What started as a fun discussion over lunch with an industry peer about the possibilities of RTB transformed into a ‘what if’ discussion about turning the ad buying concept upside down – and the potentially significant ramifications of this approach. What if algorithms were tweaked to purposely lose, and, pushing this further, lose at the highest possible price?
Let me preface this post by saying I don’t know if this “tactic” is actually being used today or not. Nor do I know if it is practical. However, the bidding model does allow for this to happen. It’s safe to say that considering what the industry has done so far and the innovation that has come from it, the idea is probably technically possible. I wouldn’t condone the approach, but if done correctly, who’s to say this isn’t a tactic?
If you are building an algorithm for RTB, you are looking to get the widest scale at the best possible price. Advertisers determine their price, gauging data on audience and quality of impressions, and then looking at performance based on goals. With what they know, they can buy as much as they can at their price. When advertisers max out scale (based on budget and bids), there is no more for them to buy unless they can pay more. The higher the advertiser is willing to pay per bid, the more volume it gets. Conversely, the lower the bid price, the less volume they get. This is the winner’s philosophy.
But can advertisers still win if they are purposely trying to lose? Here is where a ghost bidder would sneak in. What if a buyer sets up an algorithm to submit a losing bid at the highest possible losing price? The goal would be to bid a lot and lose most of the time at the highest price, thus driving up the price of inventory for everyone else in the market. If ghost bidders drive prices up, it makes inventory more expensive to direct competitors who are presumably looking for the same or similar users. For example, look at a valuable audience segment like auto intenders. A ghost bidder could submit high bids that aren’t likely to win, but will raise the price of this user segment. One plausible use cause might be for a marketer attempting to run a substantial campaign in a narrow vertical to exhaust its competitor’s budget in a specific time frame just before the buy, so that the marketer could theoretically go in and purchase the same audience at a lower price immediately after, therefore having a major competitive advantage.
This practice would come with great potential for peril. It would be risky for any advertiser because they could lose their budget almost instantly if bidding isn’t precise. Bidding would be based on the assumption of a higher counter bid, if that higher counter bid did not occur – perhaps due to a maxed budget – the ghost bidder would buy the impression. But if we can optimize to win at the lowest possible price, surely we can optimize to lose at the highest.
This type of competitive price “warring” has some precedence. Companies have clicked on competitor ads in paid search just to get the other company to exhaust their budget. In auction reality shows like Storage Wars on A&E, competitors regularly mess around with each other by bidding up the price to make the other pay more when trying to win a storage locker. Sometimes the gamble doesn’t pay off as an aggressive bidder who doesn’t want the product ends up with it.
Again, I don’t condone the practice and this is just a thought experiment. I bring it up because we, as an industry, spend so much time thinking about how to win bids. Why not think opposite? Why not think in reverse? What can we do with a losing algorithm? And what could it teach us about building smarter strategies in general? I can’t imagine I am the only one thinking about this – there are too many great minds in the industry. While there’s no reason to think this is happening, there’s no reason to think this is not happening already either. What do you think? Could you find an effective way to leverage a losing algorithm?