Query templates: Expanding the scope of topical authority

Topical authority isn’t built by covering a topic alone. It also depends on how well your content covers the different ways people search for that topic. Query templates provide a way to map those variations and build a stronger, more connected content network.

There are two main methodologies for building topical authority: covering every entity and its attributes within a topic or covering every variation of a query template. Examples of both methodologies are below.

• Entities of the same type share the same set of attributes. “Calorie” is an attribute of every entity in the “food” class, just as “symptom” is an attribute of every entity in the “disease” class. Processing an entire entity class through its shared attributes signals comprehensive coverage of the topic to the search engine.
• Query template variations, on the other hand, don’t require topical relevance to one another. WikiHow, for example, holds authority for the “how to” query template, which allows it to rank across many unrelated topics at the same time. In this case, the authority attaches to the query format rather than to a single topic.

The hybrid methodology is the strongest of these approaches. It covers all entities from the same class, with all of their attributes, across all query template variations, and it unites topical depth with query format breadth in a single content network.

The first case study, Visual semantics: The missing piece of topical authority, explains how to use web components and design elements to improve query responsiveness along with query relevance. 

The second case study, How semantics and topical authority improve local SEO, covers the “Query Deserves a Page” principle and its framework across 13 different local SEO projects. 

Reading both will make it easier to follow the concepts and results in this mini case study.

Why does Google use structural similarity between queries and documents?

Because it’s cheaper.

The cost of retrieval is always central to understanding Google’s ranking decisions because Google is primarily designed to save costs rather than to serve the best. If a website’s quality is 6/10 and its cost is 7/10, it’s not worth retrieving. The principle of cost of retrieval is explained in one sentence:

“The cost of ranking a site can’t exceed the cost of not ranking a site.”

Query templates help Google satisfy more users and drive more clicks while organizing more sources with lower computational needs. 

If a website satisfies a query and there is another similar query, Google triggers a test for ranking purposes. This created the concept of semantic content network, meaning that a group of web documents is semantically connected to cover all of the semantic query network to trigger a re-ranking.

Query templates for a QR code generator

The first example project in this case study doesn’t appear in the previous two articles.

It’s a QR code generator that gained over 1 million extra clicks in three months, mainly through microsemantic changes, supported by technical SEO improvements and a CMS migration. 

To keep the focus on the query template aspect, I’ll keep the technical improvements and checks brief.

Technical improvements

Besides the query template- and query semantics-related reasons and the historical click data signals, the main technical improvements are listed below.

• The website was migrated from WordPress to Next.js with Sanity as the back-end CMS.
• During the migration, the image-to-image, HTML-to-HTML, CSS-to-CSS, and JavaScript-to-JavaScript migration principles were followed, meaning every asset type was mapped to its equivalent on the new stack.
• All non-indexed URLs were removed from the website to prune the source’s crawl profile.
• Response times were improved to increase crawl efficiency.
• No resources or URLs were left behind during the migration. Everything was moved to the new system.
• Structured data was updated.
• It was ensured that the “centerpiece annotation” of the website, the QR code generator itself, is served without requiring JavaScript rendering.
• The only change in the CMS migration was the back-end infrastructure. The content, layout, and URLs were all kept the same.
• The HTML structure was cleaned and the DOM size was reduced.

To show the importance of handling redirections at the back-end infrastructure level, I can use the image crawling and image ranking changes. I have two rules for every migration:

• Change only one thing at a time.
• Never leave any URLs or resources behind so the search engine can adapt to the new system faster.

If you look at the image performance data below, we’re losing rankings because we couldn’t redirect all of the images on the site. We were only able to redirect the most important ones. 

This partial redirection is a problem in itself because my default position is never to change image or video URLs at all. Indexing is far more costly and slower for these resource types, so any changes to them take much longer for a search engine to process and trust again.

Cost of retrieval is a concept that I use as one of the inspirations behind topical authority because it isn’t about delivering more quality. It’s mainly about being cheaper relative to your quality. The cost of ranking a site can’t exceed the cost of not ranking a site. 

If ranking you is costlier than not ranking you, deindexing begins. Since we couldn’t keep the image URLs the same during the migration, you can see how this is reflected in the crawl data below.

I always try to increase the number of crawl requests per URL because it consistently improves rankings, indexing speed, and crawl delay, and increases the number of query terms the site covers. 

In this case, however, the total crawl hits per day decreased after the migration, and that’s actually a positive signal. The decrease came from the non-necessary section, the “other file types” segment, while crawl requests for the HTML section increased.

Smartphone crawl requests increased as well. The crawl ratio of indexable HTML URLs that are self-canonicalized, included in the sitemap, and supported by internal links also increased, which is positive. 

But as you can see, image requests and image rankings moved in the wrong direction together due to a migration error.

