InMarketing Today

Atom Bank's sale process is on the brink of collapse. Shareholders including BBVA and Toscafund targeted a £600mn exit, but only one bid materialised — from private equity group Pollen Street Capital, below the asking price and unlikely to rise. Yorkshire Building Society, Leeds Building Society, and several unnamed banks ran the rule over Atom and walked away. The bank has 250,000 customers, £4.2bn in mortgage balances, and £7.5bn in deposits. The substance is there. The price is not.

The brand gap is striking. Atom was among the UK's first digital banks, licensed in 2015 — ahead of Revolut, which took a decade to secure the same accreditation. Yet Revolut now has 70 million customers; Monzo has 15 million; Atom has 250,000. That is not a product failure alone. It is a brand failure: a company that built a credible banking operation but never generated the customer magnetism that commands a premium valuation. In M&A, buyers price future growth potential. Without a brand story that signals momentum, the numbers on the balance sheet rarely justify the ambition of the asking price.

The lesson for digital banking and fintech brands is direct: enterprise value is not just a financial metric — it reflects how convincingly you can articulate where you are going, not merely where you have been. Atom's valuation has slid from £459mn in 2022 to £350mn at its last funding round, and lower still in today's failed process. The time to close the brand-to-valuation gap is years before a sale process opens — not during one.

Crédit Agricole, Mastercard, and Worldline have completed the first agentic payment transaction in live production in France. A customer used a digital agent to search for festivals within defined parameters — budget, event type, location — received a shortlist, confirmed a choice, and instructed the AI to initiate the purchase. Crédit Agricole retained its authentication and authorisation role throughout, with specific identifiers ensuring the transaction remained secure and traceable.

The significance here is not the technology itself — agentic commerce has been discussed for months — but the milestone of "in production." This is no longer a proof of concept. When AI agents make purchasing decisions on behalf of consumers, the discovery and selection process shifts from person to machine. Brands that are not legible to an AI agent — clearly defined, attributable, supported by evidence — will be invisible in that process. Crédit Agricole's transaction is an early but real signal that AI-mediated commerce in financial services has begun.

Financial services and fintech brands should treat this as a prompt for a structured audit. Which of your products or services could an AI agent plausibly recommend — and on what basis? If the honest answer is "we don't know," that is the gap to close. Define the specific attributes that make your products recommendable, and ensure they are consistently expressed wherever AI systems might encounter them: product documentation, third-party reviews, and press coverage.

Research published today in HBR finds that only 8.4% of brands appear consistently across ChatGPT, Claude, and Gemini. The study tested more than 1,000 brand mentions across 716 unique brands in 15 retail categories, using identical prompts on each platform. Of those that surface on multiple platforms, 55% are framed differently across systems — meaning AI is constructing its own positioning for the brand, regardless of what the brand intends to project. The research was conducted at Georgetown University's McDonough School of Business and the University of Virginia's Darden School of Business.

The finding that matters most for CMOs: AI systems do not reward awareness, storytelling, or emotional positioning. They reward interpretability — brands whose attributes are clearly named, measurable, and supported by credible, independent evidence. The researchers introduce a new metric: AI recall share, defined as how reliably a brand is retrieved when its attributes genuinely match a user's query. Nike, the world's largest athletic brand, appears far less consistently in AI recommendations than Brooks Running — because Brooks has spent two decades building technical specificity and third-party credibility. The brand that wins in AI environments is not the most recognised; it is the most legible to a reasoning system.

Three actions follow directly from the research. First, replace subjective claims — "high quality," "industry-leading" — with verifiable specifications: measurable features that can be named, compared, and linked to specific user needs. Second, cultivate high-authority third-party validation — independent reviews, expert commentary, clinical evidence — the material AI systems actually draw on. Third, run the diagnostic today: query ChatGPT, Claude, and Gemini with the category-level and problem-specific queries your customers most commonly use, and note what comes back. The gap between what you find and what you intended to project is the work to do.