I am going to be clear from the beginning: in this post, you are not going to find patents, their schemes, and speculations about algorithms.
If you are looking for that type of content, I invite you to visit SEObytheSea.com and the Go Fish Digital blog and soak yourself in the immense legacy of the recently deceased Bill Slawski.
No, in this article I am going to ask you two questions, present my answers to them, and hopefully make you reflect on how you are thinking about SEO.
So what are these questions?
Here you have them:
- What if the classic funnel (Top, Middle, and Bottom of the Funnel or TOFU, MOFU, and BOFU for short) did not exist?
- What if the dreaded “0 click SERPS” were more the result of outdated strategic SEO thinking than anything else?
If it is true that we SEOs, including myself, love to analyze Google’s patents to try to discover how it will evolve, it is also true that a more reliable source when it comes to understanding where the Mountain View giant is aiming at its strategy is their Think with Google website.
Why? Because it is a site whose target audience is not SEOs or developers, but entrepreneurs and company marketing managers, the messages (recommendations? guidelines?) are explained much more clearly and cover all aspects of digital marketing and the Internet presence of a company or brand.
For example, today the importance of detecting, targeting, and optimizing the search intents of the content of our website is a dogma established in SEO. Likewise, today it is very common to evaluate an SEO strategy based on the concepts of E-A-T (Expertise, Authoritativeness, and Trustworthiness).
But, Google was already recommending it back in 2016, when it introduced the concept of Micro-Moments, which – using Google’s definition – are “moments during which a person uses a medium (eg, search engine) to know how to get to a site or get information or knowing how to do something or, finally, buy something”.
Always talking about Micro-Moments, Google recommended that the best way to respond to these search intents was to be useful with relevant content, which clearly shows expertise and authority, and which is “accountable”, that is to say, “responsible” and therefore that can be trusted.
Has Google come up with something new afterwards that we SEOs should have taken more seriously? Yes: the Messy Middle concept.
In 2020, Google released a nearly 100-page study, presenting what according to him, is the most correct way to analyze the behaviour of people when it comes to satisfying a need, from the moment that something triggers this need until the moment in which the person manages to solve it.
The study is especially focused on the need to purchase something, but its concepts are valid for any other type of need, such as learning to do something.
If the Messy Middle model is equivalent to how people behave when solving a problem, then companies must not only know what its phases are but also what to do in each of them.
However, when talking about search engines, everything must be translated into algorithms.
Let’s see, then, how the evolution of Google Search over the last 7 years has been a constant and unstoppable path towards “Artificial Intelligence”:
- RankBrain (2015)
- SyntaxNet (2016)
- Image Search Tags (2017)
- Neural Matching (2018)
- BERT (2019)
- Passage Ranking y Subtopics (2021)
- MUM (2021)
SyntaxNet, Image Search tags, Neural Matching, and BERT (but also RankBrain), are all algorithms whose purpose is either to better understand the real meaning of the searches performed by users or the entities/attributes of entities underlying these searches and, therefore, to be able to present the most pertinent answers to the implicit and explicit intentions that have given rise to the query in Google… as you can see: Micro-Moments.
Passage Ranking and Subtopics and MUM, on the other hand, seem rather be instruments with which Google tries to make navigation in the Messy Middle easier for its users.
Is Messy Middle the same as the classic conversion funnel?
The concept of funnel tells us that a person begins his journey towards conversion starting from a great blue sea of information and that he gradually descends into the funnel, restricting more and more the field of his searches until he reaches the one that leads him to convert.
The Messy Middle, on the contrary, tells us that, once a need is triggered, a person enters into a loop between the exploration phase and the evaluation phase, with the risk – for us – that she will remain trapped inside this loop or that in the constant reiterations of this loop our visibility vanishes.
If we think about it, the Messy Middle is the model that best represents the behaviour of people when looking for something and that is very reminiscent of the concept of “surfing the web” that was used many years ago.
At first glance, it might be easy to say that triggers and pain points are the same thing.
In part, it is so, but it would be reductive.
A trigger is something that makes us want to satisfy a need that maybe until that moment we didn’t know we had.
