The era of Personal Finance Management or “PFM” is over!

a.k.a. “It's about Context, Stupid!”

Despite big investments in time, energy and money, Personal Finance Management has failed to engage customers at most traditional retail banks.

Just look at the numbers. On average, only 12% of a typical bank’s customer base actually use any PFM features such as pie charts, graphs and planning guides. And in private conversations, banks admit this percentage is inflated: an average of only 1 in 20 users, or 5%, engage with any PFM features on a frequent basis.

Why are consumers so disengaged? After all, money matters. Financial problems, according to Forbes Advisor, are one of the top 5 reasons people divorce. And when looking at other, more positive, life events such as marriage, birth of a child, buying a home, relocating for a new job, etc., financial services play a critical role. Given that around 60% of European consumers still manually manage their money, and roughly half of them live paycheck to paycheck, clearly there is a big, unmet need in the retail banking market.

So what’s wrong with PFM? At Snowdrop, we believe the problem lies in the name: Personal.

Card payments and transactions are already personal. The problem is people can’t readily understand them. This lack of clarity creates frustration, disengagement and even anxiety for consumers. This is the first thing that must be fixed - create clarity with highly accurate data enrichment so people can visualize and understand their spending habits. Without clarity, there is no foundation of trust between the consumer and the bank.

Once clarity is established, banks should recognize that people often need assistance within a specific context. How much should one spend on their housing relative to their disposable income? How much should they save per month to plan a holiday? For a wedding? To buy a home? For their retirement? Helping someone pull together different elements across a wide range of products, services and disparate sets of data and information sources is difficult. This is the big unmet need: understand the context and related choices - what to do in any given moment and place?

How do other sectors tackle this? Search provides a convenient example. When a consumer searches for “money” on their laptop while at home, Google generates one set of results. When that same user searches for “money” on their mobile phone whilst traveling abroad, Google generates a different set of results. Why? Because Google contextualizes a wide range of datapoints to help the search engine better understand the user’s intent behind the search term. It’s not about more personalisation, but rather contextualization - identifying the user “signals” so the system can ascertain what the user’s intent is.

However, a lack of contextualization isn’t PFM’s only problem; many PFM features are not delivered in an intuitive way. As stated earlier, poor adoption numbers show that the vast majority of people don’t enjoy deciphering pie charts, graphs and calendar planning. Instead, people seek simple summaries, clear visuals prompts, or for more complex things, listen in to a plain spoken conversation.

Take podcasts for example. According to the UK agency Ofcom, just over 20% of UK adults now listen to at least one podcast a week. That’s 11.7 million people in the UK alone. Furthermore, almost three quarters say they listen to between one and five podcasts each week; not many pie charts here.

How can banks deliver intuitive, engaging services that really help people understand their shifting financial needs - especially for key life events? We believe this is where multi-agentic generative AI plays a role.

Today, consumer AI services like OpenAI’s ChatGPT, Google’s Gemini and others don’t just deliver different results based on a user’s shifting context. Instead, they take this further by “parrotting” what is the most likely response based on a vast number of similar questions / responses pairings. For example, a user can ask “what is the best way to get a mortgage?” and a generative AI tool will summarize information across various trusted sources in a coherent, concise manner in the media format you want rather than just display a bunch of text and hyperlinks.

More importantly, maturing AI tools are also trained to minimize mistakes by “grounding” their datasets and showing the underlying “thinking” in a number of ways. Why is this important? First of all to avoid producing mistakes - so-called “hallucinations”.

Also, as agentic AI rapidly gains acceptance, such agents will increasingly supplant human expertise and insights in terms of ability, credibility and above all efficiency. For example, a well-trained AI agent can process 95% of a complex S1 form used by Wall Street investment banks in just a few minutes. Previously, the same form would take a team of six experts two full weeks to complete. Given the power of AI, this means mistakes and underlying biases may also scale quickly if not carefully checked. It also means that human intelligence still plays a critical role to fully understand a consumer’s context.

At Snowdrop, we have built an AI tool that is grounded largely in reality - meaning we leverage the power of Google Search plus Google Maps to ensure that our generated AI outputs are based on real merchants, homes, places, routes and locations. Why does this matter? When someone wants clear insights about complex, inter-related often opaque areas (eg. “what do I need to do to afford this home?”), our AI platform provides clear, contextual multi-modal grounded information in an intuitive way to consumers in their native language.

In short, PFM doesn't need more personalisation. Instead, banks should implement tools to intuitively engage people 24/7 based on their specific contextual needs, in a language and media format that suits them best in that moment. This approach will generate customer trust, new revenue streams and long-term value for banks.