By Ruben Van den Bossche, CEO at Gorilla. Gorilla won the the ‘Best SaaS Product for Energy, Utilities or Telecoms’ category at the 2026 SaaS Awards.

Earlier this year we commissioned the research house Censuswide to survey 600 senior executives from energy retailers across Belgium, France, Germany, the Netherlands, Spain and the UK. The aim was a benchmark: not how confident the sector feels about margin but how well it can actually see, understand and act on it. The results were published as the Energy Margin Intelligence Index 2026.

The headline finding was 94 per cent confidence in margin visibility. On its own, that reads like a solved problem.

The rest of the data suggests otherwise. The majority of those same executives can reconcile reported sales against actual financial margin only ‘usually, but with some gaps or delays’. Almost three quarters, 74 per cent, named siloed teams as a barrier to improving margin. Most estimate that between six per cent and ten per cent of annual revenue is at risk because of poor margin visibility and slow decision-making. And most said it takes their organisation between 13 to 18 months to move a new energy product from inception to billing at scale.

We called this the Margin Confidence Paradox. You might read this as a sign that the confidence is misplaced but look closer at the details and there are rational reasons for the paradox to exist. I think the more useful question is why this is the case.

Five Correct Answers, Five Conflicts

Ask five teams inside an energy retailer what the margin on a given contract is and you will get five defensible answers.

The pricing team’s number is right, given the cost stack and volume assumptions used at quote. The forecasting team’s number is right, the trading desk’s number is right, billing’s number is right and at the end, finance’s number is right.

Every one is correct inside its own system but none of them describes the same contract. Each team’s confidence is a true report about its own conditions, which is why the confidence survives contact with reality for so long. Nobody in that chain is wrong, they simply lack the holistic vision required.

So high confidence and weak reconciliation are not contradictory findings. They are the same fact observed from two different perspectives.

What a Decade of Software Delivered

The dominant enterprise buying pattern of the past decade was rational: pick the strongest tool for each function and integrate later. For a pricing team, a purpose-built pricing engine beats a generic module. For settlements, the same.

What that pattern did not produce was agreement, or alignment. Reporting surfaces multiplied far faster than shared definitions did. A retailer can now dashboard every one of those five numbers with a clean interface and still have no mechanism to reconcile the differences.

This is where I would push back on how the industry talks about data maturity. Volume of reporting is treated as evidence of control. In practice, more data increase the feeling of control but leaves the actual capability untouched.

Three Years to Bill an Idea

The finding that most surprised me was what we uncovered with product launches: more than three years from inception to billing at scale.

That number is not a statement about creativity or commercial ambition. It is a measurement of how many systems have to agree before a new product can exist. Price it, sell it, hedge it, bill it, forecast it, report on it. If those six things live in six places with six versions of the truth, the launch timeline is set by the slowest reconciliation.

For anyone building or buying cloud software, that is the cost of integration debt expressed in a way anyone can understand: time to revenue.

The products the energy transition needs are flexible contracts, corporate power purchase agreements, time-of-use structures and products that reward customers for shifting load. All of them are harder to price, harder to forecast and harder to bill than a fixed tariff. A sector with a three-year product cycle cannot supply them at the speed the transition requires.

AI Raises the Value of the Foundation

AI is the reason this stops being a slow-burn problem.

I want to be specific about where I think the genuine value sits because sweeping claims are not much use to anyone. Two things look real to me. The first is compression of the experience curve. In this industry, explaining a shift in non-commodity costs to a customer, or reading a renewal conversation correctly, has historically taken ten years to learn. An analyst two years in, with the right tooling, can now work closer to that standard. The second is throughput: portfolio analysis that took an analyst three weeks can be produced in seconds, across thousands of accounts rather than a sample.

Both of those depend entirely on the inputs. Point a capable model at five inconsistent versions of margin and it will not flag the inconsistency. It will produce a fluent, fast, confident answer built on whichever version it was handed and it will do it at a volume no human reviewer can spot-check. Worse, it removes the friction that used to catch errors, which was an experienced analyst frowning at a number that looked odd.

The scarce input for AI in commercial functions is not model quality. It is an agreed, auditable definition of the thing being optimised. Organisations that have this will compound their advantage quickly.

Measure Capability, Not Confidence

A better dashboard solves none of this and the fix is not complicated, though it is operationally hard.

It means one place where pricing, forecasting, hedging, billing and finance logic meet, rather than five places that each publish a view. Every figure should be traceable to source because a number a pricing manager cannot explain is a number they cannot act on, however accurate it is. And it means closing the loop: comparing realised margin against forecast margin at contract level, continuously, so that any deviation surfaces ASAP.

That last point is the one most often skipped. Plenty of organisations can tell you what a contract was expected to earn. Far fewer can tell you what it is earning now or which of the two numbers to trust.

If you lead a commercial function, there is a diagnostic worth running this quarter. Pick one contract. Ask how long it takes to produce a margin figure that pricing, billing and finance will all sign their names to. If the answer is measured in weeks, or if it requires a meeting, that is the real benchmark.