Part 2 of Smarter Online Advertising is a big one.
The focus of this part is on ad fraud. That is, clicks (and sometimes conversions) that come from bots rather than real people.
Ad fraud is a big worry for online advertisers, because while it costs them money all the same, bots are obviously worthless to any business.
Many advertisers search for complex and over-engineered ways to block bots from viewing their ads, but in this part, I’ll share with you a surprisingly simple solution to ad fraud that was outlined in the landmark February 1923 book, Scientific Advertising.
(Yes, 100 years ago, long before bots were even invented.)
For some background, in part 1, I explained that re-targeting is a sneaky trick that Facebook and other ad vendors use, to make inexperienced advertisers spend more than they should.
There’s a second sneaky trick they use, and it’s automated bidding.
The idea behind automated bidding is that if you set a manual bid, it might be too low to get you any impressions. Automating your bid ensures your ad continues to get impressions.
Why is this a bad thing?
If you’ve ever bought traffic at a fixed price (for example, 100 clicks for $1 each,) have you ever received an “overdelivery” of traffic (for example, 121 clicks for the price of 100) and wondered why this happens?
It’s because the traffic vendor knows how many impressions they can generate and how much to sell them for, but they can’t know how many clicks they will generate and what their average revenue per click is going to be. It’s always going to vary based on whether people are clicking or not.
So, they price a little bit above the average, and overdeliver to fill in the gap. They’re not really overdelivering at all, just using some accounting trickery to keep their prices fixed.
Back to automated bidding. If you average out an automated bid (not the actual cost, but the bid itself) over the duration of the campaign, the figure you arrive at is the equivalent of a fixed price per conversion.
Since it’s a little bit above what you actually paid per conversion, it’s like you got an “overdelivery” of conversions.
The key difference, though, is you DON’T KNOW what your automated bid will end up being until the campaign is over.
That’s because the price of ad impressions is updated in real time to reflect the natural fluctuation in the market, rather than estimated in advance.
But you should know how much a conversion is worth to your business, and if Facebook is pricing your conversions higher than what they are worth to you, then you should actually WANT the ad impressions to stop. Why would you want to buy conversions that you know you can’t afford?
Here’s what this has to do with ad fraud.
Ad fraud, that is, clicks that come from bots, obviously drives up your average cost per conversion, because bots are never going to convert.
(Sometimes they do convert, but I’ll get to that later.)
So, what is the easy solution to ad fraud? It’s to set a manual bid, equal to the value of a conversion to your business. If the level of ad fraud rises above what you can afford, then the impressions will stop. That’s what you want.
(Not really what you want, but better than the alternative.)
If you had allowed automated bidding, any sudden increase in fraud would have caused the automated bid to go up in response. That’s definitely NEVER what you want.
Even if you are advertising on a platform that uses fixed pricing, the solution is the same. Know how much you are willing to spend per click or per conversion, and don’t buy from anyone who is charging higher than you can afford.
What about more sophisticated ad fraud, that can fake conversions as well as clicks (e.g. with stolen credit cards?)
In this case, your average cost per conversion WON’T go up, and it may even go DOWN, so you CAN’T detect it immediately.
But while you may get the first conversion, you won’t get any repeat purchases from bots. So, if you can track the origin of each of your customers back to a traffic source, you can measure the return of each campaign over the customer lifetime, and you may find that the return is negative (or relatively low) as a result of the fraud. In that case, you know NOT to go back to the same traffic vendor.
100 years ago, long before bots were invented, Claude Hopkins wrote Scientific Advertising and described this principle as the use of COUPONS.
In the old days, a coupon was the most reliable way to track a customer acquisition back to its origin. If you placed a different coupon code in each different ad, then whenever a customer handed you their coupon to redeem their discount or free sample, you could note down the coupon code and reliably know which ad they came from.
Then, you could compare the return of different ads, even over the whole customer lifetime.
You can still do this today with online advertising, only the coupon codes can get applied automatically, so you don’t even need to offer any discounts or free samples for the customer to redeem. Easy!
The online equivalent of “coupons” is the way to compare the true cost of different traffic sources. Ones with high levels of ad fraud will naturally give you a lower return over the customer lifetime, but so will ones with poorly targeted traffic.
At the end of the day, it doesn’t really matter whether the traffic is bots or just poorly targeted people. It matters how much your conversions cost on average, and they’re going to cost higher in both cases.
Whenever you see that, just stop advertising with that vendor and go somewhere else. Stopping the bots is their responsibility, not yours.
That wraps up part 2 of Smarter Online Advertising, so what can you expect in part 3?
I’m going to share one final tip that can save you money on your ads. The final tip is even EASIER to start implementing than the tips I already shared, and can even be used to GUARANTEE that your ads will produce conversions.
Don’t miss it!