PLA vs Search: A brief history
When Google opened up the search results pages to sparkling Product Listing Ads, the PPC community exhaled a theatrical and collective gasp. A picture is worth a 1000 words (my lazy English teacher would tell me). Our standard search ads can contain max 130 letters. How could old school search even compete?
But, from the get go, Google tried to assuage our fears. “Look”, they said, “search ads and PLAs are totes different kettles of fish.”
Search ads: they are position based, right? You test your bids, test your keywords, change things around and work out which average positions (a stat which is essentially a stand in for your bid aggression) result in the best returns for your business.
For PLA ads, on the other hand, we can’t even see this position based information. “It’s unimportant”, Google is essentially telling us. It’s not what PLAs are about.
And to a certain extent it isn’t important - Alistair Dent, our director of paid media here at Periscopix, wrote a wonderful (widely read) article about a year ago – trying to get people into the PLA frame of mind. “Impression share”, he shouted. That’s what you should care about with PLAs. Aside from optimizing your feed so that it shows up for the most relevant keyword possible – the name of the game is impression share. Test your bids and work out what impression share gives you the best return.
Impression share. Average positions. Two kettles. Maybe even more different. A kettle and a toaster.
2014: Search gets PLA-yed
However, with Google’s recent spate of PLA related announcements, it no longer seems like these PLA ads are happy to stick to their own side of things.
The first announcement concerned migrating to a new system of ad management called “Google Shopping Ads” – which Helena Clark blogs about here, and which essentially makes it easier to manage ads at product level.
The second announcement was that PLA ads will now often show above the top search ads, like so:
It’s hard to read this as working together and easier to see it as an attack. Search and PLA/Shopping ads are competing for space, competing for clicks, competing for users, now more than ever.
But still in the AdWords interface we see a big fat "Av. Pos. 0.0".
GA-me on: Finding PLA position data
Step in handsome, dashing, heroic Google Analytics.
There is a nifty (if under used) report in GA called Keyword Position.
It’s hidden in the acquisitions tab, then within the AdWords drop down, and can essentially give you keyword performance stats (sessions, transactions, goals, etc) at exact position level. Basically a slightly more in-depth version of the Top vs Other stat in AdWords (which nine times out of ten will tell you most of your clicks come from the top three positions).
However, what we can also see is this:
Yes, position based information by PLA target. My mother would be so proud.
You can change the metric at the side to show a number of key GA stats – transactions, revenue, conversion rate, goal completions and so on.
But what can we learn?
Off the bat some pretty interesting trends come to light.
You can see above that the majority of my sessions (which roughly translate to AdWords clicks) are coming from position 1. This could mean a number of things, maybe a large number of my PLAs are showing as the only Product ad on the SERP, or maybe it means the top spot (presumable first ad on the left?) gets the most clicks (as we often see with text ads). It could even be skewed by google shopping data, with a number of my ads showing in this position (which looks like it should be a form of position 1):
One PLA ad in the Google Shopping interface shows in the area highlighted by the red box, with the other PLAs competing for the same product being relegated to the grey “compare prices” box to the right. This highlighted box, then, looks like it should be another form of “position 1”. Indeed, they are at the top when you click through to compare prices:
As they aren't order by price, this suggest that John Lewis get the top spot based on bid, or a combination of bid and “feed relevancy”.
The idea that PLA position, in some cased may be able to be control by bid is a very interesting one, and is something we shall look at again later on.
Referring back to the GA screenshot: you can also see that the number of PLA positions available goes down to at least 19. The maximum number of PLAs that are realistically going to appear on a SERP is 8. What this likely means is that the sessions below this position are coming through Google Shopping. And the only real time that shopping is going to display individual PLAs in that number, is in the compare prices image above.
Now we can start to get a picture of how many people are actually using Google Shopping at this in-depth level, and it looks like the numbers are pretty small.
Interestingly, if we change the PLA target we are looking at, the trends seem to shift somewhat:
Suddenly position 1 doesn’t look like the be all and end all, but instead position 5 edges ahead (based on clicks here, but also transactions: 11 compared to position 1’s 10).
The drop in sessions between 5 & 6 makes sense, as Google are increasingly showing PLAs in the grouping of 5 shown above, rather than 8, at the side:
Maybe, in some cases, 5 gains more clicks because it is 1st on the second row of 8 PLAs. Or it may well be because increasingly Google is showing PLAs (again as the first screenshot shows) with no additional text ads to the right, meaning there is more white space around position 5, which would naturally draw the eye more frequently. Interestingly – this only seems to happen when I’m signed into my Google account. When I browse in private mode (purely for ad testing purposes – ahem), I get the full spectrum of side ads.
Taking all this information on board then, if PLA position is at least partially determined by bid, and position 5 is outperforming positions 1 through 4 – then theoretically we can now test minute bid changes at PLA product ID level, to see if that change does affect the ratio between the positions.
And that’s just the tip of the iceberg of tests we can do with this data.
What factors determine the proportion of sessions coming from the shopping tab? What sort of products are Google always showing in a 5 part ad space, what sort more commonly appear as one of 8? How does impression share affect the ratio of positions?
These are all things which, it looks like, we can now start to test.
(Stay tuned for some more analysis from us at Agency level!)