Increase Website GEO Traffic Step-by-Step Guide Tutorial: From AI Visibility to Real Visits

Quick Answer
This increase website geo traffic step-by-step guide tutorial explains how to earn more useful visits through generative engine optimization (GEO). Start by checking whether your pages can appear in AI search, then improve their answers with credible facts, original insights, and useful next steps. Measure AI visibility, website visits, and conversions separately. GEO here means generative engine optimization, not geographic traffic targeting. Neither indexing nor appearing in AI answers guarantees visitors.
Key Takeaways
- Check eligibility before rewriting content. An inaccessible or ineligible page cannot reliably compete for Google AI search visibility.
- Improve pages that already serve valuable questions. A specific, helpful answer is a better starting point than producing dozens of generic articles.
- Give people a reason to visit. Original comparisons, detailed examples, practical tools, and useful decisions provide value beyond an AI summary.
- Measure different outcomes separately. Google AI search impressions, website clicks, AI assistant referrals, and conversions describe different stages.
- Use current reporting tools. Google Search Console provides generative AI performance reporting, while Google Analytics 4 (GA4) identifies recognized AI Assistant traffic.
- Fix the actual bottleneck. An indexing problem, weak visibility, low referral traffic, and poor conversions require different responses.
- Review changes over time. A 30-day plan provides a manageable testing routine, not a guaranteed deadline for growth.
Step 1: Understand What GEO Traffic Actually Means
Imagine someone asks an AI search tool, 'How should I choose an indexing service for a new website?'
The system might summarize different approaches, explain their limitations, and provide links to relevant websites. If the person clicks one of those links, the destination may receive a visit associated with AI search.
That creates an opportunity for website owners. Instead of relying only on traditional search results, useful pages may also reach readers through AI-generated answers.
But four events must remain separate:
- Eligibility: A search system can access and consider the page.
- Visibility: The page appears in a relevant search experience or is referenced by an AI-generated response.
- Traffic: Someone follows a link and visits the website.
- Outcome: The visitor takes a useful action, such as submitting an inquiry or purchasing a product.
A page may be eligible without appearing. It may receive impressions without clicks. And it may earn visits that produce no business value.
This distinction matters because increasing AI mentions is not automatically the same as increasing website traffic.
GEO versus traditional SEO
Search engine optimization (SEO) improves how websites can be discovered, understood, and presented through search engines. Generative engine optimization (GEO) focuses on visibility and usefulness in AI-generated search and answer experiences.
The two approaches overlap substantially.
Google's official guidance states that established SEO practices remain relevant to its generative AI features. Accessible content, technical eligibility, usefulness, and clear website structure still matter.
Source: Google Search Central — Optimizing for generative AI features.
GEO is therefore not a reason to abandon ordinary SEO. It adds another discovery and measurement context.
One important limitation: Google's published requirements describe its own search features. They do not establish a universal citation-selection algorithm for ChatGPT, Claude, Perplexity, or every other AI assistant.
Step 2: Establish Your GEO Traffic Baseline
Before changing a website, determine what is happening now.
Without a baseline, an increase in reported impressions or visits may appear impressive even when it has little connection to your work.
Start with a small set of important pages rather than the entire website.
For most smaller sites, reviewing two or three pages provides a manageable starting point. That is a practical workload suggestion, not a proven optimal number.
Choose pages that meet three conditions:
- They answer a question someone might reasonably ask an AI search tool.
- They support an important business or informational goal.
- You can improve them with information you can verify.
For example, a company offering indexing services might select a page answering 'How can I check whether Google has indexed my website?'
That question is specific, useful, and closely connected to a measurable reader need.
A broad page about 'everything you should know about digital marketing' would be harder to evaluate because its audience, purpose, and possible outcomes are less clearly defined.
Record the starting numbers
For each selected URL, record the following information:
| Measurement | What it tells you | Where to look |
|---|---|---|
| Google index status | Whether Google reports the URL as indexed | Search Console URL Inspection |
| Google AI impressions | Reported exposure in supported Google generative AI experiences | Search Console generative AI performance report |
| Google AI clicks | Reported clicks from those experiences | Search Console generative AI performance report |
| Recognized AI Assistant sessions | Visits attributed to recognized AI assistants | GA4 Traffic acquisition |
| Engagement | Whether visitors interact meaningfully | GA4 engagement reporting |
| Conversions or key events | Whether visitors complete valuable actions | GA4 event and conversion reporting |
The metrics should be reviewed over comparable periods wherever possible.