However, the classic Buyer Persona analyses are still essential to identify potential triggers. They should be serious analyses, though, and not the ineffective robot portraits with pretty names (Mónica, Ricardo, Cristina…) that we see so frequently. Rather, then, they should be studied based on psychographic analyses, interest analyses, influencers and sources of influence, trends, and seasonality.
Therefore, SEOs can not get involved in this study phase, and leave it entirely in the hands of other departments, such as Marketing Automation.
Triggers live in the realm of the Predictive, and this also implies a necessary collaboration and synergy of SEO with the other areas of marketing: digital and traditional, because only thanks to this collaboration will we be able to know what type of trigger can be activated, where, how. and why.
In the specific field of search, Google Discover is the place where the trigger is most likely to appear.
Discover is completely based on a predictive and algorithmic analysis by Google of the search history, geolocation, and interests of each user.
It is clear that our brand, with its value propositions, should try to be present on Google Discover with its content or, in the case that it lacks opportunities to create the type of newsworthy and fresh content that Discover requires, actively collaborate with those news and specialized websites that we have seen to be present on Discover for those interests, which are also related to our products, services, and brand.
In other words, the classic outreach and promotion work that is normally considered only for link building has much more value than just getting a link.
This work of relying on third-party websites to position ourselves and be visible in features that are very difficult for us to obtain is commonly defined as Barnacle SEO, and you will see it suggested other times in this article.
Another place where SEOs can work to be present in this phase of triggering interest and awareness, and, therefore, influence the beginning of a Search Journey, is YouTube.
Also, in this case, the combination of owned content (the videos on our YouTube channel) and Barnacle SEO (collaborations with Creators who have a real influence on our audience) is the right strategy to follow.
Finally, for a relatively short time, there is another area from which we can generate triggers that favour us: visual search.
As can be seen in the image below, it is now possible to perform visual searches with Google Lens when visiting a website with Chrome, which currently has a global market share of 64.95%.
Visual search with Google Lens within a site is practically the online version of the classic showrooming, that is, the phenomenon by which a person is in a store sees a product that interests her, and searches the Internet if there are options to buy it online for a better price.
The “real” Messy Middle
The Messy Middle is made up of two coexisting phases:
1. Exploration phase (expansive)
2. Evaluation phase (reductive)
It is a loop that feedbacks itself, and its keyword is “What’s next”.
The success of an SEO strategy is not only in the ability to make our site present as much as possible in all the loop iterations of the Messy Middle but also in being able with our results to break these same iterations so that our potential users pass to the Experience phase, which will take place on our website.
Google retains its users in the Messy Middle.
If we get conspiranoid, we could say that Google does this so to increase the exposure of these same people to its ads.
On the other hand, we could also accept its explanation and that it does so because it considers that this is the only way it truly can help its users find the most correct information and take therefore the best decision during their Search Journey. I consider both of them true.
In any case, our objective at the SEO level does not change: to work so that “all roads lead to our website”.
Messy Middle and Search Intent
It is during the Exploration and Evaluation phases that Search Intent analysis comes into play:
As seen graphically above, the Search Intent of the Exploration phase is eminently informational.
However, we must also be precise when defining what type of information a person is looking for because each type normally corresponds to a different content format:
- Theoretical: articles, posts, white papers, guides…
- Practical: video (classics and shorts), instructographics, “How To” type posts…
- Simple: short answers that, generally, we are not interested in targeting because Google answers them immediately to its users in its search results (eg: How old is Jennifer López?).
The Evaluation phase introduces the other Search Intents:
- Commercial or “I want to buy but I need help”, which is normally targeted with landing pages, reviews, and comparison-type content.
- Navigational or “I want to go”, both online (category pages and listings) and offline (Local Search)
- Transactional or pure “I want to buy” (product pages).
Messy Middle SEO Analysis Case Study
The SERP as a landing page
As maybe you are suspecting from some of the images that are being used in this post, I am a fan of Star Wars. I am passionate about painting miniatures, so I take advantage of these two passions to present to you how Messy Middle develops itself in Google Search.
So… Gianluca is interested in painting some Star Wars minis, and he knows that the best ones are from the Star Wars: Legion tabletop game.
