I have a very old friend who’s a journalist. He’s actually a really nice chap, and he’s full of fascinating stories. A while back we were talking about the selling we do in our jobs. I was intrigued to learn how he has to compete for space in his own newspaper.

“Any old story can make it onto the Internet,” he explained, “but if you want to get your piece onto actual newsprint, you have to sell the story to your section editor.” To make the front page, the section editor sells the story (now styled as his story) up to the news editor, who then sells it to the managing editor. He decides what makes it onto the front page, but it’s the Editor in Chief who decides what fills in the coveted ‘over-the-fold’ position.

“So I have to make my story sound sexy,” my reporter friend told me. One of the ways he does that, he admitted, is by playing around with the numbers.

“I came across this paper in a medical research journal where scientists found that a certain food additive was correlated to an increase in pancreatic cancer,” he said. “Specifically, the science bods said the incidence of cancer in people who ate this stuff was one in three million, compared to one in nine million for those who didn’t.”

I shrugged. He continued. “When you’re selling a story in the newsroom, we report that as ‘Additive Triples Cancer Risk’. It’s true, right?”

Setting aside questions of statistical significance, I confessed that I understood the reasoning, but didn’t see how anyone could benefit from dressing up the numbers that way.

Something similar occurs in the pricing of technology and data for the insurance market, and it shocks me. In short, it’s incredibly common for vendors to charge their insurance customers a percentage of gross written premium. I have seen this from both sides, and it makes no sense, neither for the tech supplier, nor for the client.

Consider, from the supplier’s point of view, a single MGA with ten users writing 10,000 policies at an average premium of 1,000. That client will cost the tech provider dramatically more in support and hosting than a different MGA that writes 100 policies at an average premium of 100,000. Yet both clients’ total GWP, and therefore the fees they pay, will be the same under a GWP-linked pricing structure.

On top of that, this model creates a certainty problem for tech providers. I’ve managed enough underwriting budgets over the years to know that the way they look on Day One (or Day -90) is always dramatically different than they look at year end. Add in the sometimes-dramatic oscillation of rates, and it’s near impossible for a tech provider to ensure they will receive sufficient income under a GWP-linked model. Under a model that reflects usage, rather than premium, they can be sure of it.

Finally, a lawyer warned me that when a tech firm takes a percentage of GWP, the regulator may be inclined to view them as an intermediary. It’s great to be noticed, but no one wants that kind of attention.

Like any product-and-service combination, technology should be priced to cover its fixed and variable costs. A proportion of the sunk development costs plus the ongoing support, hosting and other underlying services, and all the other usual overheads, should be added up to create a fair and sustainable price. If necessary, add a variable cost which fluctuates with usage (of support, for example).

Pricing data is more difficult. Purchase decisions are made typically only after the client sees that the price has dipped below the flex point where buying is a no-brainer because the knowledge supplied is certain to boost the bottom line.

From the vendor’s perspective, it makes sense to price as near to that point as possible (that is, as low as costs and desired profit allow). The serious cost to the vendor lies in the collection, cleaning, and compilation of the data, not in its delivery. Low inception pricing maximises the number of purchases, and even if you’re collecting from a satellite, once the major fixed costs are covered, it’s almost all profit.

Despite this, a large number of data providers – maybe the majority – charge very much more than my pricing model proposes. Sometimes their models are based on GWP, leaving vendors to wonder why no one wants to buy their business-changing intelligence.

No doubt one of the key reasons tech companies and data providers charge fees as a percentage of GWP is that underwriters are used to buying based on this metric. Commission, after all, is the biggest line item in the expense ratio.

That doesn’t mean it makes sense for  them, either, for the same reasons, plus one more. Established tech providers understand the fluctuations inherent in GWP (and in insurers’ changing fortunes), and they price for the lowest experience. That means almost all the time, buyers of insurance tech and data are overpaying under the GWP model.

The headline percentage might look good at the outset, but the real story is bad news all around.