For example, comparing a complete previous month with only the first three days of the current month would create a misleading impression.
Also record major promotions, content changes, tracking adjustments, and site outages. These events can influence the results independently of GEO improvements.
What changed in Google AI reporting in 2026?
Older GEO guides often say that Google AI search performance cannot be separated in Search Console.
That statement is outdated.
In June 2026, Google announced dedicated generative AI performance reporting for Search Console. The rollout was completed worldwide by August 31, 2026, according to Google's documentation.
These reports help website owners examine visibility and clicks associated with Google's generative search experiences.
Source: Google Search Central — Introducing Search Generative AI performance reports.
However, a reported impression still does not mean someone visited your website.
This is why the next step is to establish whether your content can participate before interpreting low visibility as a writing problem.
Step 3: Check Whether Your Pages Are Eligible for AI Search
Technical eligibility is the foundation of this process.
If a page is blocked, unavailable, or not selected for Google's index, improving its wording alone may not solve the problem.
Start with Google Search Console's URL Inspection tool for pages you own.
Check whether Google has indexed the intended URL. If not, investigate the reported reason instead of repeatedly requesting indexing.
Common issues include:
- A page returning an error or requiring a login.
- A
noindexdirective preventing index inclusion. - A robots.txt restriction affecting crawler access.
- Canonical signals indicating a different preferred URL.
- Important content missing from the HTML Google can process.
- A redirect or page configuration that does not match the intended destination.
A canonical URL is the address Google treats as the preferred version when similar or duplicate URLs exist.
For instance, if a website publishes the same article through several URLs, Google may choose a different canonical address from the one you are inspecting.
That needs investigation before you interpret the page as invisible to search.
How to check several URLs efficiently
If you need an initial index-status inventory, the IndexNowFast Google Index Checker supports bulk URL checks and CSV exports.
Its documentation describes a workflow in which you submit public URLs, review observed results, export a snapshot, and investigate failures.
The service also states that these results are observations rather than guarantees of permanent inclusion.
For websites you control, Search Console remains useful for detailed indexing diagnosis, including Google-reported information about the URL.
Remember that an index checker answers a narrow question: whether a URL appears indexed at the time of observation.
It does not tell you whether an AI assistant recommends the page or whether visitors are arriving through AI answers.
Check Google's generative AI participation setting
Google introduced a Search generative AI control in Search Console and completed its worldwide rollout on August 31, 2026.
For the relevant property, inspect:
Search Console → Settings → Search generative AI
The control can include or exclude site links and content from specified Google generative search features. Properties may also inherit the setting of a parent property.
If the website is excluded, the affected content cannot receive visibility or traffic from those Google features under that setting.
Do not change a deliberate exclusion without considering the site's publishing and privacy requirements.
Source: Google Search Console Help — Search generative AI control.
The setting applies to specified Google features, including AI Overviews and AI Mode. It does not control every third-party AI assistant.
What technical eligibility cannot promise
Fixing indexing or access problems makes a page more eligible for consideration. It does not require Google to display the page in an AI answer.
Similarly, using an indexing submission service cannot guarantee rankings, citations, or additional visits.
Google ultimately controls its own index inclusion. AI search systems may also select different sources depending on the question and context.
That is why the next part of the process focuses on what the page actually offers readers.
Step 4: Make Your Content Worth Referencing and Visiting
A technically accessible page still needs a compelling reason to be used.
The most useful starting point is to improve existing content that answers a meaningful question but lacks enough detail to help someone make a decision.
Answer the reader's question clearly
Suppose a page answers this question:
'Why is my website not showing in Google?'
A weak answer might say:
'Improve your SEO, create quality content, and submit your website to search engines.'
That advice is broad. It gives readers little help deciding what to check first.
A better answer would explain that a missing search result may reflect indexing, canonical selection, query relevance, or ranking rather than a single cause. It would direct website owners to inspect the URL in Search Console, check whether the intended page is indexed, and investigate any reported restrictions before changing content.
This answer is more useful because it distinguishes possible problems and gives readers a sensible next action.
