We've been experiencing an interesting pattern in our branded organic traffic over the last few months. I know SEOmoz can't be the only ones experiencing this trend, so I want to call out what I've been seeing in the SEOmoz data.
We've been seeing gentle but steady organic growth in 2012, along with a small seasonal dip in early summer:
However, when we look at just our branded organic traffic, we're seeing a different story altogether:
Branded organic visits are taken from an Advanced Segment I've set up in Google Analytics. It captures any organic traffic that comes in via keywords including our brand terms (seomoz, open site explorer, etc.) and variations on our brand terms (seo moz, seo mox, OSE, etc.).
Digging into this data a bit, I compared visits from April 2012 (the first available 30-day month of the year) with September 2012 in GA and got the following results for our top four branded terms:
- "seomoz" declined 26.26%
- "open site explorer" declined 37.04%
- "opensiteexplorer" declined 28.16%
- "seo moz" declined 33.10%
Is interest in our brand declining?
I was pretty sure that the decrease in branded traffic wasn't a decrease at all. Instead, our drop in branded keyword tracking was a casualty of Google masking keyword data for some users.
However, I needed to make sure we weren't losing brand equity. Reduced search volume for our branded terms would be a bad sign for us. I put together a test to make sure our branded organic traffic (probably) wasn't actually declining.
I took a look at Google Trends data for the four terms listed above and found that while some of them have seen some volatility in interest over the course of 2012, none of them has seen a significant decline in search volume when comparing April to September (note the drop at the end from this week's incomplete data).
The term "seomoz" remains very steady:
"open site explorer" saw dips in interest in April and July, but had returned to the 80-100 level by September:
"seo moz" has seen the most significant decline, but by removing the three outlying peaks from this cart, we can see interest remaining fairly steady (especially since July):
"opensiteexplorer" has actually seen an increase in interest since late July:
For one last sanity check, I exported our rankings history from 2012. I was pretty sure I'd have noticed if SEOmoz properties had slipped from #1 for these terms, and sure enough, they haven't.
What is going on with our branded traffic?
I'm confident that I cracked the case in regards to our branded traffic. If search volume hasn't declined and we are still ranking the same, it's a reasonable assumption that our branded organic traffic has not, in fact, fallen off.
The culprit is our old pal, (not provided).
In the same period that we saw the decline in branded traffic listed above, we also saw a 42.02% increase in (not provided) traffic. In September, (not provided) accounted for 63% of our organic search traffic, compared to 51.7% of our organic traffic in April. Remember when (not provided) was only supposed to affect 5% of searches? That was fun.
Since (like most sites) our branded terms are also our most popular overall organic terms, it stands to reason that a large portion of that (not provided) traffic is made up of branded organic traffic. SEOmoz is harder-hit by this than some other sites because we have such a tech-savvy audience: our users are more likely than some other demographics to be using Firefox or signed in to Google Accounts.
What kind of increase have you seen in (not provided) traffic since the beginning of the year? Is it affecting your branded terms?
If (not provided) really is to blame then this is going to be another worry for both agencies and in-house marketers.
Showing a declining graph to your client/Director but explaining that everything's actually ok because "(not provided) data is taking away from other areas" is probably not going to fly with most, particularly those who are less tech-savvy.
It might also be the go-to excuse/reasoning at times when there might be another issue that's responsible, e.g. an actual drop in branded organic traffic, but (not provided) gets the blame.
Yet more headaches for SEOs/marketers! At least no one can ever say that our jobs aren't challenging! :-)
I think the best thing to do to reassure clients is to show them that overall traffic/revenue isn't declining. That said, it's still super frustrating.
I agree that it's dangerous to blame (not provided) without double-checking that branded traffic hasn't actually declined.
Absolutely. Client/management reassurance and education is difficult enough in our industry without the added worry of telling them that the graphs - visual data that they like to peruse and analyse - might be 'lying' to them as well, hehe!
I think sites about SEO will see this impact more than general websites. Danny Sullivan tweeted that over half of the Google search traffic to SearchEngineLand was not provided. When I look at sites not about search the % of "not provided" is much lower.
