Introducing a new (and simpler) Universal Ranking model

Things used to be a lot simpler. Google would just serve up approximately 10 blue, organic links per page (plus some Ads at the top), so to work out your organic rank, you simply had to see how far down that list of links you were.

The Challenge with SERPs

However, since Google started adding additional features to the SERPs (Search Engine Ranking Pages) intermingled with these blue links, some result types took up more vertical ‘real estate’ in the SERPs than other types and pushed organic results even further down the page. This meant that just talking about a site’s organic ranking didn’t reflect reality and it meant SEOs needed to come up with a logical way of working out what became known as a “universal rank” for each keyword. This then became increasingly complex with the introduction of features that hid results behind further user interaction (e.g. having a user navigate through a carousel or click ‘See More’ buttons). The mix of displayed SERPs features also began to reflect Google’s interpretation of a searcher’s user intent, which is one reason why we recently added User Intent in our SERPs API.

We have always reported on the SERP features that appear for any keyword you are rank tracking and have indicated where you rank for that feature.  But until today, this has not extended to showing you the ranking URL for a SERP feature and awarding that Universal SERP feature block a Universal Ranking position.

After consultation with many clients we believe we can tweak our ranking model to make it better reflect the reality of competing in the SERPs today.

What are we changing?

For clients with “Pure Rank” keyword ranking configurations there is no change as these projects will only look at your organic rankings (just the 10 blue links)

For clients with “True Rank” keyword ranking configurations, we are now reporting on your ranking positions for these Universal SERP features (in instances where Google displays a ranking URL) and showing you the ranking URL.  This enhanced view means you you will now be able to see:

  • An overall rank for a keyword which takes into account all Universal SERP features that we are tracking
  • The ranking URL (where Google displays one) for your Universal SERP feature
  • A more accurate Universal Rank.  We count each Universal SERP feature as a block, e.g. The Image block may contain 5 to 15 images but it only occupies a single Universal Ranking position in our model (otherwise the first organic ranking page could rank 16th in edge cases!)

This may cause slight fluctuations in your ranking reports if the keyword you are tracking has universal SERP features above your organic ranking.  But we feel that these ranks now better reflect the reality of what is happening in the SERPs on a day-to-day basis.

N/B: Other non-Google search engine configurations are not affected by these changes.

Simple Example:

Let’s say you were tracking the keyword “Katy Perry” and you ranked as the first organic result behind an image block and a video result.

Before this change: 

Your organic rank = 1 and we highlight that the keyword generates images and videos.

After this change: (With ‘True Rank’ configured)

Your universal rank = 3, as Images rank 1, Videos rank 2 and then your organic rank = 3.  But we can also tell you if your ranking URLs are appearing for Images and/or Videos – if they are, then your universal rank could be even higher.

After this change: (With ‘Pure Rank’ configured)

No change.  Your organic rank = 1.

Why is it critical to get this right?

Because if you don’t get this right (or at the very least take a consistent approach), there is danger that you could be reporting on a brand’s SEO visibility incorrectly. 

For example, you could easily over or under report on a site’s SEO visibility or potential visibility if you ignore certain SERP features like Featured Snippets, Videos or Image blocks, or assign them ranking positions that might not make any sense. Furthermore, just looking at organic ranks nowadays would be completely misleading, as in many examples and especially on mobile devices, the #1 organic ranking can often be below the fold of the page.

Possible solutions (models)

In deciding how to approach this issue, there are several questions we needed to consider, such as:

  • What result types should be included in the ranking model for this result?
  • Which ones do you ignore?
  • If an image block of 5 images appears does it occupy one universal ranking position or 5?  What happens if it contains 15 images?
  • Should you count SERP features that only ever show Google URLs that you cannot even optimise to try and rank for (e.g. Places, Job Finder, Destination guides, etc)?
  • What about features that are initially hidden to the user like carousels; should they count?
  • What about the Knowledge Panel in the right-hand column? Should that be ignored or counted after everything in the left hand column?

We evaluated several ranking models and there certainly is not an industry standard agreed approach to this issue.  E.g. Google Search Console has a similar approach and counts the second column after everything in the left hand column has been counted – see: What is search result position?) and most SEO platforms only count organic rankings and show you the SERP features for each keyword.If you really want to understand the pros and cons of every possible model then ask your customer success manager to set up a call with our product management team and we’d be happy to take you through it.

But, let’s look at the ranking model we chose using the query (“how to cook an egg”) to illustrate how this now works in practice.

Our new Universal Ranking “model”

How to cook an egg serps

Look at the orange numbers in the image above and you can see that each different Universal SERP feature is given a Universal Ranking position irrespective of how many different objects it contains.  The first organic listing is actually the 6th Universal Rank.