It also creates an opportunity for deeper supporting material, such as troubleshooting examples or a decision worksheet.
Neither version guarantees that an AI system will cite the page. The difference is in the quality of assistance offered to the reader.
Add information an AI summary cannot fully replace
A general explanation is useful, but readers often need more to act confidently.
Consider adding:
- A clearly sourced comparison of relevant options.
- A worked example showing how to make a decision.
- A checklist based on documented requirements.
- A useful downloadable or copyable worksheet, when appropriate.
- Relevant original observations that your organization can substantiate.
- Important exceptions, limitations, and situations in which common advice does not apply.
For example, an article explaining indexed versus non-indexed pages could include a troubleshooting matrix connecting each observed condition with a diagnostic action.
A reader might visit the source to use the matrix, even if an AI system already summarizes the broad explanation.
That creates a more defensible reason to earn a click.
Support claims with credible sources
When making claims about Google Search behavior, link to Google's official documentation whenever possible.
For product specifications, use the provider's official documentation. For laws or regulations, use authoritative public sources. For market data, identify the actual report and its scope.
Avoid copying unsupported statistics from other blogs merely because they look persuasive.
If you do not know the answer, qualify the claim or identify the question as unresolved.
Google's generative AI optimization guidance emphasizes helpful, distinctive content and established search practices rather than special shortcuts.
Source: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
Make the information easy to understand
Clear headings, direct answers, descriptive link text, and readable tables help people navigate a page.
When a topic is complex, explain the basic conclusion first and place the technical details afterward.
For example, a beginner needs to understand that noindex can prevent a page from appearing in Google's index before learning the differences between HTML meta robots directives and HTTP response headers.
Avoid splitting straightforward answers into many tiny sections merely to imitate a supposed AI-friendly format.
There is no verified universal heading count, answer length, keyword density, or content template that guarantees GEO visibility.
Step 5: Strengthen Discovery Without Chasing GEO Tricks
Once the content is useful, make it easier for readers and search systems to find the right page.
Internal linking is an important part of that work.
Suppose a website has a detailed guide about diagnosing indexing problems and a separate page for checking index status.
A contextual link between them helps someone move from understanding the problem to checking a real URL.
The link should describe the destination accurately rather than use a vague phrase such as 'click here.'
A practical approach is to:
- Identify pages that naturally introduce the target subject.
- Add links where a reader would genuinely benefit from more detail.
- Use concise, descriptive anchor text.
- Confirm that linked pages are public, useful, and current.
- Remove or correct broken destinations.
Do not add irrelevant links simply to increase link counts.
Do you need llms.txt or special GEO markup?
For Google's generative search features, Google states that special AI-specific markup and an llms.txt file are not required.
That does not mean structured data is useless. Appropriate supported structured data can help eligible Google Search features when the page genuinely matches the relevant requirements.
But adding invented schema types or mass-producing AI-focused keyword variations is not a reliable substitute for useful content.
Google documents these distinctions in its generative AI optimization guide.
Other platforms may publish different crawler or discovery requirements. Evaluate those from each platform's actual documentation rather than assuming Google's controls govern the entire AI ecosystem.
Do backlinks or indexing submissions increase GEO traffic?
Not automatically.
A genuine editorial link can help readers discover a source, and traditional search systems have their own ways of evaluating links. However, no verified rule establishes that purchasing links or submitting backlink URLs will produce a particular AI citation or traffic outcome.
IndexNowFast's Backlink Indexer documentation explicitly separates processing and optional index checks from Google's final index inclusion.
This distinction matters: submission is an action, indexing is an observed search outcome, AI visibility is another outcome, and visits are another.
Treat any service promising guaranteed AI citations or GEO traffic from indexing alone with skepticism.
Step 6: Measure AI Visibility and Website Visits Separately
Measurement is where many otherwise useful GEO strategies become confusing.
Someone may see a website mentioned by an AI system, take a screenshot, and conclude that traffic has increased.
That conclusion is not supported by the screenshot.
A mention can demonstrate possible visibility in a particular response. It does not establish how many people saw that response, whether anyone clicked, or whether the resulting visitors converted.
Use different reporting systems to answer different questions.
Measure Google's generative AI search visibility
Open the appropriate property in Search Console and review the available generative AI performance reporting.
Look for impressions, clicks, and relevant page or query breakdowns supported by the reporting interface.
