With the end of 2011 in sight, I thought it would be a nice time to reflect on the good old days. Remember the good old days? Back then, men were Real Men, women were Real Women, and SERPs were Real SERPs. Good, decent, God-fearing SERPs had 10 listings – no more, no less. You could count them on your calloused fingers after a hard day’s work, and you knew that all was right with the world.
Then Google got fancy, like some city boy wearing $100 pants and a man-purse. Now, decent folk everywhere are subjected to images, news, videos, and coffee shop listings, when all they want is a good, old-fashioned link. Worse yet, you can’t even count on a SERP to add up to 10. No sir, I don’t like it.
It’s getting harder and harder for SEOs to count rankings the old-fashioned way. While we need to let go of rankings as our only metric, being able to count a SERP reliably and consistently is still an important SEO task, and it seems to be getting tougher every day. I’d like to give a few examples of SERPs that don’t add up to 10, and discuss how to cope with Google’s new math.
Sadly, even the children aren’t safe from Google’s new math. Here’s a Google SERP for “Muppets” (logged out and with the “pws=0” depersonalization parameter added):
Let’s say we had never seen a vertical search result before, and we just counted them the old-fashioned way. These numbers are in red. If we skipped the “Top References” section (since the headline isn’t linked), we’d end up with 13 results for this SERP.
Of course, you and I know that vertical SERPs are different, so we’d be smart enough to skip Showtimes, Videos, Images, and News, right? That count is in orange, and we’d end up with – hold on – only 8 results. So, what’s missing? Well, it looks like the video results do count – even though Google owns YouTube, those are still external links. Using the green counting method, we end up with 10 results, as expected.
These results were depersonalized – what if I add in personalization, and with it, social signals? Surely my friends will save me from this counting conundrum! Sadly, they only add to the mess. On a search for “muppets” while logged in, I get an extra listing:
Although the logged-in SERP varies a bit from the example above, the social listing is an add-on. Even by our SEO-savvy green counting method, it’s #11. Fortunately, you can currently tell these apart by the “…shared this” indicator at the bottom, but expect social results to evolve dramatically over the next year. When counting your SERPs, trust no one.
But wait, it gets weirder. Let’s look at a SERP for good, old-fashioned pizza pie. This one is also depersonalized, but it’s localized to Vero Beach, FL (I discovered this one by accident, based on the location of one of my hosting companies):
Just for comparison, let’s count them the old-fashioned way (in red). If we lump in everything, we get 16 results. Obviously, we’re smarter than that, so we’ll cut out the Images and News results. That count (in orange) is 14, still 4 over our sacred 10.
The savvy SEOs among you will immediately recognize that the top results have addresses next to them – these are integrated Places results (a 7-pack, in this case). Problem solved, right? Let’s remove those Places results and, sure enough, we get (in green) – wait a minute – 7? That’s right, this page has 7 organic results.
Hold on: 2 of those Places results have Google Maps addresses and not their own URLs. Those must not be “real” results. Subtract those 2 from the orange count and we get – tada! – 12. Ok, wait a minute, I’ve got it now. Domino’s and Pizza Hut both have mini site-links – they must actually be organic results. Add those back into the green count, and we’re up to – crap – 9.
This counting thing isn’t turning out to be so easy after all. If we can’t tell which listings are technically “organic” (not Places or vertical) by looking at them, maybe the source code can help us. There are markers in the code, but they’re tough to tease apart. The most telling markers are currently as follows:
(1) The
Wrapper
Each listing’s link is wrapped in an
tag, but that header doesn’t seem to distinguish between Places and organic results. Even expanded site-links are wrapped in this
tag, so it seems to be purely a design convention.
(2) The “cd=” Parameter
(2) The “cd=” Parameter
In each listing’s Google cache link, there’s a parameter called “cd=” which counts up with the listings (cd=1, cd=2, etc.). At first glance, “cd=” isn’t very reliable, because it seems to tag vertical categories on the left-hand navigation as well (“Images”, “Maps”, “Videos”, etc.).
In the example above, though, something interesting happens. The first 3 listings are tagged with cd=1, cd=2, cd=3, but then the next listing to get tagged is the Wikipedia entry, with cd=8. The rest of the listing each have a cd= parameter. This suggests that only the top 3 Places listings (in this example) are being counted the same as regular organic results. Finally, 3 + 7 = 10.
Sorry, but it probably won’t. Google has been experimenting mad-scientist style with the SERPs in 2011, and I only expect that to continue. We’ve got little choice but to adapt and do our best to keep metrics consistent. It’s entirely likely Google could even apply the “infinite scroll” approach used in Google Images to organic listings, killing SERP pagination as we know it.
While it’s fun to dig into the inner workings, the reality is that we need to take a broader view and stop relying only on rankings. Rankings will become tougher to measure and personalization is only going to get more aggressive. If you aren’t already training your clients to look at their organic traffic, unique keywords, organic conversion, and deeper metrics, it’s time to get started.
