As communications consultants, we typically need to pitch to a marketing/executive team before we work on the business. And I’ll let you in on a little secret from those pitches: we often talk about using data to boost a fintech company’s profile. A joke around the office is that when in doubt, propose an index.

There’s a reason why we so often go back to this well – it’s incredibly effective. There’s a limited number of times that anyone will be interested in introducing a company or writing about a corporate narrative. Mergers, acquisitions and expansions are noteworthy but hardly a reliable occurrence. Data is the key to regular observations.

We pitch in poetry and have relationships in prose. Too often what is a great idea is something that never quite gets implemented. Like too many things on a project plan, it gets relegated to “someday”.

I am beginning to think we’ve collectively given up too easily. A combination of technology, new approaches and data sharing means that’s a path for every single fintech to have interesting and unique data to share with the media.

Here, I divide companies into four paths and show how they all can find a way towards interesting insights with numbers.

Scenario 1: You have a consumer-facing data product

In some ways, this is the most straightforward scenario. There’s already some sort of metric or insight that your business sells in a direct-to-consumer portal, so nothing needs to be developed. The primary concern is giving away valuable data for free. If your business is predicated on data sales, it can be a tricky proposition to open access. Plaid hits the right balance: free high-level insights for everyone while monetizing deeper analytics.

Start with your sales material – there may already be a process wherein people regularly take fresh data to share externally. If that’s not the case, consider asking for access to the tool. Play around and see what you find notable.

Logins are trickier. While journalists will want to rummage around in the system themselves, you run the risk that they find your platform too confusing or cumbersome to use. When you hook someone, it can be a ticket for regular coverage for a very long time. But mostly, you’ll still need to send insights, searches, and prompts to help with storytelling.

Scenario 2: You have an institutional product that’s not easily understood

I cannot tell you how many times I’ve been speaking to an executive where they tell me about how game-changing their software product is. “It’s nothing like what’s come before.” So I arrange a demo, and well, it’s a variety of small data entry feels on a black background. Groundbreaking. The insight derived from these terminals might truly unlock millions in value, but directly reporting those metrics will lead to confusion.

Blame the quants or blame the journalists for not being quants. For these companies, I recommend finding a way to take one step backwards. Excess kurtosis, convexity bias, Monte Carlo path dependence and the mean-reversion threshold frankly will mean nothing to the vast majority of journalists – but if you show them why a certain sector might feel a bit toppy, well, that’s news.

Don’t worry if what you wind up releasing might feel a bit simplistic to end users – sales can show them the true depth of your product.

Scenario 3: All of your work is for private, third-party

Many fintechs fundamentally work as consultants, creating custom products for each client. That means any data is proprietary and unable to be shared. And data protection means that you can’t easily combine and create insights across your data.

Work outputs and trends are your friend here. The same things that might wind up in an executive report or investor deck might be relevant to the media. “We’ve taken the average employee hours in creating a report down from 16 hours to 4 hours,” might actually be something. It’s especially true if this is an industry benchmark that’s already understood.

The AI revolution brings plenty of opportunity. Even the word “revolution” feels mysterious. I still commute into the office three days a week and work several hours at a desk. How exactly has AI changed what we are doing? Find ways to make approximations more specific. Instead of saying “we’re processing exceptions much faster before”, try to find a way to say, “the average compliance employee in 2025 is doing the equivalent of 4.3 FTE in 2021”. That’s something that will go in a story.

Scenario 4: There really is no data

Sometimes, it’s just an empty well. For whatever reason, you ultimately don’t have any quantitative insights from your work.

Here’s where I think you can get a bit more creative. Traditionally, this is where a research or survey agency can help. These are experts at places like the FT-owned Longitude who will design, commission and provide analysis for a particular audience. They are excellent at what they do – there’s a reason why the same set of providers are brought back year after year for major consultancies and professional services firms.

But the price point can be a barrier for even the largest fintechs. No matter what the somewhat deceptive initial prices say, it’s challenging to do this at scale for less than six figures. Reaching business leaders is expensive – how many surveys do you want to fill out – and so is compiling the data. Another option is tagging along to a much larger or ‘omnibus’ survey – using firms like Savanta or Opinium – but you get limited questions and typically a very broad consumer profile. Or consider hiring a media outlet to use its database and editorial staff to create a bespoke profile, which can mean great coverage, but just in a single outlet.

A whole group of innovative disruptors have entered this market, claiming to offer cheaper, quicker solutions. These are worth exploring, but with a partnership with a professional communications firm. We have a better idea of what will fly with journalists – individual reporters will have standards about sample size, geographic distribution and any preference on the types of questions.

One tip is to ask reporters what they want to see. If they feel they have contributed to the commissioning of the report, they are more likely to write on the outcome.

Large language models also make it possible to experiment with finding noticeable trends inside unsorted data. This is the work of many fintech companies but I find it underutilised as a way towards finding data. Frankly, start by creating private environments and asking a series of models what they find noteworthy or interesting. Try a prompt like: “What do you see here that isn’t broadly found elsewhere on the internet?” or “Are there any things here that answer questions that people in the field have asked?”. Or just get straight to the point, “What in here could generate X reporter and Y publication to write a story? What else would be needed?”

Finding data is a bespoke and highly iterative process. You can’t expect to come up with the next Hamburger Index on the back of a 30-minute brainstorm. But treating your data programme like any other technology – with great potential, worthy of ongoing refinement – can unlock rewards.