in Data & Insights

The Click-Through Proxy. Busted.

If you only treat symptoms, you can’t cure a disease.

That’s why, no matter how many fraud protection tools we develop and how much “progress” we make, our industry never fully defeats click-fraud. True, we have made strides in reducing the numbers, but estimates on how much fraud actually exists vary. Tech innovation marches forward for the good guys, but fraudsters make progress too — developing workarounds and taking advantage of opportunities in new frontiers.

For instance, in Thailand police just uncovered a large-scale click-fraud operation involving a wall of 500+ smartphones and 350,000 SIM cards. Perpetrators rigged the bot phones to inflate the number of clicks on ads in messaging apps. And unfortunately, these guys are not alone.

This problem will not go away — until CTR as a metric does. Here’s why.

We move toward what we measure.
Incentives are powerful. They guide progress and innovation. But even those created with the best intentions can have unintended consequences.

Take colonial India, for example. The place had too many cobras. Venomous snakes were everywhere, creating a public health menace. So the British came up with a plan: pay people to bring in dead cobras. Snake corpses came in at a clip, but people were still getting bitten. In fact, snakebite reports didn’t drop at all.

The British investigated and discovered that they had unwittingly created a cobra-farming industry. People bred the snakes, killed them, and then brought the dead snakes in to collect the reward.

So the British canceled the cobra bounty. And the snake breeders set all of their farmed cobras free.

It’s time to rethink the click.
Click-through rate as a proxy suffers from the “cobra effect.” It fails at its purpose. In fact, it is doing our industry harm.

Click-through rate is the gold standard for search. When we want to measure display ad effectiveness? Not so much. Because of the relevance clicks have in a search context, we use them as behavioral shorthand for “this ad interested me” and “I might like to buy this in the future.” But this conflation leads down the wrong path. Consumers visit sites to focus on content — not to be redirected by clicking on ads. And the fact that consumers choose not to click doesn’t mean an ad had no influence.

Bots don’t buy stuff. A conversion costs money, but a click is free, and this creates a powerful incentive for fraud. Ad buyers focused on clicks optimize to media partners delivering clicks; media partners, themselves optimizing for clicks, seek out traffic sources offering more clicks. Fraudsters at the ground floor just answer the demand.

So how do you find the right metric to incentivize?
Here are a few tips to get started:

Stop settling for the status quo. “But we’ve always done it this way” is not a good mantra for composing a digital marketing plan.

Work with reputable technology. Find partners with the depth, capabilities, and integrity to tie attribution to your real business goals.

Test the incrementality of your campaigns. Allocate resources to the partners and tactics that deliver actual, measurable results.

And most importantly…

Measure what matters. Incentivize the decrease in snakebites instead of the number of dead snakes.

 

 

Quantcast