One of the hardest parts of consulting, in my opinion, is figuring out which metrics matter to a business, and then reporting on those metrics to the client. Oftentimes, in my experience, businesses are not even sure which metrics to report on, so they report on as many as possible so as to not miss out on any important metrics.
Over time, I’ve better learned (though I’m admittedly still not incredible at it) to help my clients realize which metrics are important to their businesses, and more importantly often their bosses.
One metric that I have increasingly moved away from, though it can be helpful in very specific circumstances, is Impressions Data from Google Webmaster Tools. Many of you are probably familiar with graphs that look like this:
Let me show you an example of where this Webmaster Tools impression data is misleading and can actually be harmful to your attempts to manage client expectations.
Impressions vs Traffic vs Users
For almost the past year, I have had a client that is carving out a new niche online, yet is competing against some over well-established brands and websites for SERP real estate. Many of the keywords are moderately competitive, but as this site has grown and we have conducted various link outreach and content campaigns, the site has become stronger and thus able to rank for these terms. Traffic has been up and to the right.
Here’s a traffic graph:
And as you would expect, as traffic has gone up, so has the number of conversions. Put the two side by side, and they look thus:
You can see that in this case, a strong correlation exists between the number of visits and the number of referrals. You may even argue that this is causation, and I do not think I would disagree with you on that.
But now, let’s set these two metrics against Webmaster Tools impression data:
Why Is This The Case?
We need to ask now what the Impressions data within Webmaster Tools is really showing us. According to Google, this impression data exists
The number of times pages from your site appeared in search results, and the percentage increase/decrease in the daily average impressions compared to the previous period.
So, to put it in non-Google terms, you are seeing the TOTAL number of impressions, and the average ranking comes from all of the rankings that your site has been seen at. So, you may have all of your important terms within the top 10 rankings, but many more longtail terms are being seen in the 20+, 30+, and even 90+ range. Your average rankings are skewed and your number of impressions will be drastically off, as your site ranks for terms that will not drive traffic, much less targeted traffic, to your site.
Allow me to recommend that you not report on Impressions data taken from Webmaster Tools, as this can skew the opportunity of traffic available. The only time I might recommend reporting on this data is with a new website where a KPI is market traction and market share, in which case you should report on this metric against competitors.
Thoughts? Disagreements? What other metrics have you found that clients or bosses want reported, yet will not tell them the story that they think it will?
Also, I recognize that I’ve just given you something NOT to measure, so what should you measure? That’s a whole other post that I’ve been meaning to write, but until then I’ll refer you to this awesome post by Marty to get you started.
18 thoughts on “Why Google Webmaster Tools Impression Data is Worthless”
I totally agree John, great post that NEEDED to be written. I just had to explain to some people why this was data was good to look at to get a general feel for total impressions, but it really didnt mean anything.
In my experience, Image impressions are usually the ones that cause this problem. When searching images users normally see 20+ images, so each one of those domains get one impression with the ranking they have there. If the site has a lot of impressions in image search because of a generic search term in which they appear in the position 20, that distorts the metrics you see in GWT or in GA.
In GA you can filter the impressions data taking out the Image results, and by doing that the numbers should be more “normal”. In fact, we should take a look to that data separately for each type of Google property (mobile is very different from web in my experience too)
Take into account that users normally don’t go to page 2+ of search results, so impressions on page 2+ should be a very low number.
Anyway, I agree that this is not really accurate data and we should treat it like that, but it will be the only data we have for keywords once (not provided) increases to percentages above 50%. The problem is how to export the data in a way that it can be segmented later (as Google only shows the last three months).
Let me know if numbers are more accurate by filtering out image impressions.
+1 to the filter idea. Some of my sites have had cleaner data when those numbers are filtered out, but then we have to go back and add mobile to get even closer.
I think Christian makes a great point about image impressions affecting this. However, I don’t think impressions is a totally useless metric. From my PPC days, number of impressions was a very important metric. If you had a high number of impressions but a low number of clicks, it could tell you that your ad wasn’t performing well by enticing people to click. In some instances this meant that the copy needed a change, but it could also mean that you were going after the wrong keywords. You may think your site should go after a certain keyword because it is related, but it may not be a keyword that will get clicks or conversions. If you consistently receive impressions for a specific keyword but a low number of clicks, it may be time to divert keyword focus elsewhere.