Google creates landing page and image pairs to rank web documents. For that reason, image rankings affect a website’s overall rankings far beyond just image impressions or clicks.

According to the purpose, platform, and type of the QR code, the query semantics change, sometimes slightly and sometimes heavily, depending on the contextual domain. 

A contextual domain is the collection of all the context vectors that can be created from a term by adding one or two more words through vectorization.

Query component	Position	Variations
Contextual noun	Beginning of the query	PDF, PNG, JPG, URL, Phone, Website, Menu, Hotel, Event, Facebook, Instagram, Dynamic, Static, Trackable
Central entity	Middle of the query	QR Code
Tool synonym	End of the query	Generator, Creator, Maker, Designer

For example, the predicates that we can use for “Facebook QR Code Generator” and “PDF QR Code Generator” are mostly shared, but they differ in ways that matter for relevance. 

The predicate “download” is semantically close and highly relevant to “PDF QR Code,” whereas it sits much further from “Facebook QR Code,” where predicates like “share” or “follow” dominate. 

Here’s a comparison of the index sizes that Google returns for each of these co-occurrences.

Every ranking query, along with its historical click and impression data from web search or generative AI search results, increases the ranking chances of the other variations in the same query template.

Earlier, I explained my Query Deserves a Page principle, which decreases the cost of retrieval and increases the ranking signals per web document. 

According to this principle, we open a new page only if the query’s search demand exceeds a threshold and the query involves a different entity or a different pattern that has low semantic similarity to the existing pages. 

In other words, query templates help Google classify the web faster and choose a single website to satisfy more queries from the same topic or the same template.

The Query Deserves a Page principle is applied to query templates with a similar mindset by grouping the predicates that can be used with the specific entity in each query group. 

• If the predicates of a specific query template variation differ from the predicates of the others, it’s usually a signal to open a new page. 
• If the predicates are highly similar but the entities in the query variations are different, it should still be a separate page.

That’s why we use the micro- and macrosemantics concepts to differentiate textual and visual semantics optimizations from each other on a relevance level. 

For example, “QR Code Generator for Facebook” related SERP candidates always have the same functionality, with very similar layouts and annotations. The difference in the internal factors comes from the micro-differences. 

Google’s “Determining User Intent from Query Patterns” patent includes instances of changing relevance and rankings from the small differences in the documents or query interpretations.

To explain microsemantics, I usually use a simple example. Check the two sentences below.

• Financial independence is achieved by families with the help of financial advisors.
• Financial advisors help families achieve financial independence.

Both sentences state the same fact, but depending on the query network I’m targeting, their relevance scores will differ significantly. Micro differences in word order and in the dependency tree create relevance differences in exactly this way, and these micro differences are what I call microsemantics. 

If the query has “financial advisor” as the subject, your sentence should also have it as the subject. If “financial independence” carries more weight in the queries, then “financial independence” should sit in the subject position of your semantic triples.

A triple is simply the “subject, predicate, object” order of a statement. Matching the query augmentation models with the declarative facts you host on your web documents is a microsemantic improvement that maximizes your relevance. 

It looks like a small improvement, but if you’re trying to rank for a query template with 3,000 variations, that small difference is multiplied by 3,000 and becomes a major ranking factor.

Query augmentation is a concept I explained in our previous case study, which comes from Anand Shukla’s design, and it’s a direct equivalent to what Google calls “query fan-out.” 

How Google augments a query term impacts the “relevance weight” of a term in a query and how you should structure your sentences. 

Sentential, or sentence-level, semantics are part of “microsemantics” to increase relevance in a marginal way to be a main candidate document for an entire query network. 

Below is the “Query Suggestions Templates” anatomy for the same “query-template” scoring and ranking.

Bill Slawski’s last article was about “Generating Query Answers.” 

The patent design here explains that certain query templates require certain answer templates, and the search engine was looking for documents that have these templated answers in a better formatted way to decrease the cost and increase the quality and efficiency of index construction.

If there is a query template such as “When is [entity] born,” Google looks for answer annotations to classify a page as useful or non-useful, and it eliminates the non-useful sources faster. 

Ask yourself this question: “What type of visual or textual template does Google look for in my industry?” to evaluate my document network as efficient or costly. 

Google refers to these semantic conditions as constraints. When a document satisfies these constraints, it’s considered template-efficient. 

If you’re targeting a specific query template, your content should be structured with template-efficient sentence patterns supported by strong microsemantic relationships.

The next example project will explore this concept in greater depth and demonstrate how these constraints can be applied in practice.

Query templates for rehabilitation as an example

When you rank a relatively weaker brand in terms of budget, historical data, PageRank, and branding, you need stronger historical click data to convince search engines to trust your web entity. Thus, in the rehab industry, it’s usually better to go after two main templates:

• Does [entity] make addiction
• Can I [drink/eat] [entity] [with, after, during, before] [entity]

These two query templates are rich enough to accumulate a high volume of historical data. Every click, impression, and search engagement, along with LLM-feeding engagement, makes the web entity more authoritative and trustworthy for the other query variations. 

