If you regularly spend any budget on Google Adwords you may be surprised by a recent study conducted by eBay. The findings suggest that spending money on Google Adwords virtually generates no increase in sales and campaigns that do generate sales are outweighed by the investment in paid search. This is a rather large statement to make considering eBay estimate that they spend $51m on Google Search a year in the US, a figure which they feel is all wasted.
As someone who has worked on the advertiser side with Google for 8 years I wanted to dissect the study and give a slightly different view. In order to do this there’s probably 4 key questions we have to ask.
1. What’s the impact of paid search for eBay?
2. Have eBay approached paid search with the right strategy?
3. What is driving their overall argument?
4. Do their findings apply to other large brands?
1. What’s the impact of paid search for eBay?
There’s no doubt that paid search has an impact on traffic for eBay. Using Hitwise UK data we’re able to estimate the amount of traffic paid search drives for eBay UK. Overall, it looks as though eBay UK has seen an increase in traffic YoY, but visits from search engines have significantly decreased in line with the reduction in PPC investment. It’s estimated that overall 20M less clicks are being driven by PPC YoY for February.
Since stopping PPC activity, eBay have seen a decline in both paid search traffic and overall traffic from search engines. Not only this but their overall market share in the auction category has fallen by 6%. This suggests that organic search hasn’t fully picked up on the loss from paid search.
2. Have eBay approached paid search with the right strategy?
Reading through the study it became apparent that PPC doesn’t drive that many additional sales for eBay. However, as eBay is an auction site, are incremental sales the right metric to be measuring?
After performing a general search on eBay for the Blackberry Curve, (see screenshot below) results show 17 items “ending” in the next 24 hours — the time items are most likely to gain a bid. Of those, only 3 don’t have any bids on at all, meaning that 82% of the items will sell regardless of whether or not I bid anything at all. In this case how can paid search drive incremental sales when there’s no incremental inventory? Instead should eBay be focusing on different KPIs. Such as:
a) New customers and their lifetime value. Something they have reviewed but judging by the article doesn’t count as success.
“We find that SEM accounted for a statistically significant increase in new registered users”
b) Reviewing incremental revenue and AOV (average order value). A buyer from PPC entering an auction system won’t generate additional sales, but it should increase the price the item went for.
There’s also the concern that the eBay could have a poor paid search campaign. Over the years eBay has been criticised for the types of keywords and ads they show within PPC — often laughable. However, just because they are bidding on incorrect keywords and showing incorrect ad copy it may not be having an effect on overall traffic. If we look at the keywords driving traffic to site last year (a 12 week period up to the end of March 2012) it starts to paint a picture about what type of traffic eBay are paying for. For the below exercise I’ve stripped out brand keywords and have only looked at keywords where PPC is responsible for over 10% of traffic. Looking at the below list there are two main types of keywords:
1) Competitor terms — which anyone who runs a PPC campaign knows these types of terms have higher CPCs and generally lower CTRs and conversion rates. i.e. are less likely to be profitable.
2) Irrelevant terms — Keywords which wouldn’t be expected to drive any sales for eBay e.g. “national lottery” and “Wikipedia”
It’s also worth mentioning that although the volume of traffic from these terms seem low (less than 1% of overall search traffic), there are an additional 2.5M keywords driving traffic to eBay which will add up to a far greater share.
3. What is driving their overall argument?
The study is broken out in two sections — branded traffic and non-branded. eBay’s study suggest that all branded traffic still comes through organic traffic when PPC is turned off. I can believe this as no one else bids on “eBay” and organic listings feature quite prominently. However, their argument that organic brand plus terms still drive 100% of traffic is flawed. In their example they use the term “ebay shoes”. Looking at Hitwise we can see that last year they owned 84% of the traffic for this term with around 50% coming from paid. Fast forward 12 months and eBay’s share has fallen to 70% — almost a 18% drop. Instead, Amazon now has 9.5% share of traffic from eBay’s brand plus term — interestingly all from PPC. We could assume that this 18% decline affects all brand plus terms, and considering there are over 90,000 brand plus terms driving traffic to eBay, it could be a significant amount of traffic lost.
The biggest flaw in their argument revolves around non brand terms. Knowing that they already lose traffic on some of their brand terms when PPC isn’t being used, we can safely assume that eBay will lose even more traffic with generic terms where it’s far more difficult to rank highly organically.
“Non-brand ads have the ability to attract users that are not directly searching for eBay but the endogeneity remains because the ads may attract users who are already familiar with eBay for which the ad provided little or no information. In short, these users may have visited eBay eventually if the ad were not present.”
I don’t know where to start with this statement. Essentially this is saying that even though they weren’t looking for eBay and didn’t find eBay, they will have visited anyway because they should already know what eBay sells. There is no proof to this argument and you could apply this logic to every single piece of media. Why spend money on TV if they may have visited the site eventually anyway?
Lastly I want to touch on their ROI modelling.
“As is evident, a simple OLS yields unrealistic returns suggesting that every 10 percent increase in spending raises revenues by 9 percent.”
Essentially they are saying that paid search revenue isn’t as high as paid search spend. I don’t want to criticise this too much as I don’t know the detail of the campaign, but this is purely down to optimisation. Are they bidding on the right terms, have they got the right negatives in place, is ad copy relevant, are bids at the right level to drive positive ROI? Judging by the above statement and the keyword list provided above, it’s probably not the case. My guess is that there is high wastage and even a quick cut back of poor performing terms can increase ROI — after all why should eBay be spending money sending people looking for the national lottery to eBay.co.uk?
4. Do their findings apply to other large brands?
In short my answer is no because the nature of eBay. However, each brand would have to take it on a case by case basis.
Let’s take the branded terms argument: “eBay” as a pure brand term has no competitors in the search space. However, when there are competitors on brand plus terms they are losing out on traffic to them. This is even more the case with other retail brands. There is only one place that will sell eBay items, but there are a lot who sell Adidas shoes, this means the impact is multiplied as more competition exists.
I don’t feel I have to tackle the non-branded search topic again as their reasoning and conclusions in their findings doesn’t make sense. I’m sorry eBay, but don’t assume that if you’re not in the search space for a generic term people will end up going to you anyway because of the size of your brand.
Whether search is efficient for them is another matter. Other businesses have different targets and models. eBay is an odd one as they charge a fee when someone sells an item. I’ve already mentioned that because it’s an auction model, items could sell regardless of whether or not a PPC customer bids or not. eBay have also not reviewed the incremental revenue a PPC customer drives. Let’s say driving an additional person to an auction will increase AOV by 10%. On an auction for something costing £20, this is only an additional £2 in revenue. Of this £2 revenue, eBay will only take around 5% — that’s an additional 10p. It’s safe to assume that the CPC will be a minimum of 10p which means even with a 100% conversion rate, any profit is nullified. This is a model almost unique to eBay. Any other retailer would sell a new item and gain their full margin on this product, therefore putting a more positive ROI on PPC.
All in all I think there are flaws in the study. I do think there are some valid arguments backed up with data, but I do think eBay’s approach to PPC has been wrong. They optimise to the wrong metrics, they bid on a range of keywords that aren’t always relevant and in some cases don’t optimise at all. Pulling out of PPC has affected them in terms of traffic, but only eBay are able to prove the profitability of their PPC. Looking at the study I don’t think they have proved the true value of PPC and I certainly don’t think it applies to most other companies who have properly structured, relevant and optimised PPC campaigns.
Email me when Chris Soo publishes or recommends stories