That being said, I definitely think visitors is a better metric overall to garner actionable information.
As others have noted above, I think there’s some value in impression data if you apply various filters – especially search type and geo filters.
But it fundamentally makes no sense to report on impressions as a raw metric, just as one should never make a point of highlighting metrics from which actionable business decisions cannot be made. So I agree with you that the raw impressions graph data “is useless.”
I think probably one of the underutilized uses of these data is in uncovering potential. If you’re ranking at position 43 for an important keyword you’re unlikely to see any traffic for that, but GMT can alert you to the fact you’ve got a zillion impressions for that particular keyword. This is particularly useful in an age where collection of ranking data is (mistakenly, in my opinion) frowned upon.
Anyway, thanks for the interesting read and stimulating the conversation around this!
That has always been my gut feeling with the impression data from WMT, good to see it though from your experience too.
Thanks for this. I agree that GWMT impression data isn’t a core metric. I do think it is some general trend evidence of how Google perceives the quality of your site.
“figuring out which metrics matter to a business”
I like to start with business metrics (i.e. revenue, sales, leads, etc).
Then work backwards to tie those to “web metrics” (i.e. traffic, on-site analytics metrics, etc).
I think we know which metrics matter. I think the real challenge is connecting the dots to understand how all those metrics relate to one another.
Totally agree, and have never understood why so many people fawn over WMT data, which I’ve always found to be misleading at best.
Just yesterday I was looking at data from a new site launched 2 weeks ago (so there was less noise to wade through) and some of the keywords appearing in that report made no sense at all. Supposed high positions that just didn’t exist in any of the normal search channels. Treat this data with great scepticism.
Impression data for our clients is almost useless. In my experience with this report, high volume, broad-based searches that a client site “might” appear for – but have no bearing on potential keyword targets or business interest – are found more often than not. As a result, we end up spending more time explaining ineffective numbers rather than uncovering potential terms we could improve upon (which I feel like is what the report is supposed to reveal).
Pingback: Search Engine Weekly Roundup #2 - Wow Internet Blog
Thanks. I was needlessly worrying too much about the impressions data. Useful info.
I like the GWT Traffic data and have always used it as a guide to trends. The data is not an accurate reflection of your traffic or ranks, but it is a good barometer. If you see a lot of keywords showing less impressions, you know you are appearing less, but you can use other metrics to work out why. Less impressions without ranking drops implies a drop in demand for those keywords (seasonal, perhaps). Ranking drops as well as less impressions implies your site has gone down the rankings generally.
Sometimes you see a range of related keywords dropping while others rise, which is typical after a Penguin or Panda refresh.
Also, if you see more click-throughs generally, this may imply an improvement in the content you are showing in SERPS.
Nice post, cant agree with you more. Recently after the panda update, I was seeing large number of impression coming in from Ukraine where as my site targets US visitors. Can you tell me what could have caused this?
I am unable to figure out just for one day my impression numbers in GWT are completely not related to what i am seeing.
I wasn’t even sure what Google Impressions were before reading your post. I do see a very similar graph like in your figure. It’s also interesting that there is this one sudden spike in impressions. I guess in the long term, it’s more important to develop valuable content.
Pingback: Google Webmaster Tools: Lots Of Impressions but No Clicks On Webpage? Why? - Quora
We found this page while investigating why the number of impressions reported in Webmaster tools dropped, when our position on Page 1 dropped from top to bottom. One theory is that Google is basing impressions on screen resolution, screen size and user behavior — i.e. no scroll means no impression count for bottom of page. At any rate, impressions have been a correlated with dramatic changes in performance.
Even though we don’t know the full basis of “impressions” and they don’t provide a lot of value in consultants dialogue with clients, impressions are part of click-through rate, which can be an important indicator of whether titles and descriptions are effective.
Excellent willing analytical eye just for fine detail and can
anticipate problems before these people happen.
And I think that Gyi Tsakalakis there is spot on: we all know which metrics matter, we just need to connect the dots and represent them to clients in the right way.
I am glad that the impression data is not so important, but I’m still wondering why it dropped for our website. It seems to have done so at the time we started a blog on the subdomain. Does anyone have any idea what might have happened there?
Comments are closed.