This is the main domain of the project that targets these two specific query templates.

The Semrush results for the same project include the Google Business Profile ranking benefits at the same time, unlike Search Console-only results.

Below is the ‘Extension Domain’ project’s first-month results with the same semantic content network that targets the same query templates with the same answer templates and semantic constraints.

The “addiction types” entity class has more than 30 entities as members, and every addiction type entity carries sub-relations to substances, habits, therapy methodologies, withdrawal symptoms, causes, risk factors, and treatment durations. 

Covering each member of the entity class with the same set of attributes and sub-relations is what turns individual pages into a topical map, and it’s the entity-attribute path to topical authority that I described at the beginning of this case study.

To implement the Query Deserves a Page principle for the two main query templates, “does [substance] cause addiction” and “can I take [substance] with [substance],” we extract the attributes to cover and connect them back to the “[addiction type] rehab in [locale]” core pages in both micro and macro contexts through internal links between the related web documents. 

This connection works in both directions. Whenever an informational templated page ranks higher, the commercial landing page connected to it also ranks better.

This diagram explains how these query template variations contribute to the rankings of the commercial pages for locale-specific queries.

---

Structuring core and outer topical map sections around query templates

The core section and the outer section are the two main parts of a topical map. Internal links always flow from the outer section toward the core section to transfer ranking signals. 

• The core section is mostly commercial, and it carries the most important concepts and entities. 
• The outer section is mostly informational, and it bridges the central entity to the other main entities of the topical map. 

The third project operates in the word games industry and ranks mainly for the “unscramble [word]” query template.

We always decrease the cost of retrieval by:

• Removing unnecessary pages through the Query Deserves a Page principle. 
• Solving the technical SEO problems that dilute ranking signals, which increases the ranking signal per web document. 

The main issue in the word game industry is that every website carries the same information.

The results for the “unscramble carry” query are always identical across sites, so it’s critical for us to implement visual and textual microsemantics better than the competition, increasing the relevance per document enough to compete against 1.82 million results.

Regarding the core and outer sections of the topical map, the “unscramble” industry requires a programmatic SEO approach. As a result, the core section is not divided into commercial and informational. It’s divided into “most common and popular words” and “less popular words to unscramble.” 

The 1,000 most important words with evergreen search demand receive the majority of the internal links, are linked directly from the homepage, and provide more unique information than the competitors at a cheaper cost of retrieval.

To structure these types of projects, using BigQuery is a must because Google Search Console loses nearly 40% of the query-based click and impression data through k-anonymization. 

Here’s a programmatic command for checking indexing and query performance.

Another useful check is a “crawls and clicks” performance comparison, which shows where Googlebot spends its crawl activity and where the clicks and impressions are actually generated. 

Below is an example of a discrepancy from another example project. The site receives a huge share of its crawl activity on tech and asset URLs, but those URLs do not generate any clicks. 

Normally, this crawl quota should be allocated to the sections that matter for historical click-satisfaction signals. This is why topical authority is always strictly connected to the technical side of search engine communication.

The examples below come from multiple projects, including the 13 websites covered in the previous SEO case study and several new sites included in this article.

Example 1 for core, outer and query templates

The project below is mainly about social media marketing and promotion, and it has five main query templates.

• [Social media platform] + promotion
• [Social media platform] + engagement metrics
• [Social media platform] + influencers
• [Social media platform] + growth
• [Social media platform] + guides

Across all of these templates, some platforms deserve their own page, while others deserve only a section or a sentence. 

However, if a query network contains a Boolean question or an explicit question query, it’s usually better to turn it into a page for the entire network as part of the outer section. 

These question pages are helpful for earning the first LLM approvals for generative AI search engagement, along with semantics-related SERP features such as People Also Ask and Featured Snippets. 

In this case, a template like “does [social media platform] pay for [engagement type]” is a useful query template to leverage for transferring ranking signals further into the core section of the topical map.

Below are the results of the main domain and its extension domain. These domains, MoreThanPanel and StreamingMafia, were explained in the context of visual semantics in our first case study.

Example 2 for core, outer, and query templates

The same logic also works for tourism and travel enterprises such as ETSTur and Ucuzabilet, with query templates like: 

• “Ancient sites in [city]” 
• “Trekking routes in [city]” 
• “Hotels in [city]” 
• “Adult hotels in [city]” 
• “Top restaurants in [city]” 
• “Top museums in [city]” 
• “Top churches in [city]” 
• And more. 

Since one of these sites is for hotels and the other is for flight tickets, they share the same outer section of the topical map, differing only in contextual vectors and micro contexts. 

