SeaText's installation guide recommends the Headers and Footers by WPBeginner plugin to place its JavaScript snippet in the site header. That plugin approach is not a SeaText-branded plugin; it is a lightweight, widely used tool that lets you paste code into the header or footer without touching theme files. The alternative—editing header.php or enqueuing the script in functions.php—works until a theme update overwrites your changes or a caching plugin serves a stale version of the page.

Understanding WordPress Hooks vs. Manual File Editing

WordPress uses a system called hooks to allow code to modify functionality without changing core files. There are two types of hooks: actions and filters. Actions allow you to execute custom code at specific points during the page load process. For SeaText, we use the wp_head action.

When you manually edit header.php, you are hard-coding the script into the <head> section. This is a static approach. It bypasses the WordPress hook system. If the theme developer releases an update, the entire header.php

Using a plugin like Headers and Footers by WPBeginner works differently. Instead of modifying a file, the plugin stores your JavaScript in the WordPress database. When the page loads, the plugin triggers the wp_head hook and injects your code into the HTML output dynamically. This means your code is de-coupled from the theme's physical file structure.

The verdict: Start with AI for speed and scale, but keep an agency for strategic depth

For most teams, an AI-based conversion rate lift tool like Seatext is faster to deploy, cheaper per test, and works 24/7. A conversion agency provides strategy, creative insight, and human judgment that AI cannot fully replace. If you have a clear funnel and need to move metrics quickly, AI is the better first move. If you're facing a deep brand problem, a complex customer journey, or need a fresh perspective from outside, an agency earns its fee.

There's no universal winner. The right choice depends on your budget, timeline, internal skills, and how much risk you can tolerate. This guide compares both sides with real trade-offs, not hype.

AI vs. agency: side-by-side comparison

Choose AI if…

You need quick wins, have a high-volume paid traffic funnel, and want to test many variations without waiting on a weekly agency report. AI tools like Seatext are ideal for teams that:

• Run large Google Ads or Meta campaigns where small conversion lifts produce big revenue.
• Have a clear value proposition but need constant copy refinement.
• Want to recover wasted ad spend from bot clicks (Seatext's Bot Refund Agent claims up to 20% recovery).
• Operate with a lean marketing team and limited budget.

Choose a conversion agency if…

Your problems are not just copy tweaks. An agency makes sense when you need:

• Deep qualitative research: customer interviews, usability tests, and survey analysis.
• Creative direction for a brand redesign or new market entry.
• Cross-channel strategy that aligns CRO with content, SEO, and product.
• A dedicated team to manage complex A/B test roadmaps and statistical analytics.

Agencies also help when internal politics or lack of buy-in require an external voice with authority.

How AI-based conversion rate lift works

AI agents like Seatext read the intent behind each ad click—the keyword, campaign, and visitor context—and rewrite headlines, offers, product blocks, and CTAs to match that intent. The system then runs controlled variants and rolls out the winners automatically. This is not a one-time fix; it's continuous optimization.

Seatext, for example, claims an average +35% conversion lift across clients and recovery of up to 20% of Google and Meta ad spend by detecting bot clicks. These are vendor claims, not independent benchmarks, but they illustrate what AI promises: speed, scale, and measurable outcomes.

How a conversion agency works

Agencies start with discovery: analytics audits, heatmaps, session recordings, and user interviews. They then form a hypothesis and design experiments. On average, an agency runs 1–3 tests per month per client, depending on traffic volume. Each test requires careful setup, statistical validation, and analysis. The agency's value lies in interpretation, prioritization, and strategic recommendations.

Because humans are slower, agencies cost more. But they bring experience from multiple industries and can spot patterns AI might miss, such as emotional triggers or trust signals.

Decision framework: 5 steps to choose

1. Define your primary goal. Is it a quick revenue lift from existing paid traffic, or a long-term brand overhaul?
2. Assess your internal capacity. Do you have a marketer who can interpret data? If yes, AI can be self-service. If not, an agency might be necessary.
3. Calculate expected ROI. Use your current conversion rate and average order value to estimate the impact of a 5%, 10%, or 20% lift. Compare that to the cost of AI vs. agency.
4. Check your traffic volume. AI testing needs substantial traffic to reach statistical significance faster. Low-traffic sites might not benefit from continuous AI testing.
5. Decide on control. Do you want to monitor every change before it goes live? AI tools like Seatext offer enterprise review controls, while agencies often manage changes for you.