In most cases if the brand is aggressively marketing then the growth in exposure & the general increase in search volume will most likely more than compensate for the "not provided" issue (at least up until Chrome also defaults to "not provided"). And then after that happens the decline will still only be a single year over year shift & then from then on the same growth patterns will show going forward, so this won't be a recurring problem for most folks, but rather a one-off shift.
Also, as people generally get more comfortable with search people tend to use the search box for navigation more for many brands, shifting from direct traffic to traffic that flows through shift. Though one issue that is likely to cause a greater decline in branded searches for some of the bigger brands is going to be the rise of apps. If eBay or Amazon get to where 25% of their purchases are on mobile devices & most of that is through their own apps, that will mean the brand search signal associated with that segment of their business is not going to be reflected in Google's data, as their app users are effectively removed from the general web search ecosystem when they use the app.
Excellent point about apps and branded search, Aaron - that's going to be interesting to watch.
Purely anecdotal, but it's interesting to note that MozCast, which is arguably a site purely for the search industry, is showing [not provided] at a whopping 64%, whereas one of my clients with a long-time B2C site is at 18%. Granted, though, MozCast organic traffic is almost entirely branded. Definitely seeing wide variability across niches and customers.
That's very interesting. I wonder if the 18% is leaving a dent in their branded organic traffic (even if it's just a small one) or if that's too small a percentage to have a noticeable effect? Sounds as though you're most definitely feeling it with the 64% for MozCast though - ouch!
(not provided) of Google is frustrating industry people more and more. Recently Danny Sullivan has tweeted that Google said user privacy is the reason behind (not provided) then why (not provided) is not in adwords traffic data. If that data is so important for user privacy then it should not be disclosed anywhere.
Great points, Aaron - especially about the apps, as Ruth's already said.
It makes sense that sites about SEO are worse off and that most other industries are in less of a predicament, but I wonder if that'll be the case in 6-12 months or beyond? (not provided) is certainly squeezing harder and harder as the weeks and months go by, more than Cutts suggested originally and I guess more than many of us would have anticipated. I'm not usually a pessimist but I fear that it's only going to get worse.
That said, this is exactly why articles like this one are so valuable - we're aware of the potential issue now, in advance, rather than when it affects us first-hand further down the line.
Yes, Aaron.
I am in tourism seo here in Germany. I can sse an increasing from July with 10% of (not provided) to october 21,8 % on a client site.
I think, this situation is increasing, since people are joining Google+
Using Google+ you need to be logged on, getting your personalized filter bubble.
Take care,
Just a question... did you cross checked also with the queries estimated volumes in Google Webmaster Tools?
Said that, considering that Chrome when logged is not passing referrals, that Firefox doesn't pass them to, as well as Safari... the not provided is going to explode and not just for tech niche sites.
I saw a similar situation in tourist portal.
Finally... have you seen the arise of "new" branded searches, like "Moz"?
I think Gianluca has a good question, are you seeing an increase in traffic for "Moz"? Especially since the addition of Rand's personal blog at moz.com/rand? Thanks for the update btw!
We have seen an increase in traffic for "Moz," but it's a small fraction of our overall organic traffic - it accounted for just 272 visits in September (up from 75 visits in April).
I did check the queries' estimated volume in GWT, but since that data only goes back to mid-July it wasn't as useful. I didn't see a significant drop there either, though.
Hi Ruth, interesting post. I've been doing quite a bit of digging around on this subject recently. There's a good article over on search engine land with resources for advanced filters>> https://searchengineland.com/turning-not-provided-into-useful-actionable-data-135800
and
https://www.customreportsharing.com/topic/16-see-all-organic-searches-with-not-provided-filtered-out/
but, although useful, they will not be a perfect fit for everyone's analytics reports.
yeah I guess thats a valid point !
What I find particularly frustrating is trying to use maths to predict the breakdown of (not provided) traffic (brand, non-brand, etc.) when the percentage gets so incredibly high. When 40%+ of search traffic has its keywords hidden, I don't feel great making guesses about its composition based upon the 60% left over, even if those guesses are based on data analysis where I have data.