So he grabs his phone, opens the Google app, and searches for “Star Wars: Legion minis.”
The first thing we can see from the immediately visible screen is the contextual Search menu.
This is a feature that is not normally taken into consideration in SEO, but, in reality, just by looking at it, it is very easy to understand what the search intents that Google considers those who search to have.
In our example, the intent is primarily transactional, which is why Google suggests Shopping as a more targeted search alternative.
But, this search, as always happens, does not have only one Search Intent.
A person may have the primary intent to buy something, but may first need some information before making a final purchase decision (Commercial search intent) and, in the case of Star Wars: Legion minis, visual search is essential (Images) as well as see opinions of experts in the miniatures niche (Video).
In addition, Google knows well how the fans always tend to look for fresh information (“Have new figures come out?”), and that is why it suggests News.
On the other hand, it also knows that some of those who carry out this search will ultimately prefer to go shopping in a physical store, and thus advance the potential navigational Search Intent by suggesting Maps in the menu.
Another important thing that we can see and consider simply with this first screen of the SERP is that we must think of the SERP itself as if it were a landing page of ours and, therefore, study and work so that we can benefit as much as possible from the possibilities that it can offer us.
In this sense, fantasyflightgames.com (the company that produces the game Star Wars: Legion) does very well:
- The first search result is from your product landing page, which, however, has been designed to also be a Pillar Page from which more specific content is linked, both informative and purely transactional.
This explains why the search result presents a very effective sitelink carousel that allows even better targeting of potential implicit searches in the rather generic query “Star Wars: Legion Minis”.
- Then, as a second result, we see an “Indented result”.
This is a page that, unlike the previous one, explicitly targets a purely transactional search intent.
What does this suggest to us? , if a query has more than one search intent, we can create different contents that target that same query as long as these contents do not also target the same search intent. In other words, the dreaded keyword cannibalization is partly a new SEO myth.
How Google itself tells us everything about the Messy Middle of a query, and we don’t pay attention to it
What we see above is something very feared by SEOs.
This is the most elaborate version of the Knowledge Panel and, in general, this is the feature that is usually considered among the main causes of the “0 clicks SERPs” phenomenon, that is, those search results that may not generate clicks. websites, but rather the searcher ends up clicking on either a Google property or another SERP.
This feature is the clearest map of the Messy Middle for the reference query… and it is Google itself that offers it to us!
Many elements owned by Google make up this specific Knowledge Panel.
In the case of our query (“Star Wars: Legion Minis”), we see both components extracted from Google Merchant/Manufacturer and others from Search and Video Search.
A necessary pause to talk about Google Merchant/Manufacturer
Considering how more and more Google Merchant/Manufacturer is being used by Google for product Knowledge Panels, it’s good to talk about its importance.
There are three factors on which we must concentrate when preparing the product feed that we will load in Merchant/Manufacturer:
- Data Sparsity, for which we must focus on using product attributes that are important for users when performing a search (eg: colour, material, type, brand…).
- Data Completion, for which we must pay attention that the structured data (but also the unstructured) is complete and consistent between feed and website.
- Data Quality, for which the reliability of the data is essential.
These three points also reveal the most serious problems when it comes to making a successful feed:
- Data Harmonization: the synchronization of the data between the web and the feed must be total, so much so that if data (eg: InStock or OutOfStock) tends to change very frequently, maybe it is good that it exists only in one of the sites to avoid very likely discrepancies.
- Data Reconciliation: Ambiguity is enemy number 1 at Merchant, and since substantially identical content can exist on a website, something similar can happen with products (eg the same shirt, but different colours). That is why Google recommends the use of the UPI (Unique Product Identifier, which is practically the rel=” canonical” of Google Merchant/Manufacturer.
Messy Middle and query rewrite
Let’s return to our example.
We’ve searched for “Star Wars: Legion Minis”, and started looking at the product’s Knowledge Panel.
One of its voices is “Stores”, and if we click on it we see how our initial query changes:
This behaviour is a constant when interacting with the Knowledge Panel.
Google, based on its Knowledge Base and Knowledge Graph, knows very well what are the successive steps that a user can take when looking for information about an entity (in our case “Star Wars: Legion”).