So in order to calculate the universal rank for any particular keyword and domain combination, we’ve also decided it will follow these rules:

  • Paid results (Ads and Shopping) are ignored altogether in the model, although icons for them will show up in the ‘SERPs Features’ column in the Keyword Ranking module
  • The model is mostly focused on reflecting result types that offer a domain a chance to rank, but for these result types, we will group them together and give them all the same rank (i.e. Find Results On, Image Packs, Local Packs/Places, People also ask, Podcasts, Recipes, Research Guides, Reviews, Top Stories, Tweets and Videos – in individual results or carousels).
  • For large Google-owned SERPs features that take up a lot of space, we have also grouped them into a single universal rank. i.e. Answer Boxes, (Stock Prices, Currency Converters, Weather cards, etc), Event Finder, Destinations/Travel Guides, Job Finder, Flight Finder, Hotel Finder, Popular Products (mobile only) and Travel Finder.
  • We skip result types that only produce another search query, since you cannot optimise to rank for these features, e.g. Top carousels, People Also Search For, Related Searches and Refine By.
  • Sitelinks are considered to be part of an organic result, so will not be assigned their own individual universal ranks
  • For desktop results, we ignore the Knowledge Panel and Local Business Listings (effectively an enhanced Knowledge Panel with additional local business details), if they appear in the right hand column (for mobile results, there is only one column so a Knowledge Panel result will be given a top ranking position if it appears).
  • We record results that were in packs or carousels but not immediately visible to a user as having a rank for that particular result type (a ‘Result Type Rank’) but unranked for the purpose of calculating a universal rank; or, in other words, we only consider results that were immediately visible to a user as having a universal rank.

How will it affect my projects and reports?

  • Existing users of what was previously branded the Linkdex platform should not notice much difference, as this ranking model is more in line with what it was using before, i.e. Ads and Shopping were ignored and we were grouping a lot of result types (image packs, e.g.) which will remain grouped under this new model. Any specific queries or concerns you may have can be discussed with your account manager.
  • Existing users of our old Authoritas platform will not see any material difference in their organic rankings, but may see some changes in their universal ranking positions.  This is because the model used on the legacy platform gave each object in a pack or carousel its own, consecutive universal rank.

N/B: You can still report on organic rankings only by creating a separate ‘Pure’ ranking configuration.  But we would advise you to wait before doing so and incurring additional keyword rank credits because as you can see from the green, pink and purple numbers in the image above that we will introduce a ‘Result Type Rank’ filter in the near future.  This will give you the best of all worlds, as you will be able to see your overall Universal Rank and see how you rank for every different type of result.

Impact on Visibility

A knock-on effect of the above model change is that some universal result objects which are visible and grouped now share the same universal ranking position. 

Obviously, it would be absurd to assume that every result in an image pack, for example, received say 18% CTR if the image pack had a high ranking position. 

However, without a detailed forensic study into CTRs for image packs, we would be left guessing as to whether the one in the top left hand corner of a large image pack typically received more click throughs than one bang in the middle or towards the bottom right hand corner of the pack. 

For simplicity’s sake, we therefore decided to split the typical CTR in such circumstances. So, here, for example, we have an image pack for “katy perry images” ranking at the top of the SERPs (Ads are ignored for universal ranking calculations):

Now, there are 17 images in this pack, so if the typical CTR was 18% for a this universal ranking position, we’d simply divide 18 / 17 and assign ~1% of the typical CTR to each ranking URL in this pack.

Update to Pure Ranking Configurations: 24th November, 2021:

We’ve since tweaked this model slightly to include Featured Snippets in Pure Ranking configurations on the platform, as these are effectively Organic results that have been promoted by Google, so now if you configure a project with a Pure Ranking configuration and you rank for a top Featured Snippet, this will now have a Pure Rank of 1, as illustrated below:

Let us Know If You Have Any Feedback

Obviously, we’ve had to make our own call on this issue and we hope you find the model easy to follow, but would welcome any feedback in the comments. Should you wish to apply your own Universal Ranking Model to the SERPs results, one option open to all our customers is to make use of our SERPs API which returns both the HTML and a parsed JSON of each result. Once you have that, you can make use of any model you design and order the ranks in any way you see fit.

Should you wish to apply your own Universal Ranking Model to the SERPs results, one option open to all our customers is to make use of our SERPs API (query fees apply) which returns both the HTML and a parsed JSON of each result.   You can send the data to a database of your choice, for example, Google Big Query which connects to Google Data Studio.  Once you have that, you can make use of any model you design and order the ranks in any way you see fit.

Featured image by Nathana Rebouças on Unsplash

Leave a Reply

Your email address will not be published.

Fill out this field
Fill out this field
Please enter a valid email address.