Topical authority describes how usefully and credibly a business covers a defined subject. Map your customers' questions and your evidence, give each page a clear job and connect related answers. There's no required article count or public Google topical authority score. A complete map doesn't guarantee rankings either.
The goal isn't hundreds of articles. It's a set of answers your customers would actually miss if they disappeared.
What topical authority means
Topical authority is a planning idea from the SEO industry, not a number you can look up.
Google does use the phrase for one documented system. In a 2023 Search Central post, Google described a topic authority system for news. It helps identify expert sources for newsy searches in specialised areas, such as health, politics or finance.
That system is about news publishers. It isn't a general authority score for small business websites. It doesn't validate any third party scoring formula either. Three related ideas often get blended together:
E-E-A-T stands for experience, expertise, authoritativeness and trust. Domain Authority and Domain Rating are metrics from SEO tool vendors. The ideas compare like this:
Concept | What it describes | What it isn't |
|---|---|---|
Topic coverage | How well a subject is answered | A Google score |
E-E-A-T | The basis for trusting content | A specific ranking factor |
Domain Authority or Rating | A vendor's link based metric | A Google measure |
Google's helpful content guidance says E-E-A-T itself isn't a specific ranking factor. It also says Google has no preferred word count. Topic coverage, trust and vendor scores all differ from a real commercial result.
Say an accounting firm has one basic cash flow article. A competitor covers the same subject with:
practical cash flow guidance
a forecasting template
a relevant advisory service
permissioned client evidence
The second firm covers the subject more usefully. That doesn't mean it must rank better.
Technical access and useful content do different jobs. Crawling is Google fetching a page, and indexing is Google storing it for search results. A page Google can crawl still needs to answer the question well. A brilliant answer nobody can reach helps no one. Being indexable doesn't mean a page is indexed, and being indexed doesn't mean it ranks.
Small and local businesses can build topical authority too. An example roofer might answer real leak, repair and maintenance questions. Evidence from actual jobs makes that coverage useful. That's possible without a national publishing budget. The wider strategy sits in our guide to SEO for small business.
Understand entities and semantic topical maps
Semantic SEO focuses on meaning and the relationships between topics, not just single keywords. It adds some useful planning vocabulary. The terms sound technical, though the ideas behind them are simple. An entity is a distinct thing, like a business, a service or a place. A topical map plans the questions in a subject and the page that answers each one.
These terms come from practitioners' writing on semantic SEO, not from Google. One more is worth knowing. A contextual bridge is a link between neighbouring topics.
None of this is a Google specification. It describes how search engines might work. A patent or a case study isn't proof that a current ranking system does what it describes. What follows is our practical application of the terms.
The main terms, translated:
Term | Practical meaning | Example |
|---|---|---|
Central subject | What the map is about | SEO for Australian small businesses |
Attribute | An aspect customers need explained | Cost, duration or suitability |
Source context | Why this business covers it | An agency explaining work it does |
Query network | Related questions and contexts | Cost, inclusions and evaluation |
Contextual direction | The angle and order of an answer | Cost as a buying decision first |
Capture those terms for any topic in one record:
Central subject: [what this map is about]
Audience: [who needs the answer]
Source context: [why this business covers it]
Attributes: [properties and decisions to explain]
Related questions: [useful changes of context]
Excluded subjects: [outside this scope and why]
This guide calls the main thing a map is about its central subject, or central entity. It's a planning choice. It isn't a guaranteed field in Google's index. A business may serve several related subjects. The map should make the chosen scope clear without forcing every page into one rigid category.
In this guide, contextual direction simply means the angle and order of an answer. That's an editorial idea. It isn't a Google score, a hidden calculation or a list of words to add.
Consider one example question: what does SEO cost? An agency answers it as a buying decision, covering scope, inclusions and trade offs. A software vendor answers it as a subscription, covering licences and features. Same question, different direction.
Choose the topic your business can support
Start with what you actually sell, who buys it and the decisions they face. Then look at the evidence you can show. Some example scopes show the idea:
Business | Possible topic scope |
|---|---|
Plumber | Blocked drains and emergency plumbing |
Restaurant | Private dining and group events nearby |
Accountant | Cash flow and accounting for ecommerce brands |
Physiotherapist | Running injury rehabilitation |
Canopy retailer | Ute canopy selection, fitment and touring setup |
These are possible scopes. A business doesn't automatically have authority to advise on them.