Thanks to this knowledge, instead of leaving the initiative to the user who should write another query in the search box (or dictate it with Voice Search), Google goes a step ahead and allows her to carry out the search journey in a very natural way. This is why the “0 clicks SERPs” skyrocketed in these past years.
In the specific case of “Star Wars: Legion Stores”, we see how the search intent changes radically, going from commercial to purely navigational.
That is why the Local Pack is present and that is why it also presents pages where you can find the product online.
Also for this query, Fantasy Flight Games does a good job. Despite being a manufacturer and not owning its stores, it anyway has a local strategy with demos of its board game and presentations of its miniatures in the stores that sell its products, and it has created a specific page to publicize these local events on its website.
What does this suggest to us?
- If we have a physical presence, then it is the case to start once and for all to pay the same attention to Local SEO that we pay to SEO for our website. This also means having pages about our stores in a specific section of our site and not just providing them with Google Business profiles.
- If we do not have a physical presence, it would be equally appropriate to see how we could be present in searches with a local/browsing intent.
For example, an idea could be to create a well-organized directory at the taxonomic level where all the stores that sell our products are presented.
Another alternative path that the Messy Middle of “Star Wars: Legion Minis” can have is specifically in the Reduction phase: exploring the reviews to have clearer ideas about the product, its quality, and details (eg: factions or heroes) than what is known. better give us ideas to refine our final search.
Also in this case Google rewrites the query for us:
We, therefore, find ourselves with another secondary query to take into account in our SEO strategy and thus target the Messy Middle.
What should we do to be prominently visible?
Creating a “What our customers say about us” page is surely a valid idea, but it is not the most effective.
If we want to be present effectively, we must do Barnacle SEO (I told you that I would come back to this concept).
- What are the sites that tend to be more visible both in the “Critic Review” box and in search results?
- How can we work so that the reviews – although honest – are in some way influenced by us?
- Have we developed a strategy so that these sites, which are a source of influence equal to or more than any classic Instagram or TikTok influencer, decide to publish reviews of our products?
Google also presents as Reviews the aggregate ratings of the Google Merchant feed, so – and this is something that tourism marketers are well conscious of – we should be aware of the reviews that our products are receiving on the sites where our products are sold and, where possible, respond to these Reviews.
This is even more relevant in the case of reviews that people can write directly through the form that Google offers for it. Again, all those who are used to tourism marketing (and Local Search) know very well the importance of having these types of reviews under control, in addition to the fact that – as confirmed by Google for Local Search – it is not unlikely to think that in the future Google will begin to use sentiment analysis algorithms to be able to better judge the quality and reliability of the products (the T of Trustworthiness of the E-A-T).
Continuing with the analysis of the Messy Middle of our example, a third turn that the loop can take is that of the videos.
The query rewriting now shouldn’t surprise us anymore, and in fact, we find it again:
If we have done our job of targeting the areas where triggers can potentially be generated, then this part of the Messy Middle will be largely covered by content about us.
At a strategic level, the most important thing that the more and more frequent presence of videos (and images) in the SERPs teaches us is not so much the fact that people find in videos the most suitable type of content to satisfy their information needs, but – as SEOs – that we must stop thinking of Google as synonymous only with Search.
Google is an ecosystem of products that constantly interact with each other.
This means that ranking is not just a matter of classic search results or optimizing our website, but optimizing for the highest possible visibility on all those products (YouTube, Merchant, Images, News, Discover…) with the same degree of priority as our website.
But, the Knowledge Panel presents us with one more surprise: the suggestion of related searches that are based on the Knowledge Graph.
In the case of “Star Wars: Legion Minis”, an example of these searches based on related entities is “Star Wars: Legion Accessories”.
Is this type of analysis also valid for exploring the Messy Middle of other types of searches, which are not product searches?
The answer is yes, as we can see in the screenshots below:
The secret that was always under our eyes
If the Knowledge Panel is the clearest example of how Google is evolving to accompany its users throughout the Messy Middle, the reality is that this is not the only element that Google uses to facilitate the journey in the Messy Middle.
The truth is that every feature present in the SERPs has this function.