45% is the worst I've seen lately, and that was spread over the course of many months. Other sites I have a lot to do with are in the 20% range, and it grows - slowly - week on week.
Would make sense to switch to another analytics solution just for this reason...IMO.
Because Google is encrypting the referral data, switching to another analytics solution would not help with this problem.
Sad to say, but the days of keyword level analytics metrics are already over. Not provided has just totally gutted the data.
I've already moved to reporting SERP visibility changes + adjustments in natural search traffic to the landing pages I've actively built/tweaked/added content to/etc.
Great analysis and thanks for sharing, I have just check out the keywords ranking in Google which you mention as a SEOMOZ branded keywords. All have doing well except"OSE" its on 8th position on Google.Com so might be lookout for that. And yes after Google Introducing secure search its very difficult to collect keywords data or analysis through Google Analytic. also as klarke mention many iOS 6 searchers in Safari are appearing as direct hits so it's hard to calculate purely organic traffic.
In addition, I have notice that for many brands name competitors adds will come up in SERP which may be the one of the major point to decrease brands keyword traffic.
Here is the proof
I was actually just preparing a presentation for client reviewing their search trends for 2012. I saw the same exact trend with their branded terms (they are a big player in enclosures and data racks).
When i started to pull the report i was thrilled to see they had a 28% increase in organic traffic over the year. However when i dug deeper their branded searches were dropping at a consistant rate and the non-branded exploded (almost doubled 5k to 10k a month).
My only question to the logic is how are non-branded keywords an exception to the (not provided) pitfall? Wouldn't that include any search query?
Cheers - Kyle
I guess they are not, I think it would probably just depend on the proportion of traffic that a site gets via branded keywords vs. non branded keywords (and in SEOMOz's case, how likely those searchers are to be logged in - as search savvy folk are likely to be logged in to their Google account when they quickly search in Google to navigate their way to Moz).
Generally speaking though I'd expect the proportion of users behind https to be consistent regardless of keywords, so if a site relies heavily on branded traffic as many good sites are then this could be the reason for the branded visits seeming to decline.
It still sounds like an assumption rather than fact? Wouldn't their non-branded traffic be coming from the same exact audience that you call "search savvy folk"?
I just don't see it as a causation. Maybe correlation?
You're correct - we're certainly seeing a decline in specific non-branded queries as well. However, since (not provided) falls into our non-branded segment, for our reporting purposes it doesn't appear to be declining. I'm certainly not saying that we haven't seen non-branded traffic affected by (not provided) as well. Since our branded terms are also our highest-volume terms, the decline is more noticeable. I'm sorry if that wasn't clear from the post.
Might it not be better to separate (not provided) out from both Branded and Non-Branded traffic, and then redistribute it amongst the two, based on the ratio of Branded to Non-Branded traffic?
Leaving (not provided) in the Non-Branded segment for reporting will be inherently problematic, because a lot of that traffic is actually Branded in origin (we just can't see it). This misattribution to Non-Branded traffic will thoroughly bias (and potentially invalidate) any perceived Non-Branded organic growth, particularly for a company with as much brand recognition as SEOmoz!
That is certainly another way to segment, and one that's worth looking into. Thanks!
It's definitely a bit more work, but I highly recommend it.
What a great idea allforJesse. Even as the % of "not provided" increases, the remaining data, as long as it is sufficient to be statistically significant, should provide an ample sample that is both clean and representative.
I wanted to tell you it's probably because they are increasingly shifting into the (Not provided) but I see you figured it out. :)
It's becoming so hard to deal with keyword data from analytics. The best thing you can do is have a look at Google Webmaster Tools data for your branded keywords.
Not provided + investigate number of direct visitors. I never type in "seomoz" except if I remember some old post I want to revisit - then I type some query + seomoz. As for GWC vs GA data - Google is becoming hard, I agree, but let's face it - this is their overall strategy to flame SEO :)
Kyle,
You'll find that unbranded traffic has been affected, however more than likely to a lesser extent in most instances.