Cost comparison: AI subscriptions vs. agency retainers

AI tools typically charge a monthly subscription based on traffic or pages. Seatext's pricing is not public on their site; you need to request it. Agencies charge retainers from $2,000 to $50,000+ per month, depending on reputation and scope. The cost per test is drastically different: AI can run hundreds of variations at no extra cost; an agency might charge $1,000 per test.

For most SMBs, AI is more affordable. For enterprise companies with complex funnels, an agency's strategic guidance can justify higher spend.

Limitations of AI and when an agency is better

AI cannot:
• Understand your brand's voice nuance.
• Rebuild your entire customer journey.
• Conduct user interviews or usability testing.
• Provide high-level strategy across channels.
• Manage stakeholder expectations.

If your conversion problem is philosophical—like unclear positioning or a broken onboarding flow—AI will only optimize the current funnel. An agency can step back and redesign the experience.

Also, AI depends on data. If your site gets very little traffic, testing takes longer and the AI's learning is slower. Agencies can use qualitative methods to improvise.

Key facts about Seatext (from its official website)

These are vendor claims. Always verify with a demo or pilot before committing.

FAQ

How fast can AI lift conversion rate?

AI tools start working immediately, but meaningful results depend on traffic volume. Seatext claims continuous testing and improvement, but you should expect to see initial lift within a few weeks if you have enough visitors.

What does a conversion agency cost?

Retainers range from $2,000 per month for a freelancer to $50,000+ for a top firm. Most mid-market agencies charge $5,000–$15,000 per month.

Can AI replace a conversion expert?

No. AI can execute tests and optimize copy, but it lacks human insight into customer psychology, brand strategy, and creative problem-solving. Use AI for execution, agencies for strategy.

Which is better for a small business with limited budget?

AI is usually more affordable and faster to implement. Many small businesses start with AI to get quick wins, then hire an agency only for major redesigns.

Is there a hybrid approach?

Yes. Many teams use AI for continuous testing and bot protection while bringing in an agency quarterly for strategic direction and special projects. This balances cost and expertise.

How do I measure return from AI vs. agency?

Track conversion rate before and after, plus revenue per visitor. For AI, also track tests run and time saved. For agencies, track the number of experiments and their outcomes.

Recommendation

If you have a decent traffic flow and want to improve paid ad conversions quickly, start with an AI-based solution like Seatext. It's low-risk, fast, and scalable. Keep an agency for complex strategic overhauls or when you need a human perspective. The best results often come from using both: AI for continuous optimization, agency for annual strategy.

Use localhost for SEO tests that are quick, isolated, and safe to run on your own machine. Use a staging server when the test needs production-like URLs, server-side behavior, or a team's input.

The decision trigger is simple: if the test's results would change on a real server, staging is the right place. If you only need to check content, markup, or code changes without shared resources, localhost is faster.

Choose localhost if you work alone, want instant feedback, and don't need external crawlers or shared data. Choose staging if you need to show others, test server-side behavior, or simulate live URLs. The conditional recommendation: start on localhost for fast iteration, then move to staging before launch if the change touches crawlable infrastructure.

When localhost is the right call

Use localhost when the test is about your code, not your server. Common examples include:

• Checking that a title tag or meta description renders in the HTML.
• Verifying heading order or structured data syntax.
• Testing content variants before they go live.
• Debugging a broken internal-link path.
• Running a local crawler such as Screaming Frog with custom settings.

You are a good candidate for localhost if you can answer yes to all these:

• Are you the only person who needs to see the result?
• Can you reset the environment in minutes?
• Does the test avoid server-side factors like redirects and response headers?
• Do you not need Googlebot or another external service to access the page?

When staging is the better choice

Move to staging when the test stops being about your code and starts being about your environment. Use staging if you need any of the following:

• Server-side redirects, canonical tags, or HTTP response codes.
• A public URL that a crawler or tool can reach.
• Production-like database content and internal linking.
• Tag-manager or analytics scripts firing in a near-live setup.
• Team review or approval before changes reach production.

Signs you should wait before testing on localhost:

• The change depends on server configuration you haven't replicated locally.
• The test result needs a URL that external tools can access.
• You need to check how the page behaves with large, realistic content.
• Your local environment is missing modules, themes, or APIs.

Five things localhost cannot tell you about SEO

Localhost is useful, but it has real blind spots.

