In an agentic banking market – should prices become more complex or more simple to win or hold customers?

Complex pricing – a driver for market share

We have previously summarized at length, how banking thought leaders agree that more personalized pricing – whether this comes in the form of bundles, subscription or custom pricing – is the way forward for banks to capture both, a higher share of wallet and in return market share – as well as improve overall margin.

But will these options actually work when AI agents make the pricing decision? What do pricing experts believe?

What we found is, that there is very little reliable research on the topic at this point – even when we look beyond banking related research and journey into more general pricing research. Similarly there are very few hypothesis out there from leading consultancies on the topic.

But here is what we found and what we are concluding from it.

How consumers are searching for price comparison matters

One of the most recent and most comprehensive research we could find regarding agentic search and pricing was conducted by McGill university in Canada (Shopping by algorithm: How agentic AI deploys human heuristics as a surrogate consumer – published in September2026).

The research showed that while under a specific goal prompt, acquisition costs already lead the LLMs to ommit smaller details in their search, under a vague goal prompt the acquisition costs lead the LLMs to significantly ommit pricing details in their search. In the research case it meant, that under a specific goal prompt some models already stopped considering the cent part in unit prices, falling for the old marketing cue of pricing at 4,99 instead of 5,00 USD. Under vague prompts, the models even stopped to compare unit volume information.

The conclusion is, the more vague the prompt and the more difficult it is for the AI to find the necessary comparison information, the more it tends to simplify the pricing comparison.

“Retailers determine the cost of retrieving product attributes through platform design, and consumers determine, through the specificity of their delegating prompt, which attributes the AI agent treats as diagnostic. The combination of costly retrieval and vague goal prompts lead AI agents to truncate their information search and choose suboptimally.”

Consumer search becomes more specific

A small study of consumer search in banking by Bain & Company (How AI Rewrites the Rules of Brand Discoverability in Banking – May 2026) showed that comparison prompts are becoming more complex and that consumers often start by copying there current conditions into the prompt as a starting point for comparison.

A similar behavior is also quoted by Simon-Kucher in their blog post (When ChatGPT becomes a banking advisor: Rethinking products for the AI era), citing “customers photograph the letter, attach it to a prompt, and ask the AI for an assessment of the new offer”.

And a study from Yale from March 2026 (Agentic Markets: Equilibrium Effectsof Improving Consumer Search), indicates that there is potential for businesses, including banks to set higher prices, if they can see a price search includes more complex pricing parameter.

Because the search will reduce less suitable offers and therefore reduce overall competition – speaking in favor of the margin impact of more complex pricing models like bundles or subscriptions pricing, even in agentic markets.

So what speaks for price simplification to address agentic search?

Simon-Kucher in the above named blog post advocate for product simplification, including pricing: “Simplify products and position them clearly: The clearer the pricing structures and value propositions, the more easily customers and AI systems can grasp their benefits.”

Which is in line with the findings from the Canadian research team, which showed that current AI models quickly drop even basic price parameter in their comparison when the costs of information acquisition becomes too high.

Both address the issue banks face when they are faced with the task to compete as “challenger” in an agentic price comparison run with the current LLM models. As bank you can – at the moment – simply not stand out in agentic price comparison with your full width of product arguments, you will be reduced to simple price comparisons online.

What speaks for continuing the road of price personalisation?

We have seen and continue to see rapid developments, both in search cost reduction, with costs falling 10-40x per year, as well as consumer competence when it comes to developing more precise prompts.

Also, as the consulting posts show, when it comes to comparing banking products consumers start their comparison with the offer they receive from their current bank – or the offer they first saw online, as benchmark, which mid- to long-term – as models get more able to conduct deeper comparisons, will set the benchmark for everyone competing in the comparison.

And the Yale study indicates that as LLMs become more able to conduct deeper comparisons, there will be competitive advantage in belonging to the smaller group of competitors who can offer pricing input for all the different parameter the consumer search includes.

What banks should do now

So while there is no clear evidence to favor either strategy –simplifying or further increasing complexity – the likely winning approach is a two-fold strategy:

1.) Develop pricing simplified entry level offers, which are easy to read for AI search agents to compete for new customers, while

2.) continuing on the path to develop more personalised pricing offers for existing customers to grow their share of wallet and to compete on more sophisticated pricing once the models are ready for it.

In the meantime what is clear without doubt is that “banks should already start looking at their products and communication strategies from the perspective of a customer who relies on AI” as Simon-Kucher states in the above named blog post.

‍

And they also deliver a list of steps banks should take now to achieve this:

Simplify products and position them clearly: The clearer the pricing structures and value propositions, the more easily customers and AI systems can grasp their benefits. 

‍

Provide structured product information: Many banks already offer high-quality products, but their advantages are scattered across different channels. Going forward, the key will be to provide information that is consistent, transparent, and easy to find.

 ‍

Test AI perception regularly: Banks need to analyze on an ongoing basis how AI systems describe and assess their products. Initial tests often deliver surprising insights into which strengths are recognized and which unique selling points have so far remained largely invisible.”