If you are/have a brand, people will use navigation search queries like [seomoz] to return to the site instead of typing in the domain. In a lot of cases, you can/could group traffic off clearly brand terms like that as a direct visit and a search engine is just a familiar way for users to navigate the web.
Businesses that have a brand and especially a community built around that brand, will have users within that community visit the website (through whatever means if convenient for them) more often than a business that doesn't have a brand/community in general.
The important thing to consider in this scenario is that since SEOmoz already has a tech savvy audience, if a percentage of those users are also returning regularly to seomoz.org through navigational brand searches, it compounds the issue.
To illustrate that, consider two similar scenarios:
1) SEOmoz as you know it today
2) SEOmoz without the blog/community (ie, a pure software as a service)
In the latter example, you'd find that the percentage of (not provided) is going to be much higher than say a gardening website at 15% because of the technology focus. However in the former example, instead of having (not provided) at circa 40% (as an example) - the frequently returning community members using brand navigational search queries increase (not provided) into the 60% range.
Least that is what I'd expect to be happening.
Al.
Maybe I am being too cynical about Google.. but the creeping invasion of (not provided) is surely them clawing back what they have been giving away for free for so long. i.e. lovely valuable data. Mark my words, one day we'll be able to buy it back! Great post Ruth.
When I saw this headline, I was back at school again with my hand up saying "ooh ooh I know this!". 'Not provided' data has caught me out several times this year. I wrote about it last month: https://www.mediarunsearch.co.uk/blog/3-google-analytics-traps-caused-by-not-provided-keyword-data.html
The other major way it's caused errors for me is in creating the impression that long tail keywords are declining. Sites that had, say 5000 referring keywords at the start of the year will likely have fewer now because so many get grouped under 'not provided'.
I wonder if we're going to need to start reporting branded vs. non-branded as %s of total traffic, as opposed to raw numbers. In other words, look at how the brand/non-brand mix changes, assuming that [not provided] is spread across both groups. Of course, even that's a big assumption.
(not provided) is probably one of the most annoying things Google has released in recent times, I feel the more tech savvy your user base is the harder your analytics is going to be hit with a lack of data, I work across various niches so I notice the trends from specific niches. Lets hope Google wakes up one day and stops using it.
Hello,
Would it be something to do with Google Analytics... ? Do you have another analytic solution (i.e. Omniture) that seconds what you are saying ?
Hey remiturcotte, You will find that these issues will affect all other tools too, not just GA.
You're correct - since the information is encrypted when Google sends the referral string, not when Google Analytics receives the data, other analytics tools will be similarly affected.
Hi Ruth.
We noticed that this had started having a big effect around 6 months ago - we were giving increasingly inaccurate non-brand traffic and conversion/revenue reports because our reporting tool was pulling (not provided) in as a non-brand keyword.
Our solution was to take the KNOWN ratio of brand to non-brand traffic and conversions/revenue and apply that to (not provided) to factor it back in. It's not 100% water-tight, but it's the most accurate assumption we can make.
Sorry if someone else has already posted this - there were 67 comments before me and I couldn't read them all :)
Mike
Oh finally Rutt you answered a question for me… actually I am working for one of the giant real estate company (non US) and they are started to notice a dip in branded keywords traffic even when they are ranking #1 for all the keywords… this is was a disturbing question that you have answered..
Now when I am back in my Google Analytics data I have noticed that its not that my traffic towards the braded keywords has been lost but its (not provided) by Google that has increased significantly over the past few months…
I believe Google is going to push Google plus more and encourage people to sign up and stay sign in and this will impact the keyword data and we all will witness the increase in % of (not provided)
Sad but true! :(
Yes, in my opinion Google+ this is the main reason. The brand or non-brand discussion here is only a sideway, but interesting.
I wonder if the auto suggest has a little to do with this as well? I know that now I only need to get about 4 characters into the search for open site explorer before Google dishes up what I am looking for :)
Auto suggest visits should still count as search traffic, though. When you type in part of a keyword and Google shows you results for that full keyword, the keyword that will show up in your analytics will be that full keyword (that's why you don't see a ton of search referrals with just the first few letters of your keyword). Unless the keyword is (not provided), then that's what you'll see.