An impression records exposure according to Google's measurement rules. A click represents a reported interaction leading toward a destination.
Neither metric is identical to GA4 sessions.
Google's dedicated 2026 reporting update is explained here:
https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
If impressions rise while clicks remain flat, do not assume the website has gained traffic.
Instead, ask whether the page answers a relevant question and whether visiting the source offers a meaningful next step.
Measure recognized AI Assistant sessions in GA4
In May 2026, Google Analytics introduced an AI Assistant default channel for recognized assistant referrers.
This makes it easier to distinguish some assistant-referred visits from other website acquisition sources.
Start in GA4's Traffic acquisition reporting and inspect the AI Assistant channel where available.
Then review session source/medium, landing pages, engagement, and relevant conversion or key-event data.
Source: Google Analytics Help — What's new in Google Analytics.
Google's default channel group definitions explain how traffic is classified.
There is an important limitation: AI Assistant reporting does not capture every situation in which an AI-generated answer influences a visit.
Some people may copy a website address into their browser, return through a separate search, or arrive without a recognizable referrer. Those visits may be classified differently.
Therefore, zero reported AI Assistant sessions does not prove that nobody discovered the business through AI-generated answers.
Should you create a custom AI traffic channel?
Not necessarily.
First inspect the default AI Assistant channel and source/medium values. For many beginners, that is the simplest place to start.
GA4 also supports custom channel groups, which can organize traffic using matching rules.
A custom group can be helpful when an organization needs additional source classifications or more detailed reporting.
However, it introduces another configuration to maintain. Poorly designed rules may produce misleading comparisons or overlapping categories.
Do not simply add custom-channel totals to default-channel totals. Review the underlying sessions and classification rules first.
Why Search Console and GA4 numbers will differ
The systems measure different things.
Search Console reports Google Search activity. GA4 measures activity on a website through its own tracking and attribution system.
Their reporting scopes, attribution rules, dates, privacy conditions, and collection methods may differ.
A Google AI search impression and a recognized third-party assistant session should never be added together into a single 'GEO traffic' number.
Use each metric for the question it can answer.
Step 7: Use the GEO Traffic Diagnosis Matrix
The following matrix connects observed conditions with the most useful next action.
It is an editorial decision framework, not an experimental finding or a prediction of Google's ranking behavior.
| What you observe | What it may mean | What to do next |
|---|---|---|
| Important page is not indexed | Google Search eligibility may be limited | Inspect indexing reports, canonical signals, access and noindex directives |
| Page is indexed but receives little AI visibility | The page may not be selected for relevant AI answers | Review reader-question fit, factual usefulness and reporting scope |
| Google AI impressions increase, but clicks do not | Visibility may be improving without additional visits | Assess the reader's reason to visit and relevant destination content |
| AI Assistant sessions increase, but conversions remain weak | Acquisition may not be producing useful business outcomes | Review landing-page relevance, user experience and conversion paths |
| GA4 reports no AI Assistant sessions | No recognized sessions appear in the selected report | Review source/medium classification and tracking limitations |
| Several metrics improve after multiple simultaneous changes | The combined outcome may be positive, but causation is uncertain | Record changes, identify confounders and avoid attributing results to one edit |
The table helps avoid a common mistake: prescribing more content whenever GEO traffic looks weak.
If the real problem is a noindex directive, publishing another article does not fix the affected page.
If the page already receives visitors but few inquiries, the next priority may be a better landing-page experience rather than additional AI exposure.
A worked measurement example
Consider a hypothetical business page that begins receiving more impressions in Google's generative AI performance reporting.
At the same time, GA4 shows no increase in recognized AI Assistant sessions.
What can the owner conclude?
They can say that reported Google generative search visibility increased during the selected period.
They cannot say that total GEO traffic increased simply because impressions went up.
They also cannot assume that a particular content edit caused the impression change. Other factors, such as changes in search demand or Google's presentation of results, may affect reporting.
A sensible next step is to inspect Google AI clicks separately, review the page's usefulness for the next reader decision, and monitor the relevant website acquisition data.
This example uses no invented performance numbers. It illustrates how to interpret the relationship between reported exposure and visits.
Step 8: Build a Reusable GEO Traffic Worksheet
A worksheet turns GEO from a vague marketing goal into a set of observations and decisions.