For our worked example, the subject is SEO for Australian small businesses. The audience is owners deciding whether and how to improve their unpaid search visibility. The commercial link is real SEO work. The map doesn't need to absorb every Google Ads, web design or global enterprise SEO question.
Record what you're leaving out, and why. Typical reasons include:
a service you don't offer
a foreign regulatory setting
a specialist advice boundary
a question better answered elsewhere
Revisit an exclusion when something real changes, like your offer or your capacity to maintain more pages.
How many pages does a topic need? No evidence based universal range exists. A tightly scoped map might have eight destinations, as our example below does. Eight isn't a target for every business.
Include questions people ask after buying, plus your service limits. That coverage can cut unsuitable enquiries or help existing customers, even with little search traffic.
Research the questions and evidence
Good topic maps start with real questions, not a keyword export. Combine these sources:
approved customer questions and sales objections
service records and team knowledge
Search Console queries, the searches shown in Google's free site reports
actual search results for your area
competitor pages
Strip personal details from any research notes going to a writer. Then record where each question came from:
Question: [customer wording]
Origin and date: [interview, query report or suggestion]
Audience and place: [relevant context]
Confidence: [observed, inferred or not yet checked]
Evidence needed: [source, expert or original work]
Action: [new answer, improve, connect or defer]
Here's a filled example record. The question about needing a new website before SEO came from three sales calls in September. Confidence is observed, because customers asked it in their own words. The evidence needed is a technical review of the current site. The action is to improve an existing answer rather than write a new page.
A search result is one observation of one context. An AI generated list is a set of candidates to check. A plausible generated question is a research lead, not proof of demand or a fact to publish.
Not every gap needs a new page. There are four kinds:
no useful answer exists yet
an answer exists, but the evidence is thin
an answer exists, but people can't find it
the information is out of date or conflicting
Only the first necessarily suggests new coverage. The others might need a correction, a link or a better section.
Ask whether a question changes a decision or removes a real barrier. A competitor having a page isn't reason enough. Keep meaningful low volume questions, especially about safety and aftercare.
For regulated or technical claims, record the primary source and the reviewer needed. Don't draft unsupported advice just to fill every cell of a map. A common mistake is treating an AI generated question list as proof that people search for those questions.
Work through an Elev8d topical map
Our own subject makes a useful worked example. This proposed Elev8d SEO map uses existing pages and planned ones. The guide at the top acts as a hub, an overview page linking to the detail. The map doesn't claim every link exists yet. Nor does it claim the cluster, our group of pages on this topic, has produced measured gains.
Each page owns a question, with its status recorded:
Question | Proposed owner | Status or evidence needed |
|---|---|---|
How should we approach SEO? | Small business SEO guide | Planned parent, not live yet |
What does SEO cost? | SEO cost guide | Live. Needs approved current terms |
How should topics be planned? | This guide | Live. Holds the evidence decisions |
How should pages connect? | Site architecture guide | Live. Check the final content |
What belongs on a service page? | Service page SEO guide | Live. Check the final content |
What does structured data do? | Schema markup guide | Live. Check the final content |
How do we improve local visibility? | Local SEO guide | Planned, not live yet |
Who can deliver the work? | SEO service page | Live. Match the real service scope |
Each row then gets a full record:
Question and decision: [what the reader must resolve]
Attributes: [necessary details and exceptions]
Owner URL: [existing or planned page]
Evidence: [available proof and remaining gaps]
Related destination: [next useful answer]
Status and reviewer: [live, planned or held, and owner]
The table is a record of who answers what, not a keyword list to repeat.
A tree can't show every useful link. In this proposed map, the parent links to detailed guides where useful. The cost guide links to the service page for scope and a package explainer for buying detail. The architecture guide links to service page planning and schema. Not every supporting page has to route through the parent.
Keep entity relationships separate from URL relationships. An agency offers a service, a guide explains a decision and a template records evidence. A tree diagram can't show every one of those relations.
Develop one map row into a writer assignment
Take the cost row and turn it into a writer's assignment. The question is how much SEO costs for an Australian business. The central subject is buying SEO. The source context is an agency explaining a real service it sells. Its attributes include:
scope and starting point
recurring versus one off work
internal effort from the business
exclusions
how the spend will be judged
Our SEO costs guide owns the full answer. Searches about monthly SEO cost can share it when the intent is the same. A package inclusions guide can stay separate if it helps people compare deliverables. Don't split pages for every wording variation, like cheap or affordable.
