Call me a pedant. Every quarter for the last few years Criteo has bragged about a 90 per cent retention rate. Every single quarter. Not 91 per cent. Not 89 per cent. 90 per cent precisely, every quarter. We have never seen anything like it.

It’s quite a performance. Except that it’s not. Not quite.

Which-50 even checked with Criteo’s peers and it would be fair to say they are … impressed … by the sheer consistency of this result over the years. It does not reflect their own experience of the ebb and flow of the performance advertising market. Nor of a good many other markets.

Yet take some time to work your way through the Criteo market announcements back catalogue and its earnings call transcripts and there’s that 90 per cent figure each time, immutable.

A few points between friends

For instance, in its Q1 2016 statement to the market, Criteo said “We added over 760 net clients in Q1, the second largest quarterly addition in our history while maintaining client retention at 90 per cent.”

Except that it didn’t. How do we know? From Criteo’s own data.

Criteo’s presentation to investors this week includes a chart that clearly indicates a dip in retention in Q1 2016. This is the first such acknowledgment I’ve seen in my time covering the company.

So either the information Criteo gave the market was wrong then, or what it told the market this week is wrong. We have asked Criteo to clarify which statement was incorrect.

There’s a second smaller dip in Q2, 2017.

Now, to be clear, we are not disputing the 90 per cent figure (well, apart from Q1 2016 and Q2 2017) over the last few years. We are happy to take Criteo at its word. Just like all those investors.

But here’s the thing about retention rates: they are excellent when you are doing an analysis, especially when the company you are analysing says that its retention rate is set in stone.

Happily, when you freeze the retention rate in concrete it provides you with an anchor point to start calculating all kinds of other interesting numbers.

For instance, net new customers — which Criteo also likes to highlight.

The 820 net new customers Criteo says it added in the last quarter is actually the worst performance on this measure since Q1 2016. We know this because the retention rate allows us to calculate and compare these numbers ourselves. In fact, quarterly net new customers have been dropping for the last year, after a big spike in Q4 2016.

You can also easily calculate the total number of new customers (not just net new customers)  or the relative number of new customers compared to returning customers — also in decline each quarter since Q4 2016, following years of stability.

Of course let’s not be bloody-minded. There is unambiguously good news for Criteo in its latest results as well.

Its Q4 revenue grew 19 per cent (or 16 per cent at constant currency) in the quarter to $674 million, while EBITDA increased 45 per cent (or 36 per cent at constant currency) to $120 million (or 43 per cent of revenue ex-total acquisition costs).

That has helped drive a big recovery in its share price from a low of $22.44 earlier this month to $30.51 at the most recent close of trading. Still, that is a long way shy of its high of $54.9 in May last year.

Are we being pedantic? Sure. But numbers matter. And at least twice in the last two years, Criteo told the market its retention rate was 90 per cent, when it wasn’t.

Previous post

How Do You Make Sense of Your Martech Stack? Visualise It.

Next post

Cover Story: Why Australia's New Data Breach Laws 'Miss the Mark'

Join the digital transformation discussion and sign up for the Which-50 Irregular Insights newsletter.