By Richard Owen & Maurice FitzGerald
Field Notes on Customer AI · Edition 020 · September 15th, 2026
Each Tuesday, Field Notes on Customer AI surfaces what we're seeing in the field: patterns from implementations, ideas worth stress-testing, and the occasional inconvenient truth about how Customer Revenue Prediction programs succeed or stall. No abstractions. No product pitches. Just the working knowledge that tends to matter.
This edition is the third in a series that we are calling "The Great Sorting." It's about the end of the traditional models of CX and Customer Succcess strategies and measurements, and what replaces them.
(Image generated using DALL-E)
If you like what you read, please forward this to your colleagues and friends suggesting they learn more about customer retention and growth by subscribing here.
The Field Read
Britain Made Its Banks Publish Their Customer Scores. Nine Years Later, It Is Switching the Scoreboard Off - Richard Owen
In 2017 Britain required its banks to have their customers surveyed independently and to publish the results twice a year. The theory was embarrassment: put the scores on the wall and competition will discipline quality. Nine years later the Competition and Markets Authority has provisionally decided to abolish the requirement. Almost nobody has written about it.
The results are not what either side predicted. The new entrants won: Monzo and Starling rank first and second, at 79 and 76 percent. Monzo has 15.2 million customers, added three million in a single year, and turns a profit on 1.7 billion pounds of revenue. Roughly one in five British adults bank with them. The institutions supposedly being embarrassed did not change much. Royal Bank of Scotland went from 46 to 48 percent in seven years.
Here is the finding that breaks the theory: if the scoreboard worked, the highest-ranked banks would be gaining customers. They are not; Nationwide, ranked third, leads switching by a wide margin. Starling, ranked second, lost switchers on net. The mandated score does not predict the behaviour it was built to drive.
A metric with a compliance purpose grows a function whose job is the report. A metric with an operating purpose grows a function whose job is the change. They look identical on a slide and behave nothing alike inside an organisation. The state can mandate measurement. It cannot mandate the use of measurement.
Read the full article: "Britain Made Its Banks Publish Their Customer Scores" → Here
The Field Dispatch
The Advice Business Can't Take Its Own Advice – by Richard Owen
Forrester Research has told its investors to expect revenue to fall nine to twelve percent in 2026. Its client retention stands at 77 percent. It has taken a goodwill impairment and is sunsetting its strategy consulting business. In the same publishing cycle, its CX practice shipped three 2025 guides teaching customer experience leaders how to prove the value of their function, defend their budgets, and build ROI models. Across all three documents, the words "artificial intelligence" appear a combined total of zero times.
The corporation is restructuring itself around AI research while its CX practice sells the measurement doctrine of roughly 2015 to the professionals with the most AI exposure of anyone in the building. The navigation maps are beautifully drawn but the coastline has moved.
The incentive structure explains the omission without requiring bad faith. Analyst subscriptions are bought, in large part, by functions that need external validation for their budgets and metric choices. The product being sold, under the layers of methodology, is reassurance, and reassurance that remains sellable must not obsolete the buyer. Follow that constraint through the guides and their contents stop being puzzling. Forrester's investors are being told the truth. Forrester's subscribers are being told about obsolete methodology.
Read the full article: "The Advice Business Can't Take Its Own Advice" →
The Practitioner's Take
The score that predicted nothing – by Maurice FitzGerald
I have done three studies comparing employee happiness on Glassdoor with customer satisfaction on the American Customer Satisfaction Index. The gospel at the time was that happy employees create happy customers. The data said otherwise: Overall there was almost no relationship, except in industries where the majority of employees have direct contact with customers.
I remember presenting that finding and watching the audience look for the escape hatch. They liked the principle and they did not like the data. The principle survived, because it felt right, and because nobody had a budget line attached to the data.
Both of Richard's articles this week describe the same reflex at industry scale. Britain's banks published their scores for nine years and the scores predicted almost nothing about which banks customers actually chose. Forrester publishes CX guidance that its own restructuring contradicts. In both cases, the measurement is real, the publication is real, and the change is absent. The score becomes the output. The actions the score was supposed to cause become optional.
So therefore: take one metric your team reports quarterly. Ask not whether the number is accurate, and whether anyone changed a decision because of it. If the answer to the second question takes longer than five seconds, the metric is decoration.
The Field Tactic
Three ways to test whether your measurement drives change or compliance
1. Track the decision count. For every metric your team reports, log how many times in the past year it triggered a specific action by someone outside the CX or CS function. If the number is zero, you have a compliance metric. Compliance metrics survive audits. They do not survive budget reviews that ask what the function produces.
2. Run the prediction test. Take your top-line CX score and plot it against the customer behaviour it is supposed to predict: retention, expansion, referral. If the correlation is weak or absent, you may be in the same position as Britain's banks: scoring well and losing anyway. The Institute of Customer Service's index predicted switching; the mandated score did not.
3. Ask the buyer's question. Forrester's guides help you prove your function's value to stakeholders. Ask instead: would a stakeholder fund this function if the analyst guide did not exist? If the function's survival depends on external validation rather than internal evidence, the function is selling reassurance.
The Data Point
The deposits that the scoreboard did not cause
The number: 25.7 billion pounds
That is the deposit base at Monzo in FY2026, up 55 percent in a single year. Savings deposits rose 75 percent to 15.5 billion pounds. Monzo now has 15.2 million customers and reported 1.7 billion pounds of revenue; its first profitable year at scale. A bank that did not exist when the mandatory customer satisfaction scoreboard was introduced now holds more deposits than several of the institutions the scoreboard was designed to discipline. The scoreboard did not cause any of this. The mobile app, the seven-second account opening, and the twenty-second daily interaction did.
Source: Monzo FY2026 Annual Report, May 2026.
The Iconoclast Question
The Prediction Test
Your team publishes a customer score every quarter. Has the score ever predicted, in advance, which accounts you would lose? If not, you are measuring satisfaction while you should be measuring risk. The British banks measured satisfaction for nine years. The customers who left never told the scoreboard first.
If you've been reading Field Notes, you know the problem isn't awareness - it's execution. Knowing that AI can improve retention or accelerate revenue doesn't tell you how to make it happen in your organisation. That's exactly the gap The Customer AI Field Guide was written to close. Authored by Richard Owen and Maurice FitzGerald (that's us), it's a practical execution guide for CX, CS, and RevOps leaders, covering how to identify at-risk accounts before they signal churn, convert customer insights into frontline action, build the financial case that gets CFO sign-off, and design Customer AI systems your teams will actually adopt. Theory optional. Results required.
[ Get the Customer AI Field Guide → Now on Amazon]
Field Notes publishes every Tuesday. Each edition focuses on one topic - a trap, a framework, a field observation, or a pattern worth examining. If something in here resonates, or if you're seeing something different in your own programs, we'd like to hear about it.
Responses