The Rise and Fall of Remarketing (CPC’s, at least)

by

When Google AdWords rolled out its remarketing feature, many of us felt we’d stumbled on an exciting — but modest — new high-ROI targeting segment. By combining contextual targeting (keywords denoting which sites in the network our ads would show up on) with audiences (people who had done something in relation to your website in the past x days, cookied and followed around), suddenly these ads that generated some user complaints about “feeling stalked” were feeling like a pretty neat bonus for marketers, without being unduly intrusive. Early case studies showed ROI off the charts. On some accounts, as a result, we did a lot of this — implementing it carefully one ad group at a time.

At some point Google sent the word out to its reps, and through them, to agencies, that broader-based remarketing (no context, just audience, using impression caps if you like) was more practical. It offered wider reach. And if it wasn’t wide enough, we were given case studies that encouraged significant bid increases just to make sure you were “using it correctly” (paraphrasing here). These more general campaigns would work differently, but would move the needle more, and would be easier to implement.

It was at that point I felt that the majority of advertisers were being encouraged to bid unrealistically, in multiple dollars per click, on inventory that — while valuable — couldn’t support those levels when market forces (CPC’s and bid levels based on performance, and easy comparisons with other traffic in your account) kicked in.

Even with ongoing optimization, we weren’t finding the CPA’s quite within our target range. I imagine other advertisers are gradually coming to this conclusion. (And yes, the ROI for the original, narrower context-meets-audience version of this remains very high, so that’s a separate issue.)

This softness in demand for more general remarketing is indeed beginning to manifest itself. After several modest bid reductions for one retail client we work with in this area, we’ve recently seen no slackening of click volumes, and indeed a gradual improvement in ad position.

This smells every bid to me like the “reverse bidding wars” that we relish in this industry, that come along every so often until the bids reach a more realistic equilibrium.

rmkt

From time to time in this auction, market forces do work in our favor. And not even clever “best practices” Powerpoints from Google can keep us all overbidding forever.

Here are some stats in connection with this case study:

  • In recent weeks, the ROAS on the generalized remarketing effort is 2.08 – below where we’d like to see it. The conversion rate is 5.3%. Healthy, but the low ROAS indicates we’re still paying a premium for this valuable inventory.
  • By contrast, display network clicks across the account have had an ROAS of 2.52 over the same period, despite a weaker 2.57% conversion rate — indicating more economical click prices.
  • The search inventory produces much better numbers than either of these.
  • View-through conversions are worth considering — all versions of display, including remarketing, do indirectly drive recall and sales. Hard to value precisely, though.
  • The original remarketing effort — the hybrid of context and audience (ad group by ad group remarketing, with keywords) — performs far better, with conversion rates 8-15% depending on the ad group, thus meriting high CPC’s.

The simple takeaway is that the inventory is valuable and the technique offers much promise, but as with everything else, there is a market value for the clicks, and you can’t talk it up (or down) indefinitely. It will gravitate in the direction of its real worth.

As with other forms of interruption-based advertising (however targeted), the novelty effect wears off and the entry of more participants in a cluttered, finite sphere of user attention may also lead to gradual, or even sharp, declines in response rates and thus prices. That depends to a certain extent on how you use it, but it might also be a collective action problem. The Tragedy of the Commons applies to almost any ridiculously appealing, high-response form of advertising. What’s good for you is good for others, so the user can be bombarded with messages and begin to rebel.

You may also like