You can maintain it in a spreadsheet or another simple reporting system.
Use one row for each page and measurement period, or for each documented change when detailed tracking is needed.
Here is a practical structure:
| Field | What to record |
|---|---|
| Observation date | Date the measurement was collected |
| Page URL | Exact public destination |
| Target reader question | The specific question the page should answer |
| Target platform | Google AI search or a particular AI assistant |
| Baseline period | Dates used for comparison |
| Change date | When an improvement was published |
| Change type | Technical repair, content update, internal link or other action |
| Index status | Available Google indexing observation |
| Google AI impressions | Reported impressions for the relevant scope, if available |
| Google AI clicks | Reported clicks for the relevant scope, if available |
| AI Assistant sessions | Recognized assistant-referred sessions in GA4 |
| Source/medium | Available referral classification |
| Engagement | Relevant GA4 engagement measurement |
| Key events or conversions | Completed actions that matter to the website |
| Other changes | Promotions, tracking edits, outages or major site changes |
| Interpretation | What the observations support and what remains uncertain |
| Next action | The specific decision informed by the evidence |
The worksheet does not need every possible metric on the first day.
Start with fields that can be measured reliably, then expand when they help answer a decision.
How to calculate a change accurately
When two measurements represent the same metric and reporting scope, an ordinary percentage change can be useful:
Percentage change = ((new value − old value) ÷ old value) × 100
For example, moving from 100 to 120 comparable sessions would represent a 20% increase.
This is an illustrative arithmetic example, not a reported performance result for any website.
The formula cannot be used when the baseline is zero. In that case, report the change in absolute numbers rather than an undefined percentage increase.
Also avoid calculating a single growth percentage from unrelated units such as impressions and sessions.
For conversion quality, compare the same kind of conversion under consistent event definitions and measurement periods.
Most importantly, label an observed improvement as an observation unless you have enough evidence to support a causal claim.
Step 9: Follow a Practical 30-Day GEO Improvement Plan
Thirty days is a useful planning window because it is long enough to organize a focused cycle of work and review.
It is not a verified promise that AI traffic will grow within a month.
Search systems can take different amounts of time to discover, process, and surface changes. Traffic levels and reporting availability also vary.
Days 1–3: Select pages and capture the baseline
Choose two or three pages with clear reader intent and business value.
Record the target questions, existing index status, available AI search impressions and clicks, recognized assistant sessions, and important conversions.
Confirm that the analytics events you intend to use are configured and collecting data.
The goal is to understand the starting position before making meaningful changes.
Days 4–7: Resolve eligibility problems
Check each selected page in Search Console.
Investigate indexing exclusions, conflicting canonical signals, crawl restrictions, redirects, and missing content.
Review Google's generative AI participation setting if it is relevant to the property.
Prioritize genuine blockers. Do not spend this stage repeatedly submitting URLs that remain technically ineligible.
Days 8–14: Improve the content
Review how well each page answers its target question.
Make the opening answer clearer. Correct outdated facts. Add a genuinely useful comparison, decision table, example, or explanation where the existing content falls short.
Use authoritative supporting sources and add relevant internal links.
Keep a record of what changed and when.
Avoid changing many unrelated elements at once if you want a clearer basis for later interpretation.
Days 15–21: Confirm changes and observe reporting
Check that the revised pages are live and functioning.
Inspect their rendered content, internal links, and relevant index information.
Review available reporting without treating early fluctuations as proof of success or failure.
Record major external factors, such as campaigns, seasonal interest, site incidents, or broader changes to the website.
Days 22–30: Compare evidence and choose the next action
Use comparable reporting periods where possible, accounting for delays and incomplete data.
Ask:
- Did the page become eligible or remain technically healthy?
- Did Google generative search visibility change?
- Were there additional reported clicks?
- Did recognized assistant-referred sessions change?
- Did engagement or conversions improve?
- Are alternative explanations plausible?
Then select the next bottleneck.
For one page, that might mean improving a weak answer. For another, it might mean diagnosing tracking or fixing a conversion path.
If the data is sparse, extend the observation period instead of announcing an unsupported result.
The purpose of the 30-day cycle is better decisions, not a predetermined success story.
Step 10: Avoid Common GEO Traffic Mistakes
Many GEO recommendations sound attractive because they promise a simple shortcut to a complex outcome.
