A view from Dave Trott: Numbers is the new religion

De Correspondent is an online media news platform – they have an analogy for the way online advertising works.

Imagine Luigi’s Pizza wants more customers, Luigi prints lots of money-off coupons.

He employs two teenagers to hand these out to potential customers.

After a week, one of the kids seems to have done well, all of his coupons have been used.

The other kid seems to have done less well, not all of his coupons have been used.

Luigi asks them where they handed out the coupons.

The unsuccessful kid says he walked the streets giving them out to random people.

The successful kid says he gave them to people already queueing up to buy pizza.

Giving your coupons to people who are already waiting to buy is the “selection effect”.

You don’t get any new customers, just people who were going to buy anyway.

Giving your coupons to people who aren’t waiting to buy is the “advertising effect”.

You get less uptake but they are all new customers, all converts.

This is the difference between online advertising and conventional advertising.

Steve Tadelis was professor of economics at the University of California, Berkeley.

Ebay employed him to prove the effectiveness of their online advertising.

They told him it relied on “proprietary transformation functions”.

He knew this was just jargon for statistics that equated expenditure with sales.

But, as an economist, he knew correlation does not equal causation.

He persuaded them to stop spending across one third of the Google network.

After three months, the results were identical where they had spent and where they hadn’t.

So eBay had been wasting $20m on key-word advertising.

The clicks they got from paid-for online advertising were clicks they would have got anyway.

The numbers showed that they weren’t making $12.28 for every dollar, they were losing 63 cents on every dollar.

So what happened as a result – did eBay change their online advertising?

Quite the reverse, as Justin Rao of Yahoo! and Microsoft said: “No-one wants to know, no-one cares.”

He was employed to run a test on online advertising effectiveness – the client in charge of the budget said to him: “I better warn you, if the results come back negative, I won’t believe you.”

Rao said to him: “If this is about religion, I can’t help you. I have nothing against religion, but I don’t think it has a place in marketing analytics.”

So why do people keep spending vast budgets on online advertising?

I once asked someone at the local council, why all the roadworks happen after Christmas.

He said, budgets for the coming year are set at the end of March.

If you haven’t spent all your budget by then, it looks like you didn’t need it so it’ll be cut.

So, if you don’t want your budget cut, you make sure to spend it before the end of the year.

Is that the reason marketing people spent $273bn, globally, on online advertising?

David Reiley used to head Yahoo!’s economics team.

He said: “Bad methodology makes everyone happy: it makes the publisher happy, it makes the person who bought the media happy, it makes the boss of the person who bought the media happy, it makes the ad agency happy. Everybody can brag that they had a very successful campaign.”

Randall Lewis confronted the man responsible for evaluating Yahoo!’s marketing strategy.

He admitted that he omitted data if it led to the “wrong” results.

As Lewis said: “That is bad scientific practice, but it’s great job preservation practice.”

So, basically, what we already knew.

Marketers are mainly successful at marketing their own marketing.

Dave Trott is the author of The Power of Ignorance, Creative Blindness and How to Cure It, Creative Mischief, Predatory Thinking and One Plus One Equals Three