I'm wondering when Google will pull the keyword tool for non-advertisers. I feel like it has got to be coming soon with all of the other changes they've been making.
I'd like to suggest an alternative, or at least another factor that compounds the problem.
The suggest feature in the Google Chrome unified search/address bar suggests the URLs and search terms.
I haven't tested it, but I would presume that if you type seom and then click on the top suggestion https://www.seomoz.org/ it would count as a direct visit.
Because clicking on the suggested URL is easier than entering the name in search, getting the results and then clicking, I'd expect less branded traffic and more direct traffic as Chrome and others get better at predicting whether you want to go straight to a specific site, or whether to show search results first.
Also very interesting, why is Google suggesting the https URL for SEOmoz instead of http?
Visits from Google Instant/Google Suggest do count as visits from Google search. You'll simply see the keyword whose results were displaying in Instant as the referring keyword (if it's not blocked, then it would just count toward "not provided").
So in your example, if you typed "SEOm" into the search bar and it brought up search results for the keyword "SEOmoz," and you clicked on one, that would count as a search referral for the keyword "SEOmoz," not a direct visit.
I know a few people posited the same idea I have above but in different ways. One thing that comes with being a "bigger name" in the space is recognition. I have you guys on my bookmarks bar now. I know your FB likes have increased a LOT too so fewer people are searching for the latest updates. In addition, the more you get into someone's feed reader, the less they come to the site specifically to look for new blog posts.
There are a LOT of reasons (not provided) sucks and this is definitely one of them. But I would say there's a limit to the number of people who will search you and with every new Moz-aware person, they're that little bit less likely to run random Google searches for your brand.
Definitely multiple reasons but not a trend you want to see continue, for sure!
Sure, and that is always the hope with branded traffic, right? That eventually those searchers will become direct visitors. If we had seen a decrease in overall search volume for our keywords, that would definitely be among the hypotheses that I would test. Even as we bring on more wonderful Mozzers into our community, though, we're still working hard to build our brand and let more new people know about SEOmoz, which is why we haven't seen that decline in branded searches - and that's what we want.
The fact that you can see all the search terms if you are paying for them demonstrates this move towards removing data from GA is nothing to do with user privacy and much more connected to revenue. The fact we are all seeing a greater and greater portion of the data move to 'not provided' indicates the bleeding obvious...
I agree that it's an empty gesture. It gives Google something to point to to say "look we're protecting your privacy" while encouraging AdWords use.
yes, that is correct.!
I have been thinking about this move of Google since a while. Could it be that Google inc. wants that all companies should spent their money in SEM buying ads instead of SEO, in paid traffic from Adwords you do not have this problem.
I must say, this (not provided) nonsense is really getting out of hand. On some sites 65%+ of search engine traffic. What use are these "tools" if they don't help you, or only show you a part of the pie? I think that Google really should be more transparent with these kinds of things.
Ruth,
I had posted a question about this back in July in the forum. In my case we are an established website with an old domain. We have not penalized. However, for some reason we are seeing the drops in branded traffic but also in visits categorized as "direct" in GA. Is there any explanation for this? Up until this year we had seen flat or growth but we are seeing decreases of 20-30% on branded searches and the category "direct". Our sales are up and it is inflating our conversion rate (because of the reduction in traffic yet boost in sales) but it seems like there are many direct and branded visits missing from our GA reports. Any help is appreciated.
Hi Brad,
The first thing I would do is check in Google Insights or GWT to see if overall search volume for your branded terms has decreased (as I did in the example above). It could be that you're having an overall brand awareness problem, which would explain the concurrent decrease in direct visits. A branding/link building push would help there, although it's not really a quick solution.
Hi, i read the post and all comments. Thank you very much.
One question do you Ruth.
Do you want do a post how to do best practice SEO traffic and conversion report?
Or do you know some web resources for the most important KPI to include?
I am trying to move from the old Ranking Keyword SERP reports, and try to see more overall metrics.
Maybe you can ask your SeoMoz development team to include a section for (not provided) in the campaign management overview for Pro members?
