Tools for your next search decision.
Practical research tools and datasets to help you examine, compare and document what you find.
Explore the collection
SEO for your industry
Guides for IT, HR, accounting and other service businesses.
Browse the library
-
Calculate the additional clients needed to cover your modeled spend, with assumptions and limits attached.
Output: CSV or JSON scenario record
-
Separate shared requirements from repeated collection work across brands, markets and teams.
Output: CSV or JSON scope packet
-
Write one useful answer with its evidence, qualifications and optional next action.
Output: Working draft as Markdown or TXT
-
Record and review AI assistant citations for your own searches.
Output: CSV or JSON export
-
Assess keyword candidates with evidence and page decisions.
Output: CSV or JSON review record
-
Check statements, compare sources and note your editorial decision.
Output: CSV or JSON review record
-
Inspect pasted llms.txt content, then compare source passages.
Output: CSV or JSON review record
-
Explore eight sourced findings about AI search, with clear limits.
Output: Selected findings as CSV
-
Compare documented work and evidence, then keep the questions to ask your shortlist.
Output: Shortlist brief as TXT
-
Compare tools for monitoring, content and publishing, then prepare an evaluation brief.
Output: Evaluation brief as TXT
-
Tool · Research & evidence
AI Visibility Plan Reference
Inspect dated subscription prices and limits, size a monitoring workload and retain the questions still to ask.
Output: Plan CSV or JSON and a requirement record
-
Plan employer-facing HR pages, distinguish advisory work from recruitment and inspect a published service-navigation example.
Output: Practical guide
-
Clarify managed IT, helpdesk and specialist support, then use a worked page map to evaluate SEO work.
Output: Practical guide
-
Separate employer discovery from candidate reach, with a staffing page map and vacancy lifecycle guidance.
Output: Practical guide
-
Map the actual work of an accounting or bookkeeping practice to useful pages and relevant business searches.
Output: Practical guide
-
Map modernization, integration and development-team engagements to distinct pages, with evidence of the work.
Output: Practical guide
-
Turn an accurately scoped project record into useful discipline-page evidence and clear publication decisions.
Output: Practical guide
-
Use a capability inventory to choose process pages, combine material variants and hold unsupported sector claims.
Output: Practical guide
-
Separate address plans, workspace offers and location details, with distinct checks for operator and tenant eligibility.
Output: Practical guide
-
Check proposed searches against testing and advisory services, then write an accurate explanation of an accepted opportunity.
Output: Practical guide
-
Turn an unsupported operational claim into useful reasoning, with evidence to examine, alternatives and explicit limits.
Output: Practical guide
-
Sort shipper discovery from tracking, jobs and software searches, then define one accurate service-page improvement.
Output: Practical guide
-
Turn a program brief into useful public copy and distinguish completion, satisfaction, assessed skills and later outcomes.
Output: Practical guide
-
Audit maintenance, replacement and repair pages, then complete a clearer commercial HVAC service explanation.
Output: Practical guide
-
Turn an equipment list into a useful installation and integration page, with clear handover and support.
Output: Practical guide
-
Connect a testing capability and its accreditation scope to a useful service page and a clear search task.
Output: Practical guide
-
Turn a production-equipment maintenance brief into relevant search targets and a concrete service-page improvement.
Output: Practical guide
-
Turn an assessment deliverable into useful service-page content, with a completed dependency-record example.
Output: Practical guide
-
Connect a finished portfolio example, its visual and the studio's contribution to a useful service explanation.
Output: Practical guide
-
Turn a defined business-contract engagement into a clear service page, supporting article and discovery check.
Output: Practical guide
-
Complete a membership-information audit, correct access claims and publish a useful comparison of physical workspace plans.
Output: Practical guide
-
Understand AI SEO, AEO and GEO, then identify the work and evidence your situation needs.
Output: Practical guide
-
Read eight AI search findings with their sources, study dates and limits before using a statistic.
Output: Practical guide
-
Check indexing, snippets and Google AI inclusion settings, then decide whether the page needs a change.
Output: Practical guide
-
What Google says you can skip: llms.txt, chunking, AI rewrites, fake mentions, AI schema.
Output: Practical guide
-
Inspect our live-file audit, check pasted llms.txt text and decide who will use and maintain the file.
Output: Practical guide
-
What AI visibility actually needs from a service business: access, clear service pages, proof, local signals and routes.