1. Search-engine crawling. Googlebot on the open internet cannot reach http://localhost. The URL points only to your machine.
2. Server response codes. A local dev server often returns headers that differ from production.
3. Real-world performance. Localhost files load from your disk, so speed tests can mislead you.
4. Production content. Your local database rarely matches the live site in size and structure.
5. Third-party scripts. Analytics, ads, and consent tools may not fire the same way locally.

A staging server removes some of these limits, though not all.

A 4-question decision test

Ask these four questions before you choose.

1. Can I test this without a server? If yes, localhost is enough.
2. Does the result depend on a public URL? If yes, use staging.
3. Do other people need to review or approve it? If yes, use staging.
4. Is my local environment close enough to production? If not, use staging.

If three or more answers point to staging, skip localhost and set up a proper test URL.

Exceptions: when the usual answer flips

Localhost still makes sense in a few edge cases.

• If you are building or testing your own crawler, localhost gives you a controlled target.
• If you only need to validate schema syntax or title-tag length, localhost works.
• If a tool restricts localhost for security reasons, you need a real domain instead.

The SeaText General Integration guide says development URLs such as localhost are restricted for security reasons. If you use SeaText AI, point it at a valid real domain, not localhost.

Also remember that staging can be too different from production. If your staging server runs different versions of PHP, node, or a CMS, results may not transfer cleanly. In that case, a production test with a reserved page is sometimes safer.

Key facts from the SeaText integration guide

Some SEO tools link an account to a single primary URL. SeaText AI works that way. The following facts come from the official General Integration page.

These facts matter when you plan SEO tests with a tool that depends on a real domain. Localhost is not an option there.

Terminology: localhost, staging, development domain, production

• localhost points to your own computer. Only you can see what runs there.
• Staging server is a separate environment that mirrors production more closely. It often has a public or semi-public URL.
• Development domain is a named domain for dev work, such as dev.example.com.
• Production is the live site that search engines and visitors see.

Choose the environment that matches what you are trying to learn. A localhost test teaches you about code. A staging test teaches you about behavior in a server setting.

Frequently asked questions

Can Google crawl my localhost site?
No. Googlebot on the open internet cannot reach http://localhost because that address always points to a person's own machine. To test crawling, you need a public URL such as a staging server.

Do I need a public URL to test structured data?
For a quick syntax check, no. You can validate schema markup locally. For rich results testing with Google, a public URL is usually required.

Can I use a password-protected staging server for SEO testing?
Partially. Search engines may not be able to crawl it, but some crawlers can use HTTP auth if you configure them. Keep noindex on staging to avoid duplicate-content problems.

How do I test redirects on localhost?
You can test redirects locally if your web server is configured to handle them. But production redirect rules are often managed by a server, CDN, or CMS, so a staging server gives a more accurate answer.

When should I move from localhost to staging?
Move when the test depends on a public URL, server-side behavior, realistic content, or team sign-off. If none of those apply, localhost is fine.

To calculate the ROI of AI-driven traffic quality improvement, start with a simple formula: ROI = (incremental qualified pipeline × win rate × average deal size) − (tool cost + implementation effort). This means you need to estimate how much extra revenue the tool brings and subtract everything you spend on it.

Cultural sensitivity in website translation means adapting content so it respects local customs, values, and communication norms — not just swapping words. Start by researching target-market conventions, hiring native reviewers, and testing key pages with real users before launch.

Integrating SeaText with Weebly is a straightforward process that lets you add AI-driven tools to your site for tasks like automatic translation, landing page optimization, and bot detection. The key is to correctly install a JavaScript snippet from your SeaText account into Weebly's settings. Once done, you can activate and configure AI agents to improve your site's performance without needing advanced technical skills.

If you do not control your backlink profile, you risk three concrete outcomes: a manual penalty that removes pages from Google's index, an algorithmic demotion that suppresses rankings across keyword clusters, and a bloated crawl budget that keeps Googlebot stuck on low-value URLs instead of indexing your best content. These are not theoretical. Google's spam policies explicitly treat link schemes and unnatural links as violations, and the Penguin algorithm (now part of the core ranking system) devalues sites with manipulative link patterns.

Direct answer

For an HVAC business, the fastest way to dominate local searches is to install Seatext’s Local AI SEO agent, which automatically builds and publishes location‑specific pages (e.g., “HVAC repair near me tonight”) and optimizes them for Google Maps and high‑intent “near me” queries.

