Ernst and Young takes a stab at measuring UK online newspaper revenues

Rick Waghorn and Adrian Monck have already had a crack at the new Ernst and Young report (pdf) that claims the biggest UK newspaper websites made £15m-£20m in ad revenues each last year but could have made £120m-£250m by switching from a CPM model to Google's CPC model. Follows my contribution to the debate.

What an extraordinary document it is. There are a few things to like about it (for example I can hardly now quibble with the claim that "2007 could be remembered as the year when Google, a company founded ten years ago, first overtook ITV1", having broken that news myself in February) but I echo Adrian's astonishment that more than one of the surveys included therein - see page 6 - seems to have been conducted by sending an internal email around 100 Ernest and Young employees (a group representative of the newspaper target audience? Well quite).

Rick points out that the recommendation for newspapers to switch to Google-style CPC is made with no attempt to distinguish between user behaviour on news and search sites, making the estimate that shifting newspaper ad models to CPC would automatically improve newspaper revenues tenfold remarkably irrelevant. (Do people visit newspapers and search engines for the same reasons? Do they behave the same way when they get there? Do they click as much? No attempt is made to acknowledge the question is fundamental to any such analysis.)

And as for assumptions that go into the estimates of current online newspaper revenues ... where does the sell-through rate of 50% come from? Any number that round is probably a guess. In fact, let's go further - that number is certainly a guess since there are very few ways of estimating website STRs using publicly-available data and not one of them returns that result. Yet the accuracy of the whole model depends on that STR figure being right. Drop the STR to 25% instead and revenues under the model are halved: up it to 100% and they double. You can make any number you like come out at the other end just by changing this one guess - if STRs were set to 10% the model would show newspaper websites making just £2m-£4m a year instead of £15-£20m; at 90% £26m-£36m.

What we are left with is a model for estimating online newspaper revenues that if it is right is right by chance; some strategies that are either already being pursued (invest in classified websites? O rly?) or have not been thought through (the aforementioned "news sites can be monetised in just the same way as search sites" claim); and the economically unsubstantiated claim that ad revenues will follow eyeballs online (why? Online could simply introduce greater efficiencies and reduce the size of the whole ad market. Look at Craigslist for how this is already happening). To be fair it is probably quite hard to summarise a whole industry's present and future prospects in seven pages. But this is not even a very good try.

Comments

[Rick]

"...seems to have been conducted by sending an internal email around 100 Ernest and Young employees"

You would think they could at least email a statistically significant number of people. They have about 8,000 staff in the UK alone.

[awrc]

A pedant writes: you might want to get their name right: Ernst

[James Page]

I am bit confused don't many of the newspapers run ppc ad's from both Yahoo and Google anyway.

I would be far more interested in a comparison of banners vs mpu's...

[Seamus McCauley]

James - Good point about newspapers already running Google ads anyway. In 2006 publishers (ie websites not including search engines) made about £450m from display ads and about two-thirds of that sum in Traffic Acquisition Cost rebates from the search engines.

In pure CTR terms, MPUs are about three times more effective than banners. I hope to publish something proving this soon, sorry can't share it right now.

Awrc - honestly never noticed it was "Ernst" and not "Ernest" before. Guess I must have been making that mistake all my life.

[R N B]

Excellent analysis (SM, not E&Y). I'm not sure of how the trackback works, but I linked back here.