Thanks,
Thanks for your suggestion - I'll pass it along to your Product team.
Interesting article.
Wondering if your direct traffic has increased. Reckon your brand recall has probably improved so users just enter the URL directly or your fanbase have saved bookmarks i.e no need to find you through organic search
I've not gone through the previous comments so this may have already been pointed out!
Asaad
Hi Asaad,
We have seen an increase in direct traffic over time - however, since we have not seen a concurrent decrease in organic traffic, and since our branded terms are still our top non-(not provided) terms, and we haven't seen a decrease in search query volume for those terms, I don't believe that is what's happening.
My guts feeling tell me that a big chunk of that not provided traffic was as you said branded terms. People looking for your brand might be returning visitors deeper in the search funnel (navigational queries) and to me it would make sense that these users tend to be more logged-in Google+/Gmail or even searching for the blog on iOS (6 not provided).
Have you try the same exercise but for long-tail queries only?
Unfortunately since long-tail queries have low search volume in general, it's more difficult to get accurate data over whether a decrease in query volume is to blame, or whether (not provided) is to blame for lower traffic on individual long tail queries. My hypothesis is that our branded and long-tail queries are actually being affected more or less equally, but we see the impact much more in our branded traffic since those are our most popular queries.
I went through very similar observations, your analysis definitely helped me find the root source of the problem, thanks. Made no sense to see such a huge drop on our brand keywords!
My analysis, here, in french, if it could help other members: https://duprogrammeur.com/baisse-anormale-de-trafic-mot-cles-5-mars-2012/#axzz2AuCi9O6u
Is there not a Google update (PANDA) thats is pooling you down on specific keywords?
Sites that were hit by Panda would see a decrease in traffic on certain terms, but since we haven't seen an overall decrease in traffic I'm pretty sure that's not what's going on here.
We have also seen this but assumed it was due to an increase in our PPC spending on brand terms and users clicking these rather than organic.
I've never seen a PPC campaign be that successful in cannibalizing organic clicks. In fact, in most hybrid PPC/SEO campaigns I've run, having a PPC ad on the page actually increased CTR on the organic results, in addition to getting clicks on the PPC ad.
Hi Ruth,
It was so nice to see this post and be able to totally trust what I thought had been going on with branded traffic for rmoov recently (phew!)
I was shocked a few weeks ago to find (not provided) running at above 79% for our site(!) and at the same time it seemed that branded traffic had dwindled from almost half down to a paltry 10% :(
After the initial shock (actually it felt more like devastation), it occurred to me that (not provided) might be the culprit. When I thought about it that way, not such a big shock...after all, since we are a link removal tool, it's fairly likely that a lot of people searching for us by brand have just been to GWMT to download their links and are still logged in!
Damned irritating, but totally understandable :)
I suppose in a weird way, the last thing we want to see now is a fall in (not provided) for the site!
Every time Google changes something lately I'm reminded of one of my favorite passages from Hitchhiker's Guide to the Galaxy -
Zaphod Beeblebrox: There's a whole new life stretching out in front of you.
Marvin: Oh, not another one.
So did anyone manage to find the exact reason behind this steep decline ? :)
Yes - as I suspected, it was due to (not provided) traffic.
Oh haha :) I was under the impression that "not provided" carried some sort of specific meaning with it, but now I understand.
The only one who still HAS complete keyword data is Google. In my view, this looks like creating (another?) monopoly.
As a countermeasure, we are calculating the share of branded traffic after excluding all (not provided) traffic. To make it simple, we therefore created a little excel tool that uses the GA API. You can also usa an advanced segment in GA to do this. Of course, the underlying assumption of statistical representativity is not reliable. As bigger the share of (not provided) gets, as "weaker" is our calculated value. When I see values > 50% it gives me stomach pains...
Great post! I manage two e-commerce sites with radically different products and target markets. One is a 40-year-old brand the other is less than a year old. For both sites "keyword" (not provided) is now above 30% in GA. I have noticed the same trend in what appears to be a decline in branded search, but I am fairly certain that the lion's share of our (not provided) searches are branded terms. Unfortunately, there's no way to prove that to stakeholders beyond conjecture, and conjecture doesn't pay the bills.