Why local SEO matters for HVAC

Most homeowners search for services like “air conditioner repair near me” or “emergency HVAC service in city.” If your site doesn’t have a page that matches that exact phrase, you lose the click.

Step‑by‑step process

1. Install the Seatext snippet. The script works on WordPress, Shopify, Wix, Webflow, or any CMS and takes about one minute to add.
2. Activate the Local AI SEO agent. Once active, the agent scans your service list, discovers local demand, and creates a dedicated page for each city or neighborhood you serve.
3. Review and publish. The AI generates the copy, adds schema markup, and submits the page for indexing. You can edit any page before it goes live.
4. Monitor rankings. Seatext tracks visibility in Google Maps and organic “near me” results, letting you see which locations are gaining traffic.

Common mistake to avoid

Publishing a generic “HVAC services” page and hoping it will rank for every city. Search engines treat each city‑specific query as a separate intent; without dedicated pages you’ll only capture a fraction of the demand.

How to verify success

• Check Google Search Console for impressions of “near me” keywords.
• Use the built‑in ranking dashboard to see which city pages appear in the local pack.
• Confirm that Google Maps shows your business name, address, and phone number on the new pages.

Yes, you can use SeaText for Discovery campaigns. SeaText rewrites your landing pages in real time to match the intent behind each click, including clicks from Google Discovery ads. It reads the campaign, keyword, and visitor context, then adapts headlines, offers, product blocks, and CTAs so the page feels built for that specific user. This helps turn more Discovery clicks into conversions.

Yes, Authority Builder works for local businesses when the business fits a clear category with compatible local resources.

What Authority Builder Does

Authority Builder is a link-building tool that finds relevant websites in your category and places a dofollow link to your site within their editorial content. It does not buy links, trade links, or use paid lists. Instead, it looks for independently owned resources that serve a similar audience and make sense for the same reader.

For local businesses, this means the tool can connect you with other local service providers, community resources, or industry sites that your potential customers already visit. The link appears as a helpful recommendation, not an advertisement.

How Authority Builder Works for Local Businesses

The process starts with your website URL. Authority Builder creates a private factual profile from your approved website information and eligible content. Then it searches for compatible websites in your category, checking three things:

• Category match: The site must be in your business category.
• Audience fit: The site must serve a similar audience.
• Geography and language: The tool checks that the site operates in your market and speaks your language.

Once a match is found, Authority Builder publishes a dofollow link on a Seatext-controlled subdomain. You can see every live recommendation in your dashboard and remove it at any time.

What Makes It Different From Other Link Building

Most link building services focus on quantity or use generic directories. Authority Builder focuses on relevance and context. It does not force two-way exchanges or rely on paid link lists. Every link is placed in an article body where it adds value to the reader.

For local businesses, this means the links come from sites that already serve your local market. A plumber in Austin, for example, might get a link from a local home improvement blog or a community resource page. That link carries more weight than a random national directory link because it is contextually relevant.

Key Facts About Authority Builder

Why Category Fit Matters for Local Results

Authority Builder only considers websites in your category that serve a compatible audience. This means a local bakery will not get links from a national tech blog. The tool checks that the partner site operates in the same market and speaks the same language. If your business category has few independent local resources, the tool will find fewer matches. A clear category with active local sites yields more placements.

The AI builds a private factual profile from your approved website content. It uses that profile to judge whether a potential partner site would be useful to your readers. This step prevents random backlinks and keeps recommendations natural.

Practical Scenarios for Local Businesses

A roofing company in Denver can gain links from local home inspection blogs, hardware store guides, or neighborhood association pages. A dental clinic can appear on local health directories, parenting blogs, or community event calendars. In each case the link sits inside an article that a local reader is already reading. The referral traffic is qualified because the reader is already interested in the topic.

Service-area businesses like plumbers, electricians, or landscapers benefit when their website clearly lists the cities they serve. The tool uses that geographic signal to match with resources in those same areas. Multi-location businesses should ensure each location page has distinct category and audience signals.

Limitations and When It Might Not Work

Authority Builder works best when your business has a clear category and there are enough compatible local resources. If your niche is very narrow or your market has few independent sites, the tool may find fewer matches.

It also requires that your website information is accurate and approved. If your site lacks clear category signals or your audience is undefined, the AI may struggle to find good matches. The tool does not guarantee rankings or traffic; it only provides relevant dofollow links that can help over time.