A contextual vector is the order of headings and page segments that flows the context from one topic to another. One website directs the context toward “where to stay,” while the other directs it toward “when to visit” and “how to get there,” whether by booking a flight or bus ticket or renting a car.

Technically, the micro context is the representation of the outer topical map inside the same web document. The macro context, wherever it appears, mostly supports the contextual relevance toward the core section of the topical map. 

All of these route, destination, hotel, holiday, and city query networks are tied to each other through entity-attribute pairs and query templates, together with all of their variations. 

Here’s a representation of the hotel-related and transportation-related outer sections of the two topical maps and how both follow the same logic.

With the same logic, you can see the tourism-focused website’s results below.

Example 3 for core, outer, and query templates

The website here belongs to one of the biggest hosting companies, and the core and outer sections of its topical map shifted to new subjects through different query templates. 

Since “website builder” and “AI website generator” tools have become a major source of demand in the industry, the core and outer sections of the topical map had to change accordingly.

The results while migrating the topical map’s core from hosting to the website builder.

The core section moved from “hosting for [industry]” to “[industry] website builder,” while “[industry] website templates” and “how to open a [industry] business” became the main outer section subjects. 

Depending on search demand, the outer section can extend into further query templates such as “[industry] website examples,” “[industry] website headers,” “footers,” “about pages,” “sign-in pages,” “product pages,” and “pop-up ideas.” 

All of these queries and page template pairs eventually link back to the core section to transfer their ranking signals, exactly as in the first examples.

Example 4 for core, outer, and query templates

The company here is an ecommerce company focusing on rave culture and clothing. It’s also a case study project we didn’t give earlier. 

The project’s core section is directly related to the [gender] [rave] [clothe/dress/top/crop/bottom]-like category query templates, while the outer section mainly focuses on the “festivals” with two main templates.

• Festival + Checklist
• Festival + Outfit ideas

We mostly keep these pages also commercial, even if they are outer sections of the topical map. Informational documents should be commercially related as much as possible by providing direct conversion elements on the top of the page. 

Even if we provide checklist items or outfit ideas, we merge these with our products by commercializing the informational documents.

This brings us to the four main principles we apply to semantics: visualization, verbalization, commercialization, and contextualization. 

Whenever you commercialize a document and website, you have a higher chance of ranking in Google’s post-helpful content system updates, which favors functional websites with concrete and unique services and benefits, rather than just content sites.

Here’s a case study spanning two and a half years. A semantic content network that didn’t rank while it was hosted on an affiliate website began ranking better when we moved it to a commercial website, with exactly the same content and context.

You can observe how ranking signals are created in the outer section of the topical map and transferred to the predominantly commercial core through query-template variations. 

Every click, impression, user interaction, or LLM grounding generated by the outer section reinforces the contextual relevance and ranking signals of the commercial pages through the internal linking architecture.

The final point is that every new supporting document you publish creates additional contextual anchors for the core pages in your topical map. 

As these documents expand the internal linking ecosystem, they introduce fresh anchor contexts that reinforce topical relevance, contribute to content freshness, and provide additional ranking justification through Google’s anchor tag indexing system. 

In our system, we avoid repeating the same anchor text more than three times within the main content. This helps preserve anchor diversity, strengthen contextual uniqueness, and provide clearer originality signals across the internal linking structure.

Example 5 for core, outer and query templates

Do we always have to cover these query templates by opening new pages? Certainly not. A topical map, or even an entire semantic content network, can be exactly one page. 

Below is an example from our first case study, which covers all of the query template variations listed as a single page. 

Most of these variations don’t even appear in the headings, yet the page still penetrates all of these queries better than other sites. The main reasons for this are visual semantics, positive historical click data, and the higher relevance weight that comes from having fewer words on the page.

Some of the main query templates covered as a single page, despite other sites doing the same with multiple web documents, include:

• MP4 to text.
• MP3 to text.
• Video to text.
• Audio to text.
• WAV to text.

This graph shows more variations.

Query templates require consistency and uniqueness

Covering more attributes than your competitors or more query template variations than anyone else doesn’t guarantee better rankings. 

SEO is a complex adaptive system, and Google continually tests websites on the SERP to determine whether they satisfy users. Broad Core Algorithm updates often trigger these re-rankings.

NavBoost is one of the most important ranking concepts in today’s SEO because it combines clicks, topicality, and PageRank as a hybrid ranking factor. The RankLab module explains these click tests in greater detail.

The Automatic Query Pattern Generation patent explains how Google generates new query templates and constructs indexes.

Google relies on patterns because pattern-based ranking solves problems faster. Paul Haahr of Google explained this during his “Improving the Search over Years” speech:

• “If there is a query with wrong results, we won’t fix it. If this query belongs to a cluster, we will fix it. We are not interested in individual SERPs even if we know that it’s wrong.”