For credit unions, buying growth used to work. It has quietly stopped.

 

The number of federally insured credit unions fell to 4,250 as of March 2026, down 161 from a year earlier, which pushes charter counts within reach of banks for the first time. Consolidation on that scale happens to an institution rather than for it, so mergers are no longer a good path to growth.

 

What remains is organic growth – winning new relationships and deepening the ones already on the books.

 

That would be a manageable position if buying growth still worked, but it no longer does. Curinos now puts the average cost of acquiring a checking customer at $559, roughly double the 2018 figure, and the marginal cost of the next customer climbs past $1,400. Set that spend against a digital funnel that completes only 22.9% of applications, and the shortfall stops looking like a marketing problem and starts looking like a structural one. Every dollar poured into a leaking funnel simply subsidizes whichever institution converts better.

 

If organic growth is the whole game, then the question worth answering is where exactly it fails. That becomes clearer once organic growth is treated as an engine with distinct stages, each capable of breaking on its own, and each required to work for two very different audiences. One is people who are not yet members. The other is people who already are, and that group holds the largest unclaimed opportunity while receiving the least attention.

 

To put it in perspective, Accenture reports that North American consumers have roughly seven financial products but only about 3.4 of them at their main institution. Credit unions run sophisticated acquisition programs yet almost nothing comparable for the members they have already earned. The engine breaks first at the very top, before an application is ever started.

 

 

 

For two decades the top of the funnel was a search result, and that surface is now closing. Across the first four months of 2026, 68% of U.S. Google searches ended without the person ever clicking on a result generated, while banking's organic search traffic fell 27% in 2025. As search recedes, discovery is relocating into a conversation with an AI model, and in that conversation community institutions are close to absent. One 2026 analysis found that bank-owned domains supply just 6.8% of the sources AI assistants cite on banking topics, and Wikipedia, Bankrate and Investopedia are each referenced more often than every bank-owned domain combined.

 

This is a harder problem than a slipped search ranking, because there is no auction to enter. A model's answer is assembled from what it has already read, and no bid inserts a credit union into it. The obstacle is often invisibility of a more literal kind, since rates, fees and eligibility rules tend to live in PDFs, images and browser-only widgets that a model cannot read. Worse, the problem reaches existing members and not prospects. Comscore found that in early 2026 a quarter of U.S. credit card applicants had used ChatGPT in the month before applying, so a member asking which card to open may be steered toward a product their own credit union already offers, under a competitor's name.

 

Intent deserves the most urgency because it behaves unlike the stages beneath it. Origination and funding are engineering problems with known fixes that show results within a quarter, whereas intent is positional. That position is being set right now, cannot be purchased retroactively, and compounds as models lean on the sources they have already cited, so every quarter it goes unaddressed the cost of entry rises. Yet even the institutions that solve intent gain nothing if the applicants they attract cannot get through the door, which is where the next stage fails.

 

Intent is worth nothing if a member who has already decided cannot finish, and most cannot. Cornerstone Advisors’ benchmarks found 3.36 applications abandoned for every account opened online. At a typical institution that adds up to thousands of consumers a year who chose the credit union and never got in.

 

What makes this so frustrating is that the friction is usually self-inflicted and almost never diagnosed. One super-regional bank ran 60 abandonments per hundred applications against a peer average of 40, and the entire gap traced back to two cumbersome form questions that nobody had thought to examine. Abandonment stays invisible until somebody instruments it. Fraud control then compounds the problem, because where automated tools cannot separate a real applicant from a synthetic one in real time, the fallback is manual review, slow enough to cost good applicants and porous enough to miss bad ones. The stakes keep climbing, with U.S. consumer losses to new-account fraud reaching $6.2 billion in 2024, more than double the figure a decade earlier.

 

For business members, the funnel frequently does not exist at all. In one 2026 benchmark, only 17% of institutions offered digital business account opening and 20% could originate a business loan online, so the most valuable relationships a credit union can hold are routinely asked to visit a branch. Existing consumers face a quieter version of the same failure, since opening a second product often means re-keying data the institution already holds and re-verifying an identity it verified long ago. In every case the workflow was built for strangers and never rewritten for members. The encouraging part is that none of this is permanent, because conversion is a design variable rather than a fixed rate, and institutions have lifted it sharply by moving decisioning into configurable rules without replacing their core systems.

 

Any credit union can locate its own leaks in an afternoon, but the most revealing metrics are the ones that rarely reach a board report.

 

The first is the day-30 funding rate, reported separately by channel so the branch number cannot conceal the digital one. The second is how many applications the institution abandons for every account it opens, and which fields lose people. The third is whether a member who asks an AI assistant about a product the credit union offers hears its name at all.

 

Organic growth is the only growth these institutions have, so none can afford to leak at every stage at once. The one that already knows its own day-30 funding rate by channel is ahead of most of its peers, and the clock on the intent stage is running whether or not anyone is watching it.

Philip Paul is CEO of Cotribute, which provides account opening, lending, and fraud decisioning technology to credit unions and community banks. Over nearly three decades as a founder, operator, and investor, he founded Cerecons, a healthcare SaaS company later acquired by HCSC. He also serves on Biola University's Board of Trustees. Cotribute powers AI-driven digital account opening, loan origination, and agentic growth tools for financial institutions across the country, from Fortune 500 banks to regional credit unions.