I wonder if Google realizes that their Google Analytics tool is rapidly becoming useless? I once thought they had a shot at monetizing it. But now, I don't know anyone who'd be willing to pay for it, as every month it reveals less and less about what's really going on.
Something's gotta give, don't you think? It seems like Google could go two ways with it at this point:
Neither seems like a viable option. The first choice intereferes with their privacy policy and the second would devalue their Google Adwords product.
What bothers me most is Google on the one hand gives us the directive to improve our Websites, but on the other continues to reduce access to the very data that helps us do that.
I include not provided as a separate entity in branded vs non branded keyword segmentation reports, and look at further segmentation, e.g. Head vs tail terms, within provided/non-brand.
Ultimately the biggest hit so far for us is the inability to attribute a large majority of iOS 6 organic traffic because of new secured google search implementation on safari. Given the increasing share in mobile this is a data killer.
The iOS 6 thing is super frustrating!
That's a good idea to look at not provided as its own segment - conversations with our commenters on this thread are making me consider doing the same.
When I still need to know if my keywords are doing what they should be doing - ranking-, I take all the "not provided" keywords and select the destination page as a secondary dimension.
I know it's not the same, but it might be actually helpful! At least you can almost guess what the keywords are! It's a helpful indication, I think.
I do something similar - if you know what keywords a page is ranking for, and you know how much traffic that page is getting, you can kind of back in to some data. The problem there is that it's all based on correlation and assumptions, so we have to tread carefully.
Does anyone else just search for something like 'rich snippets' and look for an SEOmoz link on the results page these days?
I rarely bother with SEOmoz specific searches (e.g. 'rich snippets seomoz' ) because I know their content will probably be on page 1 anyway (and I've just saved myself 7 extra keystrokes!)
Given SEOmoz's audience (techies who are hard-wired to notice trends) - isn't safe to assume that many of their 'regulars' have noticed the same trend I have, could this be a small part of the picture?
Well if nothing else, thank you for the nice compliment :-)
I hope that people are doing that! But I don't know that that behavior is going to outweigh people who use Google as their address bar, i.e. searching for SEOmoz rather than typing www.seomoz.org into the address bar. The behavior you're describing is more likely to affect the long tail, which makes the effects harder to measure.
Thanks for the info, Ruth. It clears up a lot of our confusion when looking at our analytics. Yes, we have noticed the same trends and now it all makes more sense.
Maybe you can clarify something for me - I've posed this question to a number of people and have not gotten a clear answer. Why does Google mark so many of the results as "not provided"? I don't see what their angle is.
Thanks, and thanks for another great post from the Moz team!
Since https is supposed to signify a "secure" search, Google is masking keyword data from analytics users as a gesture to protect user privacy.
In my opinion, it's something of an empty gesture since a.) it was impossible to tell who searched for what term in GA to begin with and b.) keyword data is still visible to AdWords advertisers, but it allows Google to have something to point to when people ask how they're protecting searchers' privacy.
That (not provided) share is increasing even in our small sites...and it's really starting to get annoying and frankly really difficult to make decisions around. Not going to be much longer before clients start making stuff up because we can't prove it with data.
Glad we're not the only ones - and glad you're not losing people's interest ;-)
Interesting data.
On my Google AdWords blog at www.onlineeffekt.dk (in danish only - sorry!) roughly 65% of all organic traffic is recorded with "not provided" as keyword. It's not really that surprising given my contents heavy focus on Google stuff - it's however still pretty annoying that I'm losing track of how people find my blog.
On our portal for sharing of Google Analytics dashboards at www.dashboardjunkie.com roughly 55% of all organic visits are hidden due to Googles "privacy concerns".
I've noticed this across quite a few different sites. The greatest problem I see is in reporting for progress, because of "(not provided)" YoY non-brand traffic is seen as growing, but as you've pointed out Ruth, this isn't the whole story.
One of the biggest examples is a 40% rise in non-brand traffic, however 60% of that rise is because of "(not provided)", and branded traffic is down 30%.
