The report shows that, generally speaking, for short keywords (one to two words) it's not profitable to bid your way into the high positions. The most profitable positions are the fifth and sixth positions for these types of keywords. These types of keywords lose money in the first and second positions.
This trend will likely continue as more big brand advertisers enter the market and consider placement in those positions as an investment in branding.
However, the story is quite different for three-word keywords, where the most profitable position is the second and third position (but these still lose money in the first position). The results are so different because the big brand advertisers likely aren't interested in the lower volumes of clicks available on the longer tail keywords.
AdGooroo built a detailed model that shows this by analyzing data for 333 keywords that had at least 200 clicks during the timeframe of the study. The keywords also had to vary by at least one position during the time of the study, and came from a wide variety of industries.
Therefore, direct marketers should hover in position five or six for short keyword phrases, and push towards position two or three for longer keyword phrases. It's reasonable to assume, but not proven by the report, that the three-word keyword data would be extension apply to four- and five-keyword phrases.
Digging Deeper -- Cost Per Action
Using the report, I recreated much of the underlying data to see if we could see some other interesting things.
I've long been a fan of getting the most accurate data available to drive PPC campaigns. This can unfold in multiple tiers. For example, many people rely on CPA models to run the PPC campaigns. While this is a reasonable approximation, and it works, it comes at a cost.
Using real revenue numbers and subtracting out the cost for each particular conversion, and then feeding that to your bid management tool, is a much better model. Better still, feed your PPC management tool real gross margin numbers (sale price minus manufacturing and delivery costs) minus the cost for each particular conversion.
To illustrate this issue, I made a couple of sample assumptions: that the sale price of the product was $49 (the same assumption that AdGooroo used) and the target CPA someone might select might be $30.
As I did the analysis, the benefits were dramatically demonstrated. For one-word keywords and two-word keywords, there was little difference in the results. One-word keywords meet the CPA goal in positions six and seven, and two-word keywords meet the CPA goal in positions five, six, and seven. This is pretty much the same result as the revenue minus PPC costs per conversion data used by AdGooroo provided.
Three-word keywords, however, tell a different story. The CPA goals here are met in positions three, four, five, six, and seven. The AdGooroo data, which uses a more precise measurement model than CPA, showed that positions two and three are the best position from a profit perspective, and that positions four, five, six, and seven would net less profit for the advertiser.
An advertiser using a CPA model on long tail keywords would likely focus their position goals too low, and miss some real opportunity, resulting in lower profits to their business.
So a large brand advertiser may still want to pay the extra dollars because the direct profitability of the campaign is not their goal. But if you're a direct marketer using PPC, the position recommendations made by AdGooroo in this report, and make sure you provide your bid management systems the most accurate data (and model) that you can. It will be well worth it.