The Three Vendors That Kill Transformation Programmes
Most AI Transformation programmes don't fail because the technology was wrong. They fail because the structure was wrong. Three separate vendors, three separate contracts, three sets of incentives — and a single business outcome that no one signed up to be responsible for.
If you have ever run a transformation programme that produced a brilliant deck, partial software, and unclear results, the reason is probably not your team. It is probably this structure.
Vendor one: the strategy firm
A management consulting firm runs the diagnostic. The deliverable is a strategy deck — frequently excellent, frequently expensive, frequently $500K to $2M+ for a 3-6 month engagement. The slides are sharp, the opportunity sizing is compelling, and the roadmap is logical.
The deck is also the end of the engagement. The firm that wrote it doesn't build software. When the deck is delivered, the firm exits — with reputation intact, regardless of whether anything ever gets built.
Vendor two: the engineering firm
An IT services firm gets the deck and starts scoping the build. Skilled engineers at $200/hour. They build what the deck told them to build — but they don't tell you what to build, and they don't measure whether what was built worked.
By the time the build is complete, the strategy deck is six months old. The market has moved. The competitor has launched something different. The original opportunity has narrowed. The software arrives — for the problem that was true when the deck was written.
When the build wraps, the firm exits. Their incentive is billable hours, not business outcome.
Vendor three: the change firm
By now, the software is built but adoption is poor. A change management firm comes in. They run workshops, build adoption plans, deploy ADKAR frameworks. They are skilled. The work they do is real.
It is also too late. Adoption-friendly UX, change-aware specifications, training-driven onboarding — these all needed to be designed in at vendor one's stage. By vendor three's stage, they are bolted on. The software gets adopted partially, the metric moves marginally, and the programme is declared "completed" — though no one quite says successful.
No one is accountable for the outcome
The structural problem is at the seams. The strategy firm is responsible for the deck. The engineering firm is responsible for the software. The change firm is responsible for the workshops. Nobody is responsible for the business metric that was supposed to move.
This isn't a bug in any one vendor. It's a feature of the structure. Three vendors, three contracts, three exit points. The thing that was supposed to be transformed lives in the gaps between them.
The fix: one closed loop, one accountable system
The structural fix is to collapse the three contracts into one. Continuous intelligence (replacing the strategy firm's episodic deck), AI-augmented software delivery (replacing the engineering firm's billable-hour model), and embedded change governance (replacing the change firm's bolt-on workshops) — all under one contract, one team, one outcome metric.
For mid-market companies — $20M to $200M in revenue — this isn't optional. The math of three separate vendors doesn't work at mid-market scale. The first contract alone consumes the budget that should have funded the whole transformation.
For enterprises, the three-vendor structure is expensive but survivable. For mid-market, it is fatal.
What to look for instead
When evaluating vendors for AI Transformation, the structural questions matter as much as the technical ones:
Does the same team that diagnoses the opportunity also build the software? If not, you have at least two vendors and a handoff.
Is change management embedded in the specification, or bolted on after launch? If after, expect adoption issues.
Is the business metric named in the contract, with the vendor accountable for moving it? If not, you are paying for activity, not outcome.
Does intelligence continue after the build, or does the engagement end at delivery? If it ends, drift is guaranteed.
If the answer to all four is "yes, in one contract, with one team," you have an AI Operating System. If not, you have three vendors arranged around a wishful outcome.