Dropping Bids on Bing: Be Ready
When Bing picks up all of Yahoo’s traffic, be ready for some turbulence. Why? Because advertisers are going to learn very quickly, that the average value of the traffic coming from Bing ads is likely to take a hit. Those who drive by efficiency metrics will need to drop bids on Bing; those who spend a budget should expect a decline in ROI.
The principal is similar to what we discussed with respect to the syndication networks last year. Mixing traffic of varying values and preventing segmentation forces smart marketers to bid to the averages. This means spending a bit less on the higher quality traffic than we should and a bit more on the lower quality traffic than we should. Inability to parse the traffic and bid it differently leads to less spend at the same efficiency or reduced sales off the same budget.
Bing is rolling out controls for the Yahoo syndication network, so we’ll at least be able to segment that traffic, but part of the deal between Bing and Yahoo requires that advertisers not be able to discriminate between Bing.com traffic and Yahoo.com traffic.
We promised a while back that we’d take a look at the differential in traffic quality and that turned out to be a bit trickier than I imagined it would be.
The trouble is that done sloppily, one could easily end up comparing apples to oranges and producing bad numbers. For example, Yahoo doesn’t really have an “exact match” option, so comparing a Yahoo ad to a similar ad on Bing one has to tease out the effect of the different matching algorithms. Additionally, failure to exclude the syndication partner effect would also skew results. Finally, it is a mistake to view the data in aggregate; because the competitive landscapes are different, a greater portion of traffic on one engine may come from higher converting keywords.
My first reaction to all these complexities was to put off doing the analysis… :-)
It turns out not to be all that difficult.
Instead of comparing ads, I compared the results of identical user search strings, and restricted the results to those searches coming from the engines’ domains. Vagaries of match types could mean that some users were served less than ideal ads with less than ideal landing pages, skewing the results somewhat. We then also had to restrict the study to those user searches that led to statistically significant conversion volumes on Bing.com, so that the comparisons would be meaningful. That limits the number of cases quite a bit.
In most cases, the quality of traffic on Yahoo.com was somewhat lower than similar traffic from Bing. We found the median to be about -15 %. However, the differences seem to vary by vertical, and, sadly they even seem to vary between search strings for the same advertiser! As always, your results may vary, so we strongly encourage you to do your own research.
The traffic quality differential isn’t the only piece of the puzzle, though. The other key variable is the volume discrepancies between the engines. For some advertisers Yahoo brings in 100% more traffic than Bing does and as such when the traffic is combined, the lower Yahoo traffic quality has a disproportionately large impact on the combined average forcing the bids down more than simply “splitting the difference” would suggest.
For convenience, I built a little table showing the impact of quality differential and volume differential on what needs to happen to the Bing bids.
As an example, the highlighted cell shows that if the Yahoo traffic quality is 15% less than Bing’s, and Yahoo brings in 75% more of the traffic than Bing, then the combined impact on Bing bids would be to drop them by about 9.5%.
We can’t study this cleanly for very many user searches because the data is so sparse. Since there appear to be differences between keywords for the same advertisers, accurately anticipating the necessary bid adjustments is beyond difficult. Also, it’s not clear that the clean apples to apples view is the right view. Bing’s broad matching logic is different than Yahoo’s and that means that the same searches on Yahoo.com will fire different ads than would have been fired by Yahoo’s system. Hopefully they’ll fire more targeted ads, but the point is some of these differences don’t lend themselves to data modeling.
There is going to be a big thunk in the performance of Bing ads as the traffic goes from being 100% Bing.com searches to mostly Yahoo.com searches. The thunk will be HUGE for those who don’t separate out the syndication partners. The slow transition may keep this from raising huge fire alarms, but it will fundamentally change the quality of traffic on Bing.
EVEN BIGGER PICTURE:
This could actually end up having a material impact on Bing’s revenue. Obviously, Bing will generate some revenue they didn’t get before from ads they’re serving on Yahoo and the Yahoo syndication network. However, most of that revenue will still go to Yahoo. Couple that with the fact that advertisers will likely find themselves lowering bids on Bing by 5%, 10% or 15%, and Bing’s revenue could take a hit.
…expecting calls from the investment community any minute now :-)