It’s been quite some time since Google released a wide-ranging study on the performance of content network placements from a cost-per-acquisition (CPA) standpoint, but hat tip to Paul Bruemmer, there was a recent resurgence of discussion on the forums questioning how representative the study really was of — if not all advertisers out there — then, well, you, the only advertiser who matters to you.
As a reality check, we decided to go back and look at how our clients are faring with content in 2010 (study period is January 1 – April 30, 2010). Below are some notes from content performance on about a dozen anonymized randomly-selected client AdWords accounts. I’ll note the percentage divergence from the equivalent search performance (on a CPA basis), as well as how much the resulting performance has us spending on the network and managed placements channels as a proportion of the account as a whole (the rest being search, in other words).
[Remember -- higher CPA is "worse", lower CPA is "better".]
Client A: CPA 16% higher than search. Content as a % of overall spend: 6.6%. Rating: good. Upside potential: not huge, direct response is key for this account.
Client B: CPA 21% lower than search. Content 32.4% of spend. Rating: excellent. Upside potential: considerable. Quality of customers acquired through content slightly worse, but only slightly.
Client C: CPA 27.8% lower than search. Content 22.7% of spend. Rating: confused. Content generated leads are probably significantly lower quality, but not across the board. As we often find, better measurement is a prerequisite to resuming an aggressive customer acquisition stance.
Client D: CPA 8.0% higher than search. Content 21.0% of spend. Rating. Excellent! Upside potential: considerable.
Client E: CPA 138% higher than search. Content 11.0% of spend. Rating. Very Good, believe it or not. Content is still within ROAS targets, and volume is important. The reason that CPA’s are so low on search is that not much cost-effective volume upside exists above ad positions of 1.8 or so. Upside potential for content: considerable.
Client F: CPA 46.3% higher than search. Content 8.7% of spend. Rating: Fair. Upside? Maybe but it looks like the content as a proportion figure might shrink, not grow.
Client G: CPA 13.0% lower than search. Content 5.2% of spend. Rating: Very Good. Upside: Significant. Recent improvement in content performance here catches us off guard, should be refocusing on volume here.
Client H: CPA 428% higher than search, on very low volume. Content: 0.7% of spend. Rating: abysmal. Response: time to call in the big guns to understand why there are so few decent matches being found between customers and searchers for this particular client. We’ve looked into this one before and made no headway. Time to try again. At least the channel is not losing money because it is being managed very conservatively despite the high CPA’s for the small number of conversions.
Client I: Only one conversion this year from content. Rating: I’m not sure what or who I’m rating here, but overall: FAIL. Although the channel bears revisiting, it also seems to be the case that the product line is so niche and strange that you’d be better off creating new websites, to own them, and to run ads on. Which, come to think of it, is how media generally get created in the first place, isn’t it?
Client J: CPA 34.1% higher than search. Content is 18.2% of spend. Rating: Good. Notes: In this high-ticket B2B account, a previous agency had content running at 70% of total ad spend, at a miniscule CPA about 70% *lower* than search. Unfortunately that was window-dressing, as all the leads flooding in were bogus, nuisance, non-customer type leads. A complete and cautious rebuild was needed (after scrapping the old content builds) to address the business audience, and qualitative checks are needed on lead quality. This is the type of horror story that gives all of the content network, Google, and hapless agencies a bad name. As always, there is no substitute for proper measurement, and professionalism. Upside potential: painful research, but significant potential exists.
Client K: This high-ticket retailer has 500 or so sales from paid search ads, none from content. Whether that’s due to their ingrained anti-content prejudice or our lack of persuasion in revisiting the channel, this looks like a clear FAIL to me. To be fair: once burned, twice shy. Many retailers were burned 4-5 years ago before content “got better”.
Client L: CPA 61.6% higher than search. Content 3.3% of spend. Rating: Fair to Good. In reality, with proper attribution and appropriate attention to brand lift goals, content performance here is still on a par with search, but serving different purposes for this brand retailer who pursues multiple goals with the website. Upside potential: quite large.
Three concluding notes.
- Search benefits from certain “slam dunk” words, especially brand terms. It also benefits from prior advertising, including content targeting. The target CPA for content, in my view, should be at least 15% higher for content, if not higher than that. Only the most brutally literal performance-based approach would give search all the credit for last-click sales and leads. Depending on the advertiser, content ads come higher in the awareness and persuasion funnel. I don’t think it’s smart in most cases to just measure indirect brand lifts or to go on faith, but it could be fair to set your CPA targets much higher, and expect that a very high number of impressions does create brand lift over time. This might be especially true with large, graphical ad formats, which are a totally different animal from small text ads. And it depends on the publication or content they appear next to, and how much repetition might create brand lift and recall.
- Certainly, you should consider the ROAS, lifetime value, or other value (like lead quality) associated with any customer or conversion. Sometimes “leads” or “acquisitions” are of lower value in different channels. The above round-up does nonetheless provide real-world, mostly performance-based, feedback on how a cross-section of real advertisers are doing in 2010, with notes to explain that these are all distinct and mostly defensible figures.
- Probably the biggest concern about aggregate figures like Google’s (even looking at 25,000 accounts) is that they might represent a whole pile of advertisers who somehow don’t represent *you* – or for whom a conversion or transaction is a weak event poorly tied to real world revenues or company growth). I’m satisfied that Google’s content network performs well and transparently for *us* — that is, our cross-section of clients as managed by the Page Zero team, each facing unique real-world channels and goals.
There’s no doubt in my mind that, for the median account, content targeting works as billed. There’s also no doubt in my mind that the channel will continue to grow and improve, and offer real opportunities for publishers and advertisers to innovate and grow.