather than a number.
In practice, the first ninety days produced three artifacts: a ranked portfolio of candidate workflows scored on value and feasibility, a measured baseline for the top five, and an executive sponsor with a named business owner attached to each. The programs that skipped the baseline spent month fourteen arguing about whether anything had improved.
PHASE 04 — MONTHS 12-18
Compounding is the whole point
The first workflow rarely pays for the program. The second one — built on the same integration backbone, the same governance, the same trained team — pays for it twice over because the marginal cost of the next deployment collapses. One European telco reached a 312% cumulative ROI not from a single breakthrough agent but from the fourth, fifth, and sixth use cases riding infrastructure the first three had already paid for.
That is the case for the eighteen-month horizon. Anything shorter measures the first workflow in isolation and misses the compounding that justifies the investment in the first place.
Marcus leads enterprise assessment and roadmap engagements at Hudson Group, with a focus on regulated TMT organizations moving from pilot to production. He has overseen deployments across Switzerland, Poland, and the wider EU.