Only 4.5% of CX Programs Can Put a Revenue Number on Their Work. That Figure Hasn't Moved Since 2017.

· CX Pulse

CustomerGauge surveyed 340 CX programs. 4.5% can put a revenue number on their work, flat since 2017. The score is the voice of who answered.

Only 4.5% of CX Programs Can Put a Revenue Number on Their Work. That Figure Hasn't Moved Since 2017.

CustomerGauge published the tenth-anniversary edition of its B2B benchmark on Thursday, built on 340 customer experience programs and 10,139 answers, with six earlier editions to compare against. The programs got faster, better staffed and more automated over the decade. Only 4.5% of them have quantified and reported what their work returns in revenue, and that number has sat in the same band since 2017.

What got better

Closing the loop, which means answering a customer who gave feedback, went from a novelty to the norm. In 2021, 30% of programs set a target for it. In 2026, 62% do, and 38% close within 48 hours, up from 27%. AI arrived in a single year: 55% of programs now use or pilot it, against 2% who had integrated it in 2025. Most of that use is reading. 59% run sentiment analysis, 50% detect themes, 44% auto-summarize. 18% let it generate the follow-up action and 17% use it to predict churn.

What didn't

74% of programs can't state their impact on revenue at all. 60% don't know what share of the company's revenue their survey ever reached. Two-thirds don't track referrals or expansion revenue coming out of the program. Adam Dorrell, CustomerGauge's CEO, put it to CX Dive this way: "They don't know how much of their revenue they're reaching with their experience at all."

The 2026 question was worded more strictly than before, "quantified and reported," so CustomerGauge reads the move from 7% in 2021 to 4.5% now as no progress rather than a decline. Either way, a decade of the discipline telling itself to tie feedback to money produced a one-in-twenty club.

The people who didn't answer

The report's most useful finding is about silence. 40% of the programs surveyed hear back from 10% or fewer of the contacts they ask. The self-reported median response rate is 15%. So the score on the quarterly slide is the opinion of roughly one contact in six, and the slide rarely says which six.

One CustomerGauge client banded its accounts by average score, indexed the accounts that never answered at 100, and found a staircase: every point on the 0 to 10 scale was worth about 3.5% in lifetime revenue, with 9s and 10s at 114 and 0-to-5s at 88. The accounts that never answered behaved like a customer scoring 6.3, nowhere near a promoter. An Asian beverage bottler ran the same exercise against audited revenue across 7,200 scored accounts and got a 0.91 correlation between score and three-year growth. Accounts that never responded grew 2% less than those that did. Detractors nobody rescued shrank 5%, and rescued ones returned to par.

That's the part of the slide that's missing. The score is the voice of the people who answered. The revenue is mostly with the people who didn't, and most programs can't say how much.

If you run CX at scale

CustomerGauge offers one formula it says a spreadsheet can handle: revenue of the accounts that responded in the last 12 months, divided by total revenue. That's your reach, and it goes on the slide next to the score. If it comes back at 30%, the NPS you've been reporting describes 30% of the money, and the other 70% has been treated as neutral when the client data above says it behaves like a 6.3.

Then split the response rate by account tier. If your 20 largest accounts never answer, the program is loudest where the revenue is thinnest. The linked programs in the survey are in a different league on every measure: 46% know their revenue impact against 12%, 34% can show a positive one against 7%, 66% see promoters buy more against 34%. The link is a join between the survey table and the billing table, and the budget conversation changes the quarter you make it.

If you run a smaller business

You may not run a survey at all. You have reviews, a few replies to a follow-up text, and a feel for who's happy. The same question applies: of the customers who paid you this year, what share have you heard from in any form? Write down the five biggest who haven't said a word, and ask them this month. At your size the silent customer is the one you find out about when the orders stop.

A decade ago the complaint was that CX teams measured and didn't act. The acting is mostly fixed; the loops close in two days. What hasn't moved is the question a CFO asks first, and the report's answer is that most programs have never looked at who is missing from their own number. The score is the voice of the people who answered. The revenue is mostly with the people who didn't.

Source: CustomerGauge, "What Ten Years of B2B NPS Benchmarks Tell Us" (Oct. 8), with the Decade Edition report. Adam Dorrell's webinar remarks via CX Dive (Oct. 9).