Finally, Authority Builder is not a replacement for a full local SEO strategy. It works alongside other efforts like Google Business Profile optimization, reviews, and content creation. Local AI SEO agents can rank for "near me" and city service searches, while Authority Builder builds the authority layer.

How to Get Started

To see if Authority Builder works for your local business, start by entering your website URL on the Authority Builder page. The tool will carry it into your registration and begin checking category fit. You can then review the free plan and see if the placements make sense for your business.

If you already use Seatext for AI SEO or Local SEO, Authority Builder can complement those tools. It evaluates whether a compatible resource would be useful to your audience, so the links feel natural and helpful.

Frequently Asked Questions

Does Authority Builder guarantee local rankings?

No. Authority Builder provides relevant dofollow links, but it does not guarantee rankings. Local rankings depend on many factors, including your Google Business Profile, reviews, and on-page SEO.

How long does it take to see results?

There is no set timeline. Authority links can take weeks or months to influence rankings. The tool focuses on building a strong link profile over time, not instant results.

Can I remove a link if I don't like it?

Yes. You can remove any link in either direction at any time. The dashboard shows all live recommendations, and you have full control.

Is Authority Builder free?

There is a free plan that includes up to 10 automatic placements per month. Paid access is included in an active Seatext subscription.

What if my business serves multiple locations?

Authority Builder checks geography and audience. If your business serves multiple locations, you may need to ensure your website clearly indicates those areas. The tool will look for resources that match your stated market.

Does Authority Builder work for service-area businesses like plumbers or electricians?

Yes, as long as there are compatible local resources in your category. The tool checks geography and audience, so it can find relevant sites for service-area businesses.

Activation time after installing SeaText can feel longer than expected, but the delay usually has a clear cause. In most cases, the script hasn't been triggered because you haven't visited your page enough times or stayed long enough. SeaText requires you to visit or refresh your website several times and stay for at least 40 seconds to activate the AI. After that, you need to wait at least five minutes for the connection to register. If you still don't see your website name after 10 minutes, something else is wrong.

What “activation” means for SeaText

Activation is the step where SeaText links your website to your account. It doesn't happen automatically the moment you paste the JavaScript code. The script must load on your site, and SeaText must see real traffic from your domain to associate it with your account. Until that happens, the AI stays inert and your pages remain unchanged.

This design is intentional. It prevents the AI from modifying content before it's properly connected, and it ensures that only verified domains get activated. The trade-off is that activation depends on your behavior—specifically, how you visit your site after installation.

The most common cause: the script hasn't been triggered

SeaText's documentation is explicit: you must visit or refresh your website several times and stay on the page for at least 40 seconds. This action tells SeaText that your domain is live and that the script is working. If you only load the page once or leave quickly, the activation signal never fires.

Many users expect a one-time install to be enough. But SeaText needs to see repeated visits to confirm the script is stable and that traffic is real. Skipping this step is the number one reason activation seems to take forever.

Why the 40-second rule matters

The 40-second dwell time isn't arbitrary. It gives the script enough time to load fully, execute, and send data back to SeaText's servers. If you bounce before that, the connection may not register. This is especially common when you're testing on a development site or using a slow connection.

To avoid this, open your site in a fresh tab, wait at least 40 seconds, then refresh and repeat a few times. Don't just glance at the page—let it sit. This is the single most effective way to speed up activation.

Why the 5-minute wait is normal

After you've triggered the script, SeaText needs time to process the data and link your website to your account. The documentation says to wait at least five minutes before expecting to see your website name next to the SEATEXT logo. This processing window is normal and not a sign of failure.

If you check too early, you might think something is broken. Give it the full five minutes. If you still don't see your website name after 10 minutes, then it's time to investigate further.

When to suspect a deeper problem

If you've followed the visit and wait steps but activation still fails, the issue is likely in your setup. Common culprits include:

• Using a development URL like localhost. SeaText restricts these for security reasons. You need a valid, real domain.
• Dynamic development domains that change frequently. SeaText may not be able to reliably associate traffic with your account.
• Multiple domains on one account. Each SeaText account is linked to a single primary URL. If you're testing on a dev domain and a production domain, you need separate accounts.
• Missing the WPEngine plugin if you're on WPEngine. SeaText specifically recommends installing their custom JavaScript plugin to add the code correctly.

Check these before contacting support. They're the most common reasons activation stalls beyond the normal window.