The cost guide needs current approved terms for any Elev8d specific claims. Market benchmarks need a dated method. Where no client evidence exists, a clearly labelled example fills the gap. Don't present the guide's existing figures as newly verified market data.
Its links follow the reader's needs. A note on why scope changes price leads to what an SEO package includes. Readers who need help get the confirmed service page. The section on content investment points to topic planning. The finished page should:
answer the buying question
state assumptions and exclusions
separate example maths from a quote
support each factual claim
offer a useful next action
Search visibility gets measured after publication, never claimed in the map.
Choose the right page for each question
Every page needs a role, and one page can play more than one. There are four roles:
Role | Job | Example |
|---|---|---|
Commercial | Explain what you provide | An SEO service page |
Educational | Answer customer questions | How long SEO takes |
Decision support | Help people choose | SEO versus Google Ads |
Evidence | Show the work | A dated case study |
Roles can overlap. A service page can answer pricing questions. A guide can hold a template, and category pages can guide choices. For commercial pages, start with planning a service page.
Educational pages often live on a useful business blog, though they don't have to. A pillar page gives an adequate overview and helps readers pick their next question. Support pages go deeper on one need. No set length, card count or rule against short category introductions defines a good pillar.
Breadth and depth are different too. Breadth is how many relevant questions you cover. Depth is how well each one is answered, with evidence behind it. In our view, a small business competes best on depth for its own services.
Before creating a URL, run the question through this check:
A new URL needs a distinct useful job that someone can maintain. A narrow question with the same audience and decision may belong in an existing section. If the evidence is missing, getting it can beat publishing another outline. A common mistake is creating a new URL for every wording variation of the same question.
Search Console's query to page data is one diagnostic input. Review the content, user need and page history before merging anything. Several pages appearing for one search can be legitimate. Overlap alone doesn't mean a penalty.
Where a guide already covers a question, link to it instead of writing a competing one.
Connect answers to useful next steps
People don't move through a topic in a neat line. They can enter at any stage, skip ahead or come back later. One example route looks like this:
Each step needs different links, from early comparisons to proof and an enquiry route later.
Say an owner is chasing poor local visibility. Their route might pass through:
eligibility guidance for local results
a Google Business Profile explainer
our guide to local SEO, then a service option
Each link should explain why the next resource matters at that point. Cross topic links are fine too, like schema for a product page or measurement for a content plan. A rigid silo, where topics never link to each other, can block a real customer task.
Read more about our SEO services
If the audit reveals widespread indexing faults, review the scope of a technical SEO audit
Both versions are examples. Only link to a service that actually offers that scope.
Skip the identical sales pitch on every supporting page. Add a commercial link when the reader needs the help you offer. Link from the service page back to useful evidence or buying guidance too.
Link depth and navigation belong to website structure and internal links. So do orphan pages, the ones no internal link points to. This guide only needs the relationship between answers.
Keep the cluster accessible and accurate
A topic map only helps if people and search engines can reach the pages. Check the basics:
links Google can crawl, between related pages
public pages that are actually public
index settings that match your intentions
mobile pages with the same substance as desktop
consistent canonicals, the tags naming each page's preferred version
journeys that work end to end
Work through our technical SEO checks for the detail. A page with no internal links isn't proof Google never found it. External links or a sitemap, the file listing your pages for search engines, can still lead there.
Schema markup is code that describes a page to search engines. It can describe entities and relationships, as long as it matches the visible content. Even then, it doesn't guarantee a rich result, meaning an enhanced listing. For implementation, see schema markup.
Google says AI Overviews and AI Mode may use query fan-out, issuing several related searches across subtopics. That's one reason related questions may matter, not proof that a cluster shape earns citations. To be a supporting link, a page must be indexed and snippet eligible. A snippet is the short text under a result. No special schema is needed.
AI assisted publishing needs editorial rules of its own. Google's guidance on generative AI says it can help with research and with structuring original content. Generating many pages without adding value may break Google's spam policies on scaled content abuse. For the wider AI search picture, read about generative engine optimisation.
