When Adoption Looks Like It Stalled

In June 2015, NBC News published a story headlined “Why Has Mobile Banking Growth Stalled?”

The reporting was careful and the data was real. It drew on the Federal Reserve’s annual survey of consumers and mobile financial services, which found that 52 percent of smartphone owners with a bank account had performed at least one mobile banking transaction in the previous year. Two years earlier the figure had been 51 percent. One point of movement across two years, in a category that had been growing quickly. The story also reported that 86 percent of non-users said their banking needs were already being met without it.

Anyone running channel strategy at a bank in 2015 had good reason to read that as a ceiling. The survey was well-designed, nationally representative, and run by the central bank. The 86 percent figure looked like a demand problem rather than a distribution problem.

The FDIC was asking American households something different every two years: what is your primary method of accessing your bank account? By that measure, mobile went from 9.2 percent in 2015 to 48.3 percent in 2023. Over the same period, the share of households relying primarily on a bank teller fell from 28.7 percent to 15.1 percent.

Both measurements were accurate. Only one of them was measuring the thing a bank executive needed to know. The Fed was asking whether someone had touched the channel. The FDIC was asking whether the channel had become their main way of banking. A person could answer yes to the first for years before answering yes to the second, and the distance between those two answers is where the branch network decision lived. The number was right. The conclusion it invited was wrong.

Who is in the number?

The Fed’s measure ran across smartphone owners with bank accounts. For the question the Fed was answering, that was the right population. But smartphone ownership was still climbing steeply in 2015. As smartphone ownership expanded, new people continually entered the eligible population. A flat percentage among smartphone owners could therefore coexist with rapid growth in the total number of people using mobile banking.

The same trap appears somewhere that looks nothing like banking.

More than half the gap was never about technology. It was about who had the problem. Forty-one percent of adults aged 18 to 29 are single, against 23 percent of those aged 30 to 49, and about 69 percent of US adults are partnered.

The residual two-to-one gap is real and deserves its own explanation. The point here is narrower than “age gaps are fake.” A segment can look resistant to a product when fewer people in it have the underlying need, and the only way to find out is to run the comparison among people who have it. This does not explain most adoption gaps. It explained most of this one.

Before calling a group resistant, make sure the people in the denominator actually have the need and the ability to adopt.

What counts as adoption?

The second problem was the bar. “Did you perform any mobile banking task in the last twelve months?” is a very low one. Checking a balance once in November clears it. So does depositing a single check. Once most of the eligible population has cleared a bar that low, the measure has nowhere left to go while everything an executive cares about keeps moving underneath it.

The Federal Reserve identified this itself, though afterwards. In a 2018 note on mobile banking survey measures, Fed researchers separated general mobile banking users from what they called task-only users. Among general users, 53 percent said mobile was one of the three most important ways they interacted with their bank. Among task-only users, 8 percent said the same. Both groups sat inside the single percentage that made 2015 look flat.

A low adoption threshold can saturate long before the behavior becomes strategically important.

Online banking, meaning desktop or laptop access, peaked in 2015 at 37.7 percent and has since fallen by roughly half. Three consecutive surveys would have shown an executive a product that flattened and then dropped.

The behavior underneath both columns is reaching your bank account without going anywhere. That never stopped growing. It moved from one product to another. If you tracked online banking, you saw a category mature and decline. If you tracked remote account access, you saw uninterrupted diffusion. The two dashboards could have supported very different branch strategies.

The Federal Reserve’s 2016 report described the mobile channel as complementing consumers’ use of other banking channels. That was a fair reading of 2015 data. It was also the reading that made the coming substitution invisible, because a complement does not eat the thing it complements.

The pattern recurs

Venmo use runs about 57 percent among 18- to 29-year-olds and 15 percent among adults 65 and older, which looks like a generational fact about money. Measured at the level of peer-to-peer payments generally, the gap is closer to 84 percent of adults under 50 against 57 percent of those 50 and over. The product carries an age gap of roughly three to one. The behavior carries something nearer one and a half to one. PayPal had two decades of installed base, and Zelle arrived inside bank apps that older customers already opened. Same job, three products, three different curves.

A product can decline while the behavior it delivers keeps spreading somewhere else.

Sometimes the curve is real

None of this means adoption curves cannot be trusted. The research turned up a clean one, and it shows where the danger actually sits.

The Census Bureau has published e-commerce as a share of total US retail sales every quarter since late 1999. It counts money that actually changed hands rather than asking people what they do. The comparison base, total retail sales, does not move around. And nobody is asked whether they consider themselves an online shopper, so there is no low bar to clear early.

Then attack it. Seasonal adjustment moves the fourth quarter of 2025 between 16.6 percent adjusted and 18.3 percent unadjusted. Census revised the series in April 2025 to remove nonemployer businesses, changing history retroactively. The denominator covers retail rather than all commerce, so a broader definition would produce a different number.

Each of those moves the level by a point or two. None of them touches a sixteen-point rise across twenty-six years.

Did the capability converge too?

One more check, because a behavior can converge on paper while what people can actually do does not.

Internet use among American adults has reached roughly 95 percent. That looks like a solved problem. Underneath it, home broadband subscription sits near 80 percent and stays segmented, with rural adoption around 73 percent against 86 percent in suburban areas, and persistent gaps by income and education.

What closed the headline gap was substitution. The share of adults who own a smartphone but have no home broadband rose from 8 percent in 2013 to 16 percent in 2025. Among households earning under $30,000, that figure is roughly a third, against about 4 percent of households earning $100,000 or more. Those people have internet access by any reasonable definition. They also do tasks built for larger screens on a phone, including job applications and schoolwork.

The check is not universal. Peer-to-peer payments substituted completely: PayPal and Zelle do the core job as well as Venmo does. Workplace messaging substituted completely as well. Sometimes the substitute is genuinely equivalent, and the convergence is exactly what it looks like.

If adoption spread through a substitute, check whether the substitute actually lets customers complete the job that matters.