Mind the Bid-to-CPC Gap in Paid Search
So you’ve nailed down the value of your PPC clicks with a high level of detail and confidence and you’ve set your bids accordingly given your efficiency target. Over time you notice that performance is consistently more efficient than needed. That’s not the worst problem to have in marketing, but it may represent a lost opportunity to drive additional revenue or margin dollars.
Understanding that your bid is just the maximum amount you are willing to pay for a click and that auction mechanics often lead to observed CPCs below your bid, you set about to raise your bids in order to bring about the CPCs you know you can afford. But how do you go about it? And should you?
Tactic #1: Push Bids Across the Board
It’s tempting (and easy) to simply apply an across the board bid increase, observe the results and tweak as necessary until you’re hitting your aggregate efficiency target. This will work and it will drive additional traffic volume, but it’s not an ideal solution in many cases. The chart below helps to show why:
This is a one day snapshot of the bid to CPC gap for a set of high traffic terms in a single program. Bid/CPC is plotted against the predicted efficiency (A/S) of those terms given observed CPCs. We clearly see a wide range in bid/CPC ratios here, from just above 1:1 (CPCs nearly match bids) to 6 or 7:1 (actual CPCs are far below bids).
Naturally, if our bids are set at the amount we believe we can afford, the lower our actual CPCs are, the more efficient we expect our performance to be and vice versa. Applying an across the board bid push will end up pushing some terms into inefficiency, while others will still be far more efficient than necessary.
Tactic #2: A Progressive Approach
To avoid the pitfalls of the across the board bid push, you decide to apply a progressive approach where the keywords with the greatest bid to CPC gap get the biggest pushes while some terms may not get any push at all. Whether you accomplish this formulaically or through a more tiered approach, this may still require some tweaking in order to hit the high-level efficiency goal.
One issue is that we cannot be certain how two seemingly similar keywords will respond to the same stimulus without some additional information. For instance, going back to our data set, if we take out terms with an average position of 1.5 or higher on the page we see a similar trend as above, but lose a number of points including many of the extremes:
We intuitively grasp that our highest position terms are more likely to enjoy a competitive advantage in the rankings, whether it’s that they are of higher quality or simply that there is little competition. It makes sense that we would see larger bid to CPC gaps for these terms, but if they are already in top positions, will pushing their bids accomplish much more than driving up costs? What about the terms in lower positions, will they respond predictably to our bid changes?
Tactic #3: Consider the Margins
Ideally, we’ll be able to accurately predict not only our CPC for any given bid, but also the incremental efficiency of the additional traffic we expect to receive from a bid increase. While this can be attempted by other means, the safest bet for most advertisers is to utilize Google’s Bid Simulator data.
Bid Simulator data is available via the online AdWords interface and over API. Both provide only a discrete set of points surrounding your current bid, but we can interpolate that information to estimate points in between. While Bid Simulator has its shortcomings, its use should get you closer to the CPCs you know you can afford for individual keywords than either an across the board or even a progressive bid push.
Once you’ve come this far though, it is worth putting just a little more effort into determining the marginal returns of any additional investment. You may even find that your current bids are delivering poor results on the margins. For a more in depth examination of average vs incremental efficiency, check out our post here.
In the end, varying goals and data levels may necessitate some combination of these tactics and it’s important to have bidding systems in place that are flexible, powerful and responsive to new information as it comes in.