Build evidence people can inspect
Evidence is what separates a useful topic from a pile of articles. It comes in different strengths:
Evidence | What it can support |
|---|---|
Project observations | What happened on specific jobs |
Original tests | What a defined test showed |
Surveys | What a stated sample said |
Pricing datasets | Prices within a stated period |
Calculators | Estimates from stated assumptions |
Templates | A repeatable way to record work |
A calculator using assumptions isn't a market study. First hand experience shows in specifics, like the job, the constraint and what changed. Original work still matters. Google says it has systems to show original content prominently in search results. Record each evidence asset before you publish it:
Question and method: [what was studied and how]
Period, sample and exclusions: [defined population]
Result and limitations: [what the evidence supports]
Permissions and removed details: [approved public material]
Reviewer and owner: [responsible people]
Correction trigger: [when to update or withdraw]
A convenience sample, meaning whoever was easy to ask, doesn't represent Australian businesses. For an SEO case study, record:
the original URL set
the dates
other changes made at the same time
the search context
the outcome you actually observed
An impressions rise, meaning more appearances in search results, doesn't prove more qualified customers. A before and after comparison alone can't isolate the effect of a topic map. Before publishing client material:
get authorisation for screenshots and data
remove personal and commercially sensitive details
check small groups that could identify a client
record who approves corrections or a withdrawal
Private Search Console exports and customer messages shouldn't become article attachments by default. A useful research method doesn't authorise disclosure on its own.
Name real authors and reviewers with relevant experience and actual involvement. Don't add qualifications nobody holds. When a team member leaves, review any claims about their current role. The wider trust picture sits under demonstrating experience and expertise.
Keep external sources next to the claims they support. Independent recognition can back up expertise. No quota of backlinks, links from other websites, proves authority. A common mistake is calling a calculator's output research, when it only reflects the assumptions typed in.
Our free audit checks your content against what your customers search, then shows you the gaps.
Adapt the map to your business
The same method fits very different businesses.
Four example businesses show how coverage connects to real services:
Business | Coverage | Evidence to request |
|---|---|---|
Melbourne roofer | Leaks, repairs, delays, upkeep | Permissioned projects |
Ecommerce accountant | Reconciliation, stock, cash flow | Professional review |
Sydney private dining | Capacity, menus, booking terms | Real room details |
Outdoor gear store | Fit, setup, care, replacement | Measured specs and testing |
Each example keeps the service connection. The roofer's guides lead to real repair work. The store's articles help people choose categories and products. The restaurant needs current venue facts more than a big archive of generic hospitality articles.
Local authority needs local evidence. Examples should describe real operational differences, rather than a city name dropped in. The accounting example is only a planning aid. Professional advice boundaries and advertising rules need specialist input, as they would for any health business. For a store, ecommerce SEO covers category and product pages in depth.
Use the gap mapper as a planning tool
Our gap mapper helps you record coverage against an editable topic list.
It offers eight industry templates. Each is an editorial inventory rather than a validated map of customer demand. Treat its topics and definitions as a starting point for your own judgement. The tool says results stay on your device. Work through it in this order:
Choose the closest template.
Edit the scope to match your business.
Record the page and evidence for each topic.
Mark each topic Covered, Partial or Gap.
Review missing and partial work.
Take your chosen actions into the content plan.
The tool asks you to mark a topic Covered only when a useful page meets that intent. A complete section on an existing page can also answer a narrow question. Don't create a URL just to lift a tool score.
The score uses weights of three, two and one for high, medium and low priority. Partial earns half credit. An example three topic inventory works like this:
High priority, marked Covered: 3 of 3
Medium priority, marked Partial: 1 of 2
Low priority, marked Gap: 0 of 1
Weighted coverage is 4 ÷ 6 × 100, or 66.7%. Yet only one of the three topics is fully covered, and both statements are true.
If every topic is Partial, weighted coverage is 50% even though every topic still needs work. Filters don't change the overall figure. Keep the total topic count and template version visible when you compare.
As the tool itself says, the score isn't an authority metric, keyword volume study or ranking forecast. Don't fill every template topic just because it's there. Deciding what to fill first is covered in prioritising a content programme, along with AI publishing rules.