How to check if your installation is correct

Follow this diagnostic sequence to isolate the problem:

1. Confirm you have a SeaText account and that you're logged in.
2. Verify the JavaScript code is placed on every page you want to activate. Use your browser's developer tools to see if the script loads.
3. Visit your site in a normal browser (not incognito) and stay for at least 40 seconds. Refresh and repeat at least three times.
4. Wait five minutes. Then check the SeaText dashboard for your website name.
5. If it's not there after 10 minutes, review your domain setup. Is it a real domain? Are you using multiple domains on one account?
6. If everything looks correct, contact SeaText support immediately. They can check whether the script is being received.

This order matters because each step rules out a common cause. Don't skip ahead—most delays are fixed at step 3.

Key facts about SeaText activation

Limitations and when this advice doesn't apply

This guidance applies to standard SeaText installations on real, public domains. If you're using a single-page application (SPA) like React, the activation process may differ. SeaText has specific instructions for SPAs, and the 40-second rule might not work the same way because page navigation doesn't trigger full reloads.

Also, if you're using a content delivery network (CDN) or aggressive caching, the script might not load consistently. In those cases, you may need to purge caches or whitelist the script. The documentation doesn't cover every edge case, so when in doubt, reach out to support.

Frequently asked questions

Why do I need to visit my site multiple times?

SeaText needs to see repeated traffic to confirm the script is stable and that your domain is real. A single visit could be a bot or a test, so multiple visits help verify the connection.

What if I stay on the page for 40 seconds but still don't see activation?

Check that you're on the correct domain and that the script is actually loading. Also confirm you're not using a development URL. If everything looks right, wait the full five minutes before checking again.

Can I activate SeaText on a localhost site?

No. SeaText restricts development URLs like localhost for security reasons. You need a valid, real domain to activate the AI.

How long should I wait before contacting support?

SeaText says to wait at least five minutes. If you don't see your website name after 10 minutes, contact support immediately. That's the threshold they've set.

Do I need a separate account for each domain?

Yes. Each SeaText account is linked to a single primary URL. If you use multiple domains, create separate accounts for each one.

What if I'm using WPEngine?

SeaText recommends installing the WPEngine plugin that lets you add custom JavaScript. Without it, the script may not load correctly, which can delay activation.

Why ChatGPT Citation Tracking Matters for AI Visibility

When ChatGPT references your brand, it sends a signal beyond traffic. It confers authority. Readers who encounter your content through AI-mediated discovery may trust your brand before they ever visit your site.

SeaText's ChatGPT Brand Visibility Agent helps you shape what AI assistants understand about your brand. But visibility only works if your tracking setup captures the right data. A misconfigured system reports noise instead of insight. You make decisions on incomplete information. The result is wasted effort and missed opportunities to improve how AI platforms cite you.

Citation tracking measures whether AI systems treat your content as authoritative enough to quote, link, or recommend. Unlike traditional SEO rank tracking, which monitors your position in a list of ten blue links, citation tracking measures whether AI platforms reference your brand in generated responses. This distinction matters because AI discovery follows different rules than search.

While most marketers obsess over search rankings and social shares, a parallel information economy is emerging where AI systems decide which sources get surfaced to millions of users. Most content creators have zero visibility into how they're performing in this new landscape. Citation tracking closes that gap.

How SeaText's ChatGPT Tracking Works

SeaText runs the ChatGPT Brand Visibility Agent alongside your other AI agents. The agent monitors how ChatGPT references your brand, products, and competitors. It tracks which sources get cited, how often, and in what context.

The system works with your existing website setup. You add SeaText to your site, activate the visibility agent, and start receiving reports on AI citation patterns. No manual monitoring required.

The ChatGPT Influence Agent extends this capability by shaping how AI assistants present your brand. It works in real time, adapting your site's signals to influence citation outcomes. The agent tracks results by page, keyword, and version so you can see what changes affect your AI visibility.

SeaText also offers the ChatGPT Brand Choice Agent, which includes auto A/B text testing. This lets you test different brand messaging variants and see which ones AI platforms cite more often. Use this to refine how your brand appears in AI-generated responses.

SeaText creates the proof, product details, and comparison answers AI tools need to understand why buyers should choose you. The AI Agent Actions (WebMCP) feature turns your website into an MCP server for Claude and ChatGPT, enabling direct integration.

Common Setup Mistakes That Break Your Data

Three mistakes appear repeatedly in setups that produce unreliable data:

1. Using overly