The retention numbers everyone quotes, checked against their sources

If you have read anything about keeping customers, you have met these four numbers. We went and read what they are actually based on. Three of them do not support the sentence they are usually attached to.

This is not pedantry. If you are deciding whether to spend a Tuesday evening a week on a room full of people, you deserve to know which of the numbers pushing you into it are real.

"A 5% increase in retention raises profits by 25 to 95%"

The source is Reichheld and Sasser, Harvard Business Review, September 1990. What the article reports is that cutting defections by five points raised profit by 85% in bank branches, 50% in an insurance brokerage, and 30% in auto service. One credit-card case reached 125%.

So the range is real, the mechanism is argued, and the analyses are Bain's own client work rather than an independent study.

The 95% is not in the 1990 article. It appears in later Bain material. The safe version of the sentence is: "Reichheld and Sasser, HBR 1990, Bain client analyses, +25 to 85% across service businesses."

"Acquiring a customer costs 5 to 25 times more than keeping one"

The usual citation is Gallo, Harvard Business Review, 2014. The article's own words are "depending on which study you believe… five to 25 times."

No primary study is named. There is no dataset behind it. It is a practitioner's summary of an impression, repeated until it acquired the texture of a finding.

You can still use it — as "Gallo, HBR 2014, citing unspecified studies." That is a different sentence from the one on the slide.

"Repeat customers spend 67% more than new ones"

Bain, around 2000, in apparel: the average repeat customer spent 67% more in months 31 to 36 of the relationship than in months 0 to 6.

Read that again. It compares the same people early and late. It is a spend curve among customers who stayed, not a comparison of repeat customers against new ones. Everybody who left before month 31 is absent from it.

The popular paraphrase is a distortion of a real finding, which is the most durable kind of wrong number.

"Returning visitors are 8% of traffic and 40% of revenue"

Adobe Digital Index, 2012. The report exists. We could not confirm the figure in any accessible version, and it is aggregated client web analytics rather than a causal effect in any case.

We are not calling it false. We are saying we could not verify it, which is the honest position and the one we will keep taking here.

Why we are publishing this

We sell software to people who run recurring formats. It would be easy to quote all four of these at you, and we would probably sell more.

The problem is that we intend to publish our own measurements later, including the ones that go against us. That promise is worth nothing from a company that was loose with everybody else's numbers first.