Clearly that suggests there is an overall rise in non-brand, however, how much of that is branded terms creeping in? But beyond conjecture, there's nothing concrete you can really add to this.
Another thing ...visits from many iOS 6 searchers in Safari are appearing as direct hits. Not even "not provided", just direct hits with no referrals.
are you sure iOS 6 searchers in Safari are not referral?
Edit - sorry missed the bit in the post where you specifically mentioned the (not provided) traffic. *Goes and stands in the corner with a dunces hat on*
"Not provided" is far worse than I think Google ever intended it to be. I've seen some sites where 40% of the traffic was lumped under "not provided" which is a far cry from what was originally predicted. If you aren't looking for it and you see those numbers sliding it definitely makes your heart stop for a minute!
We're at over 60%. It's out of control.
FWIW - this mirrors our analytics very closely. I have exactly the same opinion you do...
That's reassuring, Will. Thanks!
Our (not provided) traffic hasn't spiked or experienced wild fluctuations, but it has been slowly edging its way toward 10% of organic traffic. It had about 8% in July, 9% in August, and is sitting close to 7% for October MTD. Our visitors overwhelmingly favor Internet Explorer on Windows, so the impacts of Chrome & Safari will show less in our results. A quick check of September numbers shows that 43% of (not provided) organic visits came from Firefox and 30% from Chrome. Overall, 21% of our users used Chrome and 14% used Firefox.
Thanks for the TAGFEE numbers sharing! I think it's important for us all in the SEO community to be discussing the degree to which this is affecting our sites.
It's not just branded keywords...it's ALL keywords.
A prime example of this is how we pushed a client from #3 to #1 for a MASSIVE keyword, but when we compare that keyword year over year, it looks like we only increased traffic by 10-15% for that keyword specifically.
Also, to answer your question about an increase in (not provided) traffic in 2012, highlyrelevant.com has gone from 38% of our SEO traffic being (not provided) the month of January to 61% of our SEO traffic being (not provided) the month of September...
Yes, it's definitely all keywords. My point was more that we see the impact more in our branded terms since they're some of our top traffic-drivers. Thanks for sharing your increase numbers - we're over 60% not provided as well!
As someone who's only recently been getting into SEO this seems troubling. Since it seems like this is affecting a much larger portion of the traffic than Google originally expected, do you think they will do anything to lessen the impact of this? Or is this a new reality that we all have to learn to live with?
I doubt they'll be able to do too much to roll it back without looking like they're compromising their privacy stance. I think it's a reality we're going to have to learn to live with.
(not provided) - Googlers' reason for evil cackling Monday - Friday.
(not provided) has been my number 2 keyword term for the last 4 months now. When are they going to roll out with the paid version of analytics that will show the keywords? I miss 2010...
It's our number 1 keyword, and has been for some time. I would gladly pay for Analytics to get this data, and to get a dedicated support rep!
I'd usually test my intuition - probably (not provided) and potentially (not set). Usually pretty spot on since when you work as an inhouse SEO, you know where all the skeletons are hidden in which closet. (Seasonality is something that I'd also look into if it were say, Christmas. Otherwise, it's rare to not know why.)
If not, I would've delved into GWT first and not Google Trends or SERP checks. Personal preference here, but it's nice to see the broad approach here.
Won't "not provided" hurt the Web, Matt? I hope that question makes the cut, and gets answered soon in Matt Cutts' next round of videos.
In my view, more accurate data helps website owners make better sites, which then benefits visitors.
And if it's a privacy thing (**cough, cough**), then why does Google's own Chrome browser in Incognito mode not use https while searching?
Thanks for the post Ruth, and to echo comments above I have found the same with a couple of clients I'm working with. Comparing the same period as you did (apr/ sep)
Client 1
Organic branded traffic has decreased 25% with "not provided" visits increasing by 66%
Client 2
Organic branded traffic has decreased 0.52% with "not provided" visits increasing by 122%!
lately noticed a large growth in organic traffic the site iTrabalhistas
great posting wow i like Google Analytic s & blog Thank you