Measuring SEO Performance After "Not Provided"

In recent years, the biggest change to the search landscape happened when Google chose to withhold keyword data from webmasters. At SEOBook, Aaron noticed and wrote about the change, as evermore keyword data disappeared.

The motivation to withold this data, according to Google, was privacy concerns:

SSL encryption on the web has been growing by leaps and bounds. As part of our commitment to provide a more secure online experience, today we announced that SSL Search on https://www.google.com will become the default experience for signed in users on google.com.

At first, Google suggested it would only affect a single-digit percentage of search referral data:

Google software engineer Matt Cutts, who’s been involved with the privacy changes, wouldn’t give an exact figure but told me he estimated even at full roll-out, this would still be in the single-digit percentages of all Google searchers on Google.com

...which didn't turn out to be the case. It now affects almost all keyword referral data from Google.

Was it all about privacy? Another rocket over the SEO bows? Bit of both? Probably. In any case, the search landscape was irrevocably changed. Instead of being shown the keyword term the searcher had used to find a page, webmasters were given the less than helpful “not provided”. This change rocked SEO. The SEO world, up until that point, had been built on keywords. SEOs choose a keyword. They rank for the keyword. They track click-thrus against this keyword. This is how many SEOs proved their worth to clients.

These days, very little keyword data is available from Google. There certainly isn’t enough to keyword data to use as a primary form of measurement.

Rethinking Measurement

This change forced a rethink about measurement, and SEO in general. Whilst there is still some keyword data available from the likes of Webmaster Tools & the AdWords paid versus organic report, keyword-based SEO tracking approaches are unlikely to align with Google’s future plans. As we saw with the Hummingbird algorithm, Google is moving towards searcher-intent based search, as opposed to keyword-matched results.

Hummingbird should better focus on the meaning behind the words. It may better understand the actual location of your home, if you’ve shared that with Google. It might understand that “place” means you want a brick-and-mortar store. It might get that “iPhone 5s” is a particular type of electronic device carried by certain stores. Knowing all these meanings may help Google go beyond just finding pages with matching words

The search bar is still keyword based, but Google is also trying to figure out what user intent lays behind the keyword. To do this, they’re relying on context data. For example, they look at what previous searches has the user made, their location, they are breaking down the query itself, and so on, all of which can change the search results the user sees.

When SEO started, it was in an environment where the keyword the user typed into a search bar was exact matching that with a keyword that appears on a page. This is what relevance meant. SEO continued with this model, but it’s fast becoming redundant, because Google is increasingly relying on context in order to determine searcher intent & while filtering many results which were too aligned with the old strategy. Much SEO has shifted from keywords to wider digital marketing considerations, such as what the visitor does next, as a result.

We’ve Still Got Great Data

Okay, if SEO’s don’t have keywords, what can they use?

If we step back a bit, what we’re really trying to do with measurement is demonstrate value. Value of search vs other channels, and value of specific search campaigns. Did our search campaigns meet our marketing goals and thus provide value?

Do we have enough data to demonstrate value? Yes, we do. Here are a few ideas SEOs have devised to look at the organic search data they are getting, and they use it to demonstrate value.

1. Organic Search VS Other Activity

If our organic search tracking well when compared with other digital marketing channels, such as social or email? About the same? Falling?

In many ways, the withholding of keyword data can be a blessing, especially to those SEOs who have a few ranking-obsessed clients. A ranking, in itself is worthless, especially if it’s generating no traffic.

Instead, if we look at the total amount of organic traffic, and see that it is rising, then we shouldn’t really care too much about what keywords it is coming from. We can also track organic searches across device, such as desktop vs mobile, and get some insight into how best to optimize those channels for search as a whole, rather than by keyword. It’s important that the traffic came from organic search, rather than from other campaigns. It’s important that the visitors saw your site. And it’s important what that traffic does next.

2. Bounce Rate

If a visitor comes in, doesn’t like what is on offer, and clicks back, then that won’t help rankings. Google have been a little oblique on this point, saying they aren’t measuring bounce rate, but I suspect it’s a little more nuanced, in practice. If people are failing to engage, then anecdotal evidence suggests this does affect rankings.

Look at the behavioral metrics in GA; if your content has 50% of people spending less than 10 seconds, that may be a problem or that may be normal. The key is to look below that top graph and see if you have a bell curve or if the next largest segment is the 11-30 second crowd.

