Peer Benchmarking for Banks in Late 2026: Reading the Signals Others Cannot

When the EBA released its Spring 2026 Risk Assessment Report in June, the number that circulated inside innovation offices was not the capital ratio or the cost of risk. It was the finding that operational and cyber risks are now the primary concern of European supervisors, sitting on top of what the ECB describes as an environment where more than 85 per cent of supervised banks are using AI in some form. Add the EBA's February 2026 peer review on ICT risk assessment under the SREP, and a pattern becomes clear. Regulators now know more about your peers than you do.

The problem

Peer benchmarking sounds simple until you sit inside a bank and try to do it. The information you can see publicly is mostly wrapper: earnings calls, tech partner press releases, vendor case studies with no attribution. The information you cannot see is the interesting part. Which of your regional peers actually put an agentic AI pilot into a supervised production environment in Q3? Who quietly walked away from a signed proof of value because DORA testing exposed a concentration risk that had not been priced in? Which bank in the Nordics moved from a monolithic core partner to a sidecar architecture, and why?

Under delivery pressure, these questions are treated as nice to know. They should be treated as a control. The absence of peer signal is not neutral; it makes every build, buy or partner decision more expensive, because you are pricing risk with a smaller data set than the vendor is using to price the deal.

There are three structural reasons peer benchmarking is broken inside a financial institution. First, delivery pressure has stripped travel budgets and offsite time to the bone. The head of transformation who used to swap notes at three offsites a year now attends one. Second, regulators, most visibly the ECB, the EBA and the ESRB, are asking sharper cross-institution questions, but the resulting insights sit inside supervisory colleges, not on your desk. Third, vendor noise has scaled faster than internal filtering: the average European bank now receives more than a hundred qualified fintech introductions a quarter, most of which reference peer logos with no verifiable adoption depth behind them.

Common approaches and their trade-offs

There are four common routes European banks take to close the peer gap, and each has a real trade-off.

The first is the industry survey. Consultancies publish annual benchmarks on cloud, AI, payments modernisation and cyber. These are useful for a board deck but poor at operational depth. By the time a survey is published, the adoption curve has moved. The named institutions are also, unsurprisingly, the ones with the strongest brand incentive to appear ahead.

The second is the conference circuit. Money 20/20, Sibos, The Banking Scene, Point Zero Forum and the like remain the fastest way to compress a year of market signal into three days. But they consume the calendar, and they favour the vendors who can afford the largest stands. What you see on stage is not always what is being deployed in the back office.

The third is the informal network. WhatsApp groups, LinkedIn back channels, old colleagues moving between banks. This is genuinely valuable, and often the most honest source. It also decays quickly under delivery pressure: senior peers who are heads down on DORA remediation or PSD3/PSR readiness stop replying. It does not scale, and it does not survive turnover.

The fourth is the vendor round trip: asking your suppliers what they see in the market. This is efficient, but the incentive alignment is obvious. Vendors will show you the peers who are furthest along in adopting their product. They will not show you the peers who have paused, walked away, or are quietly building in-house.

None of these are wrong. They are, however, insufficient in a market where regulatory readiness under DORA, the EU AI Act, PSD3/PSR and FIDA is now the primary driver of both buying and pausing decisions. The peer signal has to be closer to real time than an annual survey, more selective than a conference, more scalable than a personal network, and more honest than a vendor briefing.

A smarter route

The route that seems to work best inside European banks in late 2026 combines three characteristics. It is closed, in the sense that the people in the room are peers under Chatham House rules, not vendors. It is curated, in the sense that the topics track live adoption decisions rather than horizon-scanning slides. And it is written up, in the sense that the discussion produces something a busy innovation lead can circulate on Monday morning.

This is the space in which formats like Discovery Innovation Meetings, small Peer Forums and closed Roundtables now sit. Not as a replacement for the industry survey or the conference, but as the layer between them. A well-run peer forum for heads of innovation, heads of payments or transformation directors will typically pull together six to twelve institutions of comparable size and regulatory profile, set two or three questions on the table (for example, agentic AI in fraud, sidecar core banking, ISO 20022 structured address readiness), and produce a shared read-out. Finance X Magazine and adjacent editorial formats then do the follow-up work of turning that read-out into something you can hand to a stakeholder who was not in the room. The Connector has been running variations of these formats with European financial institutions since 2018.

Two features are worth naming, because they are what makes the format hold up under delivery pressure. The first is comparability: participants are selected so that regulatory regime, size band and business model are close enough that the answers are actually transferable. Comparing a Tier 1 universal bank with a payment institution on DORA testing is interesting; it is not benchmarking. The second is cadence: the format runs on a rhythm the innovation team can actually sustain, typically quarterly, rather than an annual moment where twelve months of change has to be compressed into a single afternoon.

None of this is exotic. It is closer to how supervisors already exchange information across colleges. The interesting shift in late 2026 is that European banks are starting to treat peer intelligence as an operational input rather than a networking benefit.

One useful test: ask whether your innovation office can answer, in ninety seconds, what three comparable European institutions did last quarter on the topic you are currently deciding. If the answer is a shrug, the peer benchmarking function is missing, whatever it is called on the org chart. This is not a criticism; it is the practical reality of running an innovation mandate in 2026, when supervisory expectations from the ESMA and the EIOPA on cross-sector adoption of AI and data-sharing are also being fed back into national supervisory dialogues. The gap is systemic, not local.

Why this matters right now

Three deadlines make the peer gap expensive between now and the middle of 2027. The first is the EU AI Act, whose 2 August 2026 obligations for general-purpose AI and high-risk systems are now live and being read across into internal model risk frameworks. The second is DORA, whose second full year of supervision is producing the first meaningful cross-institution findings from the ECB and national competent authorities such as the FCA, the PRA, BaFin and the AMF. The third is the PSR/PSD3 package, moving through implementation while FIDA continues its trilogue. Each of these has a peer dimension that a solo institution cannot see: how are other banks scoping the AI Act high-risk register, how are peers responding to the first DORA thematic reviews, which payment institutions are adopting the SEPA Instant Payments Regulation infrastructure ahead of the November 2026 structured address deadline.

The cost of getting this wrong is not reputational. It is procurement. An innovation lead who cannot see peer adoption depth will over-buy in one place and under-buy in another. Under DORA, over-buying introduces concentration risk. Under the AI Act, under-buying leaves capability gaps that will be visible in the next supervisory dialogue. Standards-setters like the BIS and the FSB are also nudging supervisors toward more cross-institution transparency, which raises the price of being the bank that cannot show its working.

Closing thought

Peer benchmarking used to be a soft input. In late 2026 it is a control. If your innovation office cannot show, on a rolling basis, what comparable European institutions are actually adopting, pausing or walking away from, you are not benchmarking; you are guessing. The banks that will look competent in the 2027 supervisory cycle are the ones putting a small, disciplined peer intelligence habit in place now, before the next round of deadlines lands.

The question worth putting on your Monday agenda is not whether peer benchmarking is useful. It is whether the peer signals you rely on today are close enough to real adoption to be worth the price you are quietly paying for their absence.