Output: Practical guide
-
Find which pages and measures changed, separate confirmed faults from update hypotheses and prepare a diagnostic handoff.
Output: Practical guide
-
Compare crawler policies, verify a reported block and retain the evidence for a narrow configuration decision.
Output: Practical guide
-
Separate ChatGPT search access, training preferences and user visits, with a clear check for your web team.
Output: Practical guide
-
The DUCR model: four areas to inspect when deciding what a page needs, with evidence before a fix.
Output: Practical guide
-
Turn checked material into a useful answer section, with three worked examples and a draft composer.
Output: Practical guide
-
Investigate an apparently wrong business fact, compare its sources and prepare a correction packet.
Output: Practical guide
-
Map buyer questions to useful page roles, then investigate overlap before changing what already exists.
Output: Practical guide
-
When a service, a place or an industry deserves its own page, and the rules that stop them duplicating each other.
Output: Practical guide
-
Four before-and-after cases of giving each page one job.
Output: Practical guide
-
Keep, reposition or combine: compare the buyer decisions and evidence before changing two overlapping pages.
Output: Practical guide
-
Publishing about AI search without splitting your own pages: ownership, borders, redirects.
Output: Practical guide
-
Where a search term belongs: the service page or the post.
Output: Practical guide
-
Choose where the next effort belongs: an existing page, a missing buyer answer or a problem content cannot solve.
Output: Practical guide
-
The complete checklist for a service page, without turning it into generic SEO copy.
Output: Practical guide
-
One service, and the questions buyers ask before they contact you, mapped to the pages that answer them.
Output: Practical guide
-
Choose a useful next step from one article, with link wording and destinations that answer the reader's remaining question.
Output: Practical guide
-
Plan the connections around one service: what buyers need, where the answers exist and which links deserve attention.
Output: Practical guide
-
Separate a missing-index problem from an indexed page that is not reaching the searches you care about.
Output: Practical guide
-
Location pages that serve real local visitors, and the doorway patterns Google treats as spam.
Output: Practical guide
-
City pages, area pages or one coverage page: how to choose without building dozens of thin pages that compete.
Output: Practical guide
-
How the Google Business Profile, the service pages and reviews work together for local searches.
Output: Practical guide
-
Find the first enquiry gap to investigate: contact delivery, relevant visitors, service fit or the next step.
Output: Practical guide
-
Help a suitable buyer understand the service, assess the work and decide whether to contact you.
Output: Practical guide
-
Choose evidence that answers the buyer's concern, including relevant cases, samples and confidential work.
Output: Practical guide
-
Guide · Template
Service page brief template for business owners
Seven business questions for one priority service page, with a worked example and a matching printable brief.
Output: Worked example and printable PDF
-
Read the dedicated AI report, compare the same scopes and keep unavailable export values out of the calculation.
Output: Practical guide
-
Inventory main-site, campaign and booking events before validating a hostname Include filter.
Output: Practical guide
-
Group bot requests by page role, retain coverage limits and distinguish repeated visits from pages reached.
Output: Practical guide
-
Compare monitoring, content and publishing tools by task, documented capabilities and practical limits.
Output: Practical guide
-
Inspect 30 candidate phrases, their evidence and page decisions, then keep your own research record.
Output: Practical guide
-
Review a teaching draft claim by claim, including sources, qualifications and metadata.
Output: Practical guide
-
Find recorded ChatGPT visits in the current GA4 reports, then check entry pages, completed actions and received enquiries.
Output: Practical guide
-
A working scorecard and enquiry log for deciding whether visibility is reaching suitable buyers and what to improve next.
Output: Practical guide
-
What changed in Google's AI search and ChatGPT search in the second quarter of 2026, what stayed stable, and what to do.
Output: Practical guide
-
Check readiness and model the contribution needed to cover a defined commitment before buying.
Output: Practical guide
-
Software, advice or implementation, what changes the scope and what the fee covers.
Output: Practical guide
-
What ships, what changes and what to review: delivery dates and the 42-day reading.
Output: Practical guide
-
Launch checks, relevant enquiries and the next investment: practical 30, 60 and 90-day reviews with clear responsibilities.
Output: Practical guide
-
Repair or rebuild: compare the actual constraint, what can be retained and the work each route requires.
Output: Practical guide
-
Keep useful pages, map changed URLs and review launch evidence with blank and fictional worked records.
