Surviving Procurement: Why Utility Analytics Projects Die After the Pilot

The median large IT project runs 45 percent over budget and delivers 56 percent less value than its business case promised. Roughly 70 percent of digital transformations fail to meet their goals, and by Gartner's count around 85 percent of analytics pilots never scale into production. Utilities know these numbers, and increasingly they procure as if they expect to become one of them.

That defensiveness shows up in tender design: exhaustive specifications, heavy liability clauses, payment schedules back-loaded to the point where vendors price in the financing cost. The irony is that this armor does little to prevent failure. The projects still die, just more slowly and with better documentation. Having sat on the advisory side of utility ESG and analytics tenders, I would argue the failures trace to three design choices made before any vendor is selected, and all three are fixable inside standard procurement rules.

Failure is designed in before the tender closes

The first choice is specifying solutions before validating use cases. Tenders routinely ask for platforms, dashboards, and data lakes, defined in technical detail, without evidence that any specific decision inside the utility would change if the tool existed. An analytics use case is only real when someone can name the decision it improves, the data it needs, and the person who will act on the output. Use cases that pass that test before procurement are rare, which is why so many delivered platforms become expensive reporting furniture.

The second choice is buying effort instead of outcomes. Time-and-materials contracts, or fixed-price contracts milestoned on document delivery, pay vendors for activity. Nobody in the structure is paid for the utility actually using the analytics. The incentive gap widens every month after kickoff.

The third choice is scaling before proving. Enterprise-wide scope in a single award maximizes procurement efficiency and failure risk simultaneously. The 85 percent of pilots that never scale are matched by a quieter statistic: full-scale programs that should have been pilots and consumed ten times the budget learning the same lesson.

ESG analytics compounds all three problems. The use cases are newer, so fewer have been validated anywhere. The data is scattered across systems that were never designed to talk to each other, from emissions inventories to procurement records to grid operations. And the regulatory target moves: disclosure standards, taxonomy definitions, and carbon pricing rules change faster than a two-year implementation schedule. A utility that buys a monolithic ESG platform against today's reporting requirements is specifying against a spec that will be obsolete before go-live. The case for narrow, validated, adoptable use cases is therefore strongest precisely where tenders are currently broadest.

The structures that survive

None of this is an argument against ambition. It is an argument about sequencing: ambition expressed as a portfolio of proven use cases compounds, while ambition expressed as a single monolithic award mostly does not.

The tenders that produce working analytics share a recognizable architecture. They procure a small number of validated use cases, not a platform. Each use case enters the contract with a named business owner, a defined decision it supports, and a baseline against which improvement is measured. The platform, if one is needed, is justified by the use cases rather than the other way around.

Payments are milestoned on adoption, not documents. A milestone reads: the dispatch team uses the congestion forecast in daily operations for eight consecutive weeks. That is verifiable, it is what the utility actually wants, and it aligns the vendor's cash flow with the utility's outcome. Vendors who resist adoption milestones are telling you something useful about their own confidence.

And scope is released in tranches. The first award covers two or three use cases and the minimum infrastructure to run them. Subsequent tranches are pre-negotiated options, exercised on evidence. This preserves competitive tension through delivery, keeps the exit cheap if the work disappoints, and, counterintuitively, attracts better vendors: firms confident in their delivery prefer contracts where performance unlocks growth.

What this means for decision-makers

For utility executives, the highest-return work happens before procurement: a use-case validation sprint of a few weeks, run with the business owners who will live with the results, filters out most future failures at almost no cost. Nothing in public procurement law prevents it.

For procurement teams, adoption milestones and tranche options are compatible with standard tender frameworks, including in regulated and state-owned environments. What they require is an evaluation approach that scores delivery credibility and reference adoption rates as heavily as price.

For vendors and advisors, the discipline cuts both ways. Accepting adoption-linked payment terms is the most credible signal available in a market where the base rate of failure is public knowledge. In a sector where 85 percent of pilots stall, being demonstrably in the other 15 percent is the entire commercial argument.

---

Catalyze Transition advises governments, development institutions, and corporates on the economics and market design of the Asia energy transition.