Either way, we must encourage visitor engagement. Even small improvements in terms of engagement can mean big changes in the bottom line. Getting visitors to a site was only ever the first step in a long chain. It’s what they do next that really makes or breaks a web business, unless the entire goal was that the visitor should only view the landing page. Few sites, these days, would get much return on non-engagement.

PPCers are naturally obsessed with this metric, because each click is costing them money, but when you think about it, it’s costing SEOs money, too. Clicks are getting harder and harder to get, and each click does have a cost associated with it i.e. the total cost of the SEO campaign divided by the number of clicks, so each click needs to be treated as a cost.

3. Landing Pages
We can still do landing page analysis. We can see the pages where visitors are entering the website. We can also see which pages are most popular, and we can tell from the topic of the page what type of keywords people are using to find it.

We could add more related keyword to these pages and see how they do, or create more pages on similar themes, using different keyword terms, and then monitor the response. Similarly, we can look at poorly performing pages and make the assumption these are not ranking against intended keywords, and mark these for improvement or deletion.

We can see how old pages vs new pages are performing in organic search. How quickly do new pages get traffic?

We’re still getting a lot of actionable data, and still not one keyword in sight.

4. Visitor And Customer Acquisition Value

We can still calculate the value to the business of an organic visitor.

We can also look at what step in the process are organic visitors converting. Early? Late? Why? Is there some content on the site that is leading them to convert better than other content? We can still determine if organic search provided a last click-conversion, or a conversion as the result of a mix of channels, where organic played a part. We can do all of this from aggregated organic search data, with no need to look at keywords.

5. Contrast With PPC

We can contrast Adwords data back against organic search. Trends we see in PPC might also be working in organic search.

For AdWords our life is made infinitesimally easier because by linking your AdWords account to your Analytics account rich AdWords data shows up automagically allowing you to have an end-to-end view of campaign performance.

Even PPC-ers are having to change their game around keywords:

The silver lining in all this? With voice an mobile search, you’ll likely catch those conversions that you hadn’t before. While you may think that you have everything figured out and that your campaigns are optimal, this matching will force you into deeper dives that hopefully uncover profitable PPC pockets.

6. Benchmark Against Everything

In the above section I highlighted comparing organic search to AdWords performance, but you can benchmark against almost any form of data.

Is 90% of your keyword data (not provided)? Then you can look at the 10% which is provided to estimate performance on the other 90% of the traffic. If you get 1,000 monthly keyword visits for [widgets], then as a rough rule of thumb you might get roughly 9,000 monthly visits for that same keyword shown as (not provided).

Has your search traffic gone up or down over the past few years? Are there seasonal patterns that drive user behavior? How important is the mobile shift in your market? What landing pages have performed the best over time and which have fallen hardest?

How is your site's aggregate keyword ranking profile compared to top competitors? Even if you don't have all the individual keyword referral data from search engines, seeing the aggregate footprints, and how they change over time, indicates who is doing better and who gaining exposure vs losing it.

Numerous competitive research tools like SEM Rush, SpyFu & SearchMetrics provide access to that type of data.

You can also go further with other competitive research tools which look beyond the search channel. Is most of your traffic driven from organic search? Do your competitors do more with other channels? A number of sites like Compete.com and Alexa have provided estimates for this sort of data. Another newer entrant into this market is SimilarWeb.

And, finally, rank checking still has some value. While rank tracking may seem futile in the age of search personalization and Hummingbird, it can still help you isolate performance issues during algorithm updates. There are a wide variety of options from browser plugins to desktop software to hosted solutions.

By now, I hope I’ve convinced you that specific keyword data isn’t necessary and, in some case, may have only served to distract some SEOs from seeing other valuable marketing metrics, such as what happens after the click and where do they go next.

So long as the organic search traffic is doing what we want it to, we know which pages it is coming in on, and can track what it does next, there is plenty of data there to keep us busy. Lack of keyword data is a pain, but in response, many SEOs are optimizing for a lot more than keywords, and focusing more on broader marketing concerns.

Further Reading & Sources:

• Occams Razor: Search Not Provided
• Four More Ways To Crack The Not Provided Code
• Easing The Pain Of Keyword Not Provided