Can Search Retargeting Skewer CPAs?

Now that "buying traffic" from search engines like Google has become a fairly reliable science, making the most of that downstream flow has become the next big marketing challenge. Web-hosting companies battle fiercely in a market where price wars have shaved margins to the bone. They can buy traffic to their marketing home pages, but at such a high price that the cost per acquisition is extraordinary, says Hartland Ross, president and CEO of eBridge Marketing Solutions, which handles many of these clients. Ross recently tried retargeting the traffic from his Google buys using Revenue Science's BT network. The results were encouraging, but for now this is not a solution for everyone.

BI: What is the conundrum for hosting companies?

Ross: It is an extremely competitive industry. It becomes difficult to acquire clients at an acceptable rate, particularly as the prices go down. Now, in some cases, they are paying three years worth of revenue to acquire a client.



BI: How does search retargeting work?

Ross: The hosting companies spend a lot of money on advertising on a lot of different sources, offline and online initiatives. There are two major online source--affiliate sites, and Google AdWords and other pay-per-click search engines. So, they're paying $100 to, in some cases, $300 or so to acquire a client who is paying $6 to $8 a month. So these visitors are coming to the hosting company sites, and not everyone buys.

If they came through Google, we might be paying $20 a click on the high end. Clearly there was some interest there. We have been able to see if we can retarget that person and hit that person again in some way. Through the Revenue Science model, we implant code on the site and that introduces a tracking cookie on the user's machine. When that user goes and surfs on a number of other sites that have become part of the Revenue Science network, then they are retargeted, by serving them an ad that can be in the form of either a text ad or a banner. Those people have then been re-exposed to the ad for a second time, and we found that we've been able to convert those visitors into clients and bring the acquisition down to somewhere in the neighborhood of $70 to $80.

BI: Decribe the executions. Are text or banners doing better?

Ross: The banners were not implemented until about a month or two ago, so they are relatively new. The text ads were doing very well. Primarily the text ads would have been in the margins.

BI : Is it all priced on a CPC basis? Are there frequency caps, limits on how soon after they hit your landing page they can be retargeted?

Ross: They were all cost-per-click. There are a lot of things we have identified that we would like to implement with the campaign going forward. Frequency caps are one, and the ability to target specific sites. We would like to be able to target by publisher and cherry-pick the ones that are performing better....

BI: It would seem that once you get these potential clients into the behavioral engine, you could then discover a lot about their habits to optimize campaigns in other ways.

Ross: That functionality isn't built in, but we've discussed that. I would like to explore what kind of impact that has on the sales process. Unfortunately, a lot of these companies have got sales coming from so many different sources and are seeing [enough] different types of returns in different places that their general strategy is not just CPA, but also the aggregate number of sales that are being generated on a monthly basis. Even if the acquisition costs are within some acceptable range, the problem is that unless there is a significant number at that kind of volume, they can't justify it.

BI: That begs the question of volume--and whether retargeting just search traffic that happens also to visit a BT vendor's network produces enough heads.

Ross: That's a major issue. The larger companies have a lot of advertising initiatives underway and lots of traffic, so we can leverage that...

But with the smaller companies I would go so far as to say it doesn't work. On a smaller site they aren't getting the traffic... Having said that, the issue with the large ones is that it becomes un-scaleable.

BI: Is there anything you can do at the top of the funnel to drive more in?

Ross: The easiest thing to do is for the network to increase in size, and if that were something we could do, then we could improve that overlap. However, you can also ensure that all the pages that receive a significant amount of traffic have code in them so people are not being driven to a site and not being cookied. If that host doesn't have code in all of the pages receiving traffic then people are falling through the cracks.

BI: Does the model work with sectors that don't have such exorbitant CPAs?

Ross: I think this has a lot of opportunity for other industries. It will be contingent on driving traffic to the site so there is enough overlap. So that's doesn't exist in every industry. It isn't going to work well for high-end B2B products, which are more lead generation. We're looking at more of the consumer side and focusing on small business solutions.

BI: Does it matter that a retargeted ad using behavioral tracking is hitting that user out of context, when they are not thinking of a hosting solution?

Ross: Fortunately, this particular campaign is on cost per click, so the CTR is not the relevant from a cost standpoint. It does have something to do with the quality of the people clicking. Being out of context has come up a number of times. Clients have expressed some concern that their ads will appear on sites that are not relevant. The argument is, well, these people have been to your site already so they have expressed some interest in your solution. But that out-of-context component does remain a factor and that's where I suggest that being able to cherry pick the sites, those that are performing better than others would come into play once we have enough publishers. The problem is if we really start cherry picking, it will whittle away the number of clicks that we've got. At the high end at this point it's not extremely high. We're talking about somewhere around 25 to 30 clicks a day for some of the larger clients.

Next story loading loading..