Of the many, the most important for SEO is the People Also Ask block, as we see in its structured version in this graphic below-obtained thanks to the Alsoasked.com tool:
Analyzing the PAAs, and creating a content strategy that takes them into account, is essential to have the possibility that our website is chosen as the one that offers the most relevant answers and, in this way, to be the content that is capable of breaking the loop of the Messy Middle and take the user to the Experience phase and, possibly, to the final conversion.
Other essential features to be able to develop our positioning strategy in the Messy Middle are, for reasons that should be obvious to us, the Related Searches and People Also Search blocks.
But, the most important and unfortunately most overlooked Search Feature is the Google Images tags:
The analysis of Google Images tags provides us with two great benefits:
- Being able to easily discover the entities and attributes related to our main entity (eg: “Star Wars: Legion Minis”), which Google considers most important.
- Being able to build a map of the possible Messy Middle of our main query in an even more complete way than that offered by the Knowledge Panel study, which – furthermore – is not always presented by many queries.
The future is here: Google Lens, MUM, and the Messy Middle
In 2021 Google announced MUM, a new algorithm based on Artificial Intelligence with which it intends to revolutionize the way of searching for information on the Internet.
Without going into technical details, MUM (Multitask Unified Model) allows you to carry out a search journey naturally moving from visual to vocal search and, finally, textual (and all other combinations), in addition to many other things (eg, being multilingual, it can extract knowledge from content in one language to try to find similar content in ours that, initially, was not so directly retrievable in the index).
As we have indicated at the beginning of this post, MUM seems to be in the line of adapting Search as a product to the Messy Middle concept.
The clearest example of how MUM will change people’s behaviour when searching for information can already be seen in action with Google Lens, especially with functions such as Multisearch.
Google Lens now represents a huge advance in what is defined as visual search
We have seen before, speaking of Triggers, how Chrome allows us to search within the images that are presented both in Google Search itself and in websites that we are visiting, apart – obviously – from the photos that we take with our mobile.
These features, which are being used more and more, force us to rethink the use of images on our websites and in search results as well as a potential variant of the Messy Middle loop.
Optimizing images for Visual Search also means creating images that know how to “target” a parent object and child objects in a clear way, so that they can be easily identified by Google Cloud Vision and possibly cause, by searching for the focused object, the landing page. Next search results present turn our images prominently.
In addition, considering how Lens includes OCR functionalities for recognition of text embedded in images, it is also appropriate to consider those texts that we use in our images or even have in our packaging as potential sources of new searches in the Messy Middle:
MUM does nothing more than enhance these search possibilities.
With Multisearch, for example, we can carry out a visual search for a garment (or take a photo of ourselves with the garment on while we are in a store), and search Google if it is available in another colour thanks to the possibility of refining the visual search with other text.
Or, thanks to its Multimodal nature, with MUM we can frame an object and directly ask Google things about this object:
But MUM is not just Google Lens.
In US search results, MUM is already present with two features – Broaden This Search and Refine This Search – that further expand the navigation opportunities in Messy Middle directly from within Google Search:
So how to think about SEO and do SEO in the era of the Messy Middle?
Just thinking about all the possible Search Journey variables that the Messy Middle can generate due to the Search Features, that Google presents in the search results, and the full and closer and closer arrival of MUM, the confusion is clearly huge, and so the “Google is stealing our clicks” complaints are justifiable.
But with a “colder mind”, beyond the analyses that we can make of the Messy Middle itself and that we have exemplified before, there is something that can not only give us the key to how to do SEO for the Messy Middle, but that has also existed since time and, therefore, SEOs already know how to use it to optimize the visibility of a site and a brand in Google: Entity Search.
If you have noticed during the example of the Messy Middle of “Star Wars: Legion Minis”, the concept of entity was recurring.
This is logical: entities are the basic and unequivocal element of a search, and the keywords that we target are nothing more than a verbal expression of these same entities, relationships between entities, combinations of entities and their related attributes, and actions we may take about entities.
Basing an SEO strategy on the keywords, their successive clustering work, and, finally, creating content based on those clusters is not only a less effective way to establish an SEO strategy, but it also does not ensure that the strategy is ultimately the most correct or the most suitable to target the Messy Middle.
