Every financial decision comes down to trust. A consumer trusts a comparison site's price. A provider trusts a research platform's data. A regulator trusts a firm's own account of how it treats customers. That question, of who or what lies behind an answer and whether it can be believed, is old, but it has rarely been harder to answer than it is today.
Consumers are navigating more information, from more sources, than ever before, and a good deal of it is contradictory or simply wrong. Ask an AI tool to explain your insurance policy and there's a reasonable chance it will get a detail wrong, not through deception, but because it's predicting the answer that sounds most plausible rather than checking the one that's true.
That's a vivid new example, but it sits alongside a longer-running problem: even when the source is a person rather than an algorithm, a comparison site, a review or a rating, working out which version of the truth to trust has got harder, not easier.
But there is a solution, and it is the same approach that sits behind every Defaqto rating and dataset. It is ensuring three critical qualities: independence, comprehensiveness and provenance.
When ratings and datasets deliver across all three:
- Providers can make and defend better product/pricing decisions because the ratings underpinning them are independently earned. They are not bought. And they are not marketing spin.
- Regulators can apply a dataset to a compliance judgement
- Comparison sites can use the data to justify a published price.
- Consumers can access a source of truth that hasn't been shaped by who's paying for the answer.
It’s about understanding what independence means and how it is applied.
Independence starts with who pays whom
The simplest test of independence is a financial one: does the rated business pay the rater, and does that payment change the outcome? Defaqto doesn't take commission for the ratings it awards. A Star Rating, or a product's position in Defaqto's wider data, reflects what is in that product, not who is paying for a badge or licence.
Providers can license the Star Rating badge for marketing use once they've earned it, but the earning happens first, independently of any commercial relationship, and no provider can buy a better outcome.
There's a second, quieter way independence gets compromised, and it has nothing to do with who pays whom. Some organisations that publish product or service reviews in financial services are built around a consumer campaigning or advocacy mission, existing to fight for better outcomes on behalf of consumers. That's a legitimate role, but it isn't the same thing as independence.
An organisation whose purpose includes taking a side, even the side of the customer, starts from a position rather than from the evidence. Defaqto's ratings and data aren't built to advocate for consumers, providers or anyone else. They're built to describe what a product contains and they let providers, regulators and consumers each draw their own conclusions. Evidence that starts from a predetermined conclusion, however well-intentioned, is a different kind of output to evidence that is entirely neutral.
Trust needs comprehensiveness, not a sample
A rating built on a self-selected sample of reviews, or a shortlist of products a business has chosen to submit, tells you something, but not what you might assume. Defaqto's data spans three markets (general insurance, banking, and wealth and protection) across 127 product types, and every product in a given type is assessed against the same published criteria, whether the provider wants it rated or not.
That consistency lets a regulator, a comparison site or a compliance team treat a rating or a data point as genuinely comparable across the whole market, rather than just the businesses that opted in.
Provenance is what makes trust auditable
Comprehensiveness and independence are still not enough if nobody can check the working. Every fact in Defaqto's database is tied to a specific provider, product, data table and timestamp, so any figure can be traced back to its source and to the point at which it changed. That matters to a compliance team building a Consumer Duty fair value assessment, and it matters just as much for the newest use case: grounding AI systems in facts rather than plausible guesses.
Matrix 360 is a useful example of provenance at work. It turns these same datasets into dashboards that a product team can use to see where their range sits ahead of a renewal cycle, a compliance team can cite in a Consumer Duty fair value assessment, and an executive can use for board-ready reporting on market position, all refreshed four times a day with alerts when a competitor changes a price or feature.
More than 85% of UK general insurers now use it for one or more of those jobs, and because every dashboard traces back to source, the same data that powers a strategy meeting can also stand up to a regulator's question.
This provenance makes Defaqto's underlying data usable further afield: it's used directly by the FCA and by the price comparison websites that need to stand behind the numbers they publish. As AI tools increasingly query that data at runtime, through APIs and Model Context Protocol endpoints built for exactly that purpose, provenance becomes the thing standing between a customer and a confidently wrong answer.
Why the discipline has to hold, especially when nobody's checking
None of this is difficult to claim. Most organisations that publish data or ratings would say they are independent, comprehensive and trustworthy, whether or not that’s demonstrably true. What's harder, and what earns trust over time, is building a business model, a data structure and an audit trail that means those claims can be checked rather than simply asserted.
It is that discipline in Defaqto data that lets providers, regulators, aggregators and, increasingly, the systems acting on their behalf, rely on it for decision making, including the ones nobody is watching in the moment they are made.
As those decisions move further from a human reading a policy document and closer to an algorithm summarising one, independence, comprehensiveness and provenance become ever more vital. They are the only real guarantee that the answer a customer, a provider or a regulator receives is the true one, not just the most convenient or the one that sounds the most confident.