The most costly mistakes usually come from confusing different stages of discovery or applying a tactic without identifying the actual problem.
Mistake 1: Treating every AI mention as traffic
A visible citation is not a website session. Measure clicks or visits where reporting permits, and keep manual visibility observations separate.
Mistake 2: Publishing large quantities of similar answers
Producing more pages does not automatically create more useful coverage. If several pages repeat the same claims without new information, they may add little value for readers.
Prioritize meaningful questions and distinctive content.
Mistake 3: Ignoring indexing and access problems
An excellent article can still face discovery problems when important content is inaccessible or Google does not select the intended URL for indexing.
Inspect technical eligibility before investing heavily in cosmetic GEO changes.
For broader indexing and crawlability education, the IndexNowFast SEO indexing guides provide related reading.
Mistake 4: Assuming all AI platforms work like Google
Google publishes requirements for its own search experiences. Other AI systems have different capabilities, access arrangements, and documented controls.
Do not turn a Google-specific recommendation into a universal rule.
Mistake 5: Promising results from indexing submissions
A submitted URL is not necessarily indexed. An indexed URL is not necessarily referenced in AI search. And an AI reference does not necessarily lead to a visitor.
Track each outcome honestly.
Mistake 6: Declaring victory after a short-term spike
A brief increase may reflect changing demand, a campaign, measurement adjustments, or search presentation changes.
Review more than one metric, preserve the change history, and avoid attributing results to a single edit without adequate evidence.
Frequently Asked Questions
What is GEO traffic?
GEO traffic commonly refers to website visits associated with generative AI search and answer experiences. In this guide, GEO means generative engine optimization. It does not mean visitors from a particular country or geographic region.
Can GEO actually increase website traffic?
GEO improvements may help useful pages reach people through AI-assisted discovery. Whether that exposure produces additional visits depends on the platform, query, content, user behavior, and link opportunities. No specific traffic increase is guaranteed.
Does GEO replace traditional SEO?
No. Technical eligibility, helpful content, clear structure, and other established SEO practices remain important. Google explicitly connects its generative AI optimization guidance with conventional search fundamentals.
How do I identify AI traffic in Google Analytics?
In GA4, begin with Traffic acquisition and look for the AI Assistant default channel. Review session source/medium and landing-page performance as needed. Only recognized and appropriately attributed sessions are captured by that classification.
Can Search Console show Google AI search performance?
Yes. Google introduced dedicated generative AI performance reporting in 2026. It provides relevant visibility and click measurements, but those numbers should not be treated as identical to GA4 sessions or third-party assistant referrals.
Why might my website have AI visibility but no AI Assistant sessions?
The measurements cover different sources and activities. Google's reported AI impressions do not necessarily produce clicks, while GA4's AI Assistant channel measures recognized assistant referrals. A lack of recognized sessions does not establish that no AI exposure occurred.
Does my website need llms.txt to appear in Google AI search?
Google does not require an llms.txt file or special AI-specific markup for its generative search features. Follow Google's current technical and content guidance. Evaluate separate platform requirements individually rather than assuming all AI services use identical rules.
Will getting a page indexed make an AI assistant cite it?
No. Index inclusion can be an important Google Search prerequisite, but it does not guarantee AI visibility, citations, rankings, or visits.
How long does it take to increase GEO traffic?
There is no verified universal timeline. The 30-day process in this guide is a manageable implementation and measurement cycle, not a promise of measurable traffic growth. Continue observing when data is incomplete or sparse.
Conclusion: Improve the Right Stage, Not Just the Content
Increasing website GEO traffic requires more than adding AI-related keywords or seeking citations.
The practical route is to identify valuable reader questions, verify that relevant pages are eligible for discovery, improve their answers, and measure whether reported exposure leads to useful visits.
The most important habit is separating eligibility, visibility, traffic, and conversions.
When those outcomes are measured independently, the next action becomes clearer. A page with an indexing problem needs a technical fix. A visible page with few clicks may need stronger next-step value. A page attracting visits without conversions may need a better user experience.
Start with one important page, record its baseline, make a well-supported improvement, and use the GEO Traffic Diagnosis Matrix to decide what to address next.
That repeatable process is more useful than any promise of guaranteed AI traffic.