On the contrary, if we start with an exhaustive work of ontology analysis, the definition of primary and secondary entities, taxonomy, and, finally, the keywords and queries that explicitly (textual/vocal search) or implicitly (visual search) express verbally relationships between entities, attributes, and actions, then the work will be much more efficient and capable of laying a solid and lasting foundation for our SEO strategy for the Messy Middle, which will only need relatively limited updates at the ontology level.
Do exist tools that let us do this work a little bit faster than doing it manually? Indeed. To cite a couple, Wordlift.io for Entity analyses and Keywordinsights.ai, which is a tool “for a more civilized age” that allows us to do keyword research and clustering most efficiently.
Unfortunately, though, neither the best SEO suite nor SEO tracking tool is tracking those elements (Google Search contextual menu, Knowledge Panel tabs, and Image Search Tags) that we have seen as being fundamental for understanding the Messy Middle of our queries.
And, now, it’s time to return to our “Star Wars: Legion Minis” example for the last time:
“Star Wars: Legion” is a “Miniatures Tabletop” (relationship between related entities).
In turn, every “tabletop game” is made up of elements (sub-entities) such as the terrain pad, the miniatures, the cards with the characters’ statistics, and the rules of the game.
The minis, in addition, can be of armies and heroes (again, sub-entities), and the rule is that the players paint these miniatures (action related to an entity), for which they will need the help of guides, experts, and material to paint them (other entities, attributes, and related actions).
In addition, we must not forget that “Star Wars: Legion” is a game (product) that is produced by Fantasy Flight Games (Brand Entity and Manufacturer), for which we must consider all those entities and actions that are normally related to products (and the Shopping Graph): reviews of blogs, specialized magazines, YouTube video channels, and the experts, journalists, and Creators who carry them out.
As you can see, we already have not only a basic architecture optimized for Entity Search but also the main topics that we will develop to target the different search intent and those contents. that, now yes, the keyword research will suggest us as the most effective to respond to these search intent according to Google and that feeds the Messy Middle.
In addition, speaking of Messy Middle, the entity analysis work offers us the map of all the probable paths that the Search Journey can take both forward and backwards and, thus, the guide to target all the Messy Middles related to our products and brand.
Finally, speaking of Brand, this should always be considered as our nuclear entity, with which all the others that we will be targeting with our SEO strategy are related.
For this reason, the concepts of E-A-T (Expertise, Authoritativeness, and Trustworthiness) are so essential, because the more we know how to provide our brands with positive signs of E-A-T, the more the relationships between the Brand entity and the entity’s “products”, “services”, “ things” associated to it… will be considered relevant and worthy of being rewarded not only with greater visibility in search results but also the ideal for all SEO could be realized: the synonymy between brand and products/resources in the minds of those who search in Google.
Going back to the questions I asked at the beginning, it is clear that Messy Middle tells us that the classic concept of linear conversion funnel TOFU, MOFU, and BOFU is outdated.
Another clear thing is that complaining about “0 click SERPs” is legitimate when it implies blaming some clear monopolistic behaviour of Google, but it also means that how Google has changed the very concept of search is not fully understood.
Instead of the old search process of clicking, visiting a website, returning to the search engine to introduce a more refined one, clicking again and so on, Google encompasses this search loop and pushes to make all these passages within its search engine(s).
This means that thinking of SEO as the binomial Search and Clicks is also outdated and that we should think in terms of Search Session and Search Journey within the Search Session, something that is also clearly suggested by Google with GA4 at the web analytics level.
Finally, to be able to target the Messy Middle, it is better not to start from the roof (the classic keyword research with thousands and thousands of keywords) because it is not an efficient method to do it, but the correct thing is to start it from the foundations (Entity Search) because the Messy Middle is based on Entity Search and the consequent tokenization of the content that responds in a relevant way to the searches related to entities.
Post credit sequence
Once the Messy Middle loop has been broken, Google users will finally be our users and enter the Experience phase.
Once they are landed on our site, SEO continues to be important in its most technical aspect, because everything that is defined as Page Experience (Core Web Vitals, Mobile-Friendly, and HTTPS) are those areas of SEO that contribute most directly to achieving conversions.