Output: Practical guide
-
Plan a first website with a whole-project brief, page responsibilities and a fictional worked example.
Output: Practical guide
-
Compare seven providers' scope, delivery responsibilities and evidence, including our disclosed own offer.
Output: Practical guide
-
Compare providers on work, evidence and responsibilities, with a worksheet to record their answers.
Output: Practical guide
-
Choose a public review, a focused investigation or a conversation by the decision it should help you make.
Output: Practical guide
-
Who researches, writes, implements and approves: practical answers about working with us and your existing team.
Output: Practical guide
-
An optional one-page note for the business question you want to work through. Rough notes are enough.
Output: Practical guide
-
For agencies: when white-label delivery beats hiring, and what a healthy partner model looks like.
Output: Practical guide
-
Research & evidence
AI Visibility Benchmark
Inspect our June 2026 review of 50 service-business websites, with recorded observations and limits.
Output: Dated benchmark and findings
-
Research & evidence
Reviewed Benchmark Dataset
Download the reviewed records and inspect the codebook, rubric and checksums.
Output: Dataset and supporting files
-
Research & evidence
Benchmark Methodology
Inspect sample selection, collection, manual review and reproducibility limits.
Output: Published research method
-
Research & evidence
DUCR Score Explained
Read the four-category rubric and a worked record before using the totals.
Output: Rubric and worked example
-
Research & evidence
AI Crawler Access Study
Read historical access observations and distinguish allowed flags from actual crawler visits.
Output: Historical access observations
-
Research & evidence
Crawlable, Not Citable
Examine the gap between crawler access and recorded source-quality features in this sample.
Output: Benchmark interpretation
-
Research & evidence
Citation-Worthy Service Pages
Inspect the page features assessed in the benchmark and the limits of those observations.
Output: Page-feature findings
-
Research & evidence
Answer-Source Format
Read the published answer-source format and examine what the benchmark can support.
Output: Format and supporting context
-
Research & evidence
AI Citation Pilot
Inspect the dated pilot, retained answers, incomplete attempts and source-review limits.
Output: Pilot records and protocol
-
Research & evidence
Benchmark Limitations
Separate useful observations from questions this benchmark cannot answer.
Output: Scope and interpretation limits
-
Compare the same requirement, feasibility evidence and open questions before choosing the scale of work.
Output: Blank and worked CSV comparisons with a Markdown workbook
-
Agree first-release scope, content responsibilities and the questions still to settle.
Output: Blank and worked Markdown briefs with CSV page plans
-
Record what stays, inspect changed URLs and keep unfinished launch checks visible.
Output: Blank and worked CSV records with a Markdown workbook
-
Keep report scope, aggregation and unavailable values beside a worked interpretation exercise.
Output: Blank and worked CSV or Markdown
-
Template
GA4 Hostname Validation Workbook
Compare intended event origins, candidate hostnames and actual Testing evidence before activation.
Output: Blank and worked CSV or Markdown
-
Template
AI Bot Page-Role Record
Classify synthetic requests with an explicit page inventory, status checks and coverage limits.
Output: Blank and worked CSV or Markdown
-
Template
Traffic Drop Diagnostic Workbook
Separate metric observations, confirmed faults and unresolved causes, with a fictional two-page worked record.
Output: Blank and worked CSV or Markdown
-
Template
Cloudflare Crawler Access Record
Keep intended policy, request identity, blocking evidence and the next retest together.
Output: Blank and worked CSV or Markdown
-
Template
Monitoring Tool Trial Protocol
Prepare a common task and retain conditions, outputs and missing observations. This protocol has not been run.
Output: Markdown, CSV or JSON protocol
-
Template
AI Visibility Quote Worksheet
Keep quoted scope, payment terms, sources and unresolved costs together before comparing fees.
Output: Worked review and blank Markdown or CSV
-
Template
Business Fact Correction Packet
Compare an answer claim with its sources and keep ownership, proposed action and recheck separate.
Output: Worked record and blank Markdown or CSV
-
Twelve checks across the four DUCR layers, for one service page and one relevant supporting page.
Output: One-page PDF
-
Template
Page-Role Mapping Template
Give each page one job: owner, support or proof. Keep the decisions in a working map.
Output: Spreadsheet and printable canvas
-
Template
Topic Boundary Map Template
Draw the borders between pages before overlap spreads.
Output: Spreadsheet and printable canvas
No matching resources.
Try a different phrase, topic or resource type.
Behind the tools: methods and retained evidence
Read our published research, including the June 2026 AI Visibility Benchmark and the dated citation pilot, with its retained answers and collection limits.
The method is public. This is where it gets tested.
Every page on this site says how we work: a business has to be understood twice, by the machines that decide who gets found and by the buyer who arrives. Labs is where we check that it holds. Each test is written down before it runs, read against the data, and published whichever way it went.
Nothing here is a promise about your site. It is what we have seen on real ones, with the method next to it, so you can inspect the evidence, assess the reasoning or disagree with it.
How a finding gets here
- A question worth testing, from a client site, an Outside View or our own pages.
- The test, written before it runs: what we change or measure, what we expect to move, and when we read it.
- The reading. 14 days to settle, then 28 days of data against the 28 days before. Confirmed, refuted or unclear.
- The write-up: what we saw, what it means, what we would do next. Dated, with the method.
The same loop runs on every client site. See one experiment as we write it
This site is experiment one.
The first site the method runs on is our own. The hypothesis: a site built this way gets understood by Google and the AI assistants, found for what it sells rather than only for its name, and named when someone asks an assistant who to use. United States first.
The baseline was recorded on 29 September 2026, before the rebuild. The rebuilt site went live on 2 October 2026. The first reading comes after 14 days to settle, from 16 October; the 28-day comparison completes on 13 November 2026. The result goes here, whichever way it goes.
Experiment 001, as written
- Hypothesis. Built on the method, this site is found for AI SEO and the searches around it, not only for its name, and the assistants name it.
- The change. The whole site, rebuilt to the method: one job per page, the decision pages in the buyer's words, every claim with its source.
- What we watch. In Search Console: the searches that are not our name, and the pages they land on. In ChatGPT and Claude: whether we are named for the questions a buyer would ask.
- When we read. 16 October 2026, after the settling period. 13 November 2026: 28 days of data against the 28 days before the rebuild went live.
- Verdict. Pending.
The baseline and each reading are published here as they land.
The AI Visibility Benchmark.
Our June 2026 review of 50 service-business websites records public-page features under the DUCR model: Discoverable, Understandable, Citable, Routable, scored out of 100. Inspect the recorded findings, recalculate the published figures and see what additional evidence a recommendation needs. The corrected public export retains observations and scores while removing direct identifying fields.
The whole corpus, in reading order.
Edition 2026.06, in four lines
- 50 service-business websites, reviewed in June 2026.
- Median score: 52.5 of 100.
- Weakest layer: Citable, at a median of 4 of 30.
- The 50 records store allowed flags for the five named search crawlers. These are collector flags, not observed platform visits or citations.
What the method produced.
Results live on the Proof page, each with its period and its source: the Xporate case study, with figures from the client's own Search Console; an HR advisory firm's rebuild; and a manufacturer in Wales, unnamed by agreement. When a client agrees to be named, they are named. When they don't, the result is still shown, in the words we can stand behind.
If you read three things.
The three most useful places to start, whichever question brought you here.
- Missing from Google AI answers? Check eligibility before changing the site.
Check indexing, snippets and Google AI inclusion settings, then decide whether the page needs a change.
- Track ChatGPT traffic in GA4. Then check the enquiries.
Find recorded ChatGPT visits in the current GA4 reports, then check entry pages, completed actions and received enquiries.
- What does AI visibility work cost?
Software, advice or implementation, what changes the scope and what the fee covers.
Questions about Labs
What is Labs?
Our library of research, practical guides, browser tools and working templates. Research records carry dates, methods and limits; teaching examples are labelled.
Can I use the research?
You can inspect and download the AI Visibility Benchmark's dataset, codebook and rubric. Cite the edition and export date when referring to a finding. Contact us about reuse permissions; the dataset page explains the public files.
Do these findings apply to my business?
They are observations from real sites, not a forecast for yours. If you want to know where your own site stands, the Outside View says so, in writing, free.
How often does Labs change?
When an experiment finishes. Each write-up carries its date, and we update the ones that stop being true rather than leaving them up.
Where did Insights go?
Here. Insights became Labs in October 2026, and every article kept its address.
Can I suggest a test, or a market to benchmark?
Yes. Write to us with the question or the market, and say why it matters to you.