Measure progress without inventing an authority score
Pick a small set of measures that suit your business. Avoid a composite score that hides weak evidence. Choose from categories like these:
Category | What it shows | Limit |
|---|---|---|
Coverage | Share of agreed questions answered | Depends on the list |
Visibility | Impressions and clicks for the set | Exposure, not revenue |
Relationships | Links between related pages | A count isn't quality |
Quality | Accuracy and evidence reviews | Needs a real reviewer |
Recognition | Independent mentions and citations | No quota proves authority |
Commercial | Suitable enquiries and sales | Attribution is messy |
Record a fixed set of priority questions and URLs before you change anything. Report that set separately from new pages. An example shows why:
Scope | Answered well | Coverage |
|---|---|---|
8 agreed questions | 6 | 75% |
Scope grows to 10 | 6 | 60% |
The drop reflects a bigger total, not worse content. For search evidence, keep these the same between periods:
the Search Console property, meaning the site version you track
the date ranges and comparison method
the country
the device
the search type
Note whether a report covers the fixed set or the whole growing library. Note seasonal swings. A drop after a busy season isn't automatically a fault. Annotate launches and migrations too. Set up the reports with our notes on Search Console measurement.
Search Console drops rare, anonymised queries from its tables. Query filters, including non brand filters, drop them from totals too. So a filtered view only shows part of the picture.
Chart and table totals can differ too. The chart counts the whole site. A table grouped by page counts each page.
Click through rate (CTR) is clicks divided by impressions. Don't average page CTRs or positions to build a cluster figure. Two example pages show why:
Page | Clicks | Impressions | CTR |
|---|---|---|---|
A | 10 | 100 | 10% |
B | 30 | 3,000 | 1% |
Combined | 40 | 3,100 | 1.3% |
Averaging the two rates gives 5.5%. The real combined CTR is 40 ÷ 3,100, about 1.3%. When you combine compatible pages, divide total clicks by total impressions.
Define brand and non brand queries explicitly. Note unusual changes to the results page, like a new AI Overview. Not every CTR movement is a content verdict.
Measure suitable enquiries, useful service journeys and business outcomes with your own definitions. Don't credit every later sale to the first article someone read. An impressions rise measures search exposure. A page can collect impressions and still fail the customer's task. If you get visits but no enquiries, start with why traffic does not become enquiries.
Maintain the topic before expanding it
Keep what you have accurate before adding more. Review a page when any of these change:
services or prices
rules or platform behaviour
your evidence
staff roles
customer questions
Keep review dates truthful. Changing a date without improving the answer is a freshness trick, not maintenance. Maintenance options include:
improving a weak section
merging genuinely redundant answers
repairing useful links
retiring content deliberately
Say two articles answer the same cost question for the same audience. Merging them into one stronger answer keeps the useful parts. A redirect then sends visitors from the retired URL to the merged page. Leaving both splits the evidence.
Low traffic alone isn't a reason to remove aftercare or operational information. Before expanding, check readiness:
Expand when the offer, the evidence and your capacity all line up. When they don't, record a decision to defer. That's an editorial call, not a Google pass mark. A common mistake is starting a second topic while the first one is still half answered.
A practical 90 day plan
Treat this as a flexible sequence.
Each window has a clear job:
Days 1 to 30: list existing pages, collect real questions and note missing evidence
Days 31 to 60: improve the main commercial pages and the most important answers
Days 61 to 90: connect related pages, test journeys and record a baseline
Adjust the workload to your capacity. There's no required number of new articles, evidence assets or tools.
Later, schedule a review of maintained coverage and adjacent topics. That review isn't a ranking forecast.
What we recommend at Elev8d
Map the questions before you plan the calendar. Pick one subject you can genuinely support. Give each important question one page that answers it. Then fill the evidence gaps. A smaller set of answers people trust beats a big archive nobody needs.
Frequently asked questions
How many articles do I need?
There's no evidence based universal number. Cover the decisions your customers face and back them with evidence. Stop where useful demand or your capacity runs out.
Can a useful section count as coverage?
Yes. A complete section on an existing page can answer a narrow question well. A new URL needs a distinct job.
Does every keyword need a page?
Not usually. Related wording with the same intent can share one answer. Split only when the decision or audience genuinely differs.
Does a higher tool score prove authority?
Not at all. A coverage score measures your own records against an editorial list. It says nothing certain about rankings or trust.
When should I expand to another topic?
When the offer, evidence and maintenance capacity line up. Keep the current map accurate first.
Next steps
Start with one question and see it through.
Pick one unanswered question
Choose a question your customers ask that nothing on your site answers well.
Check for an existing owner
See whether an existing page should own it before you create a URL.
Record the evidence needed
Note the proof, source or reviewer the answer needs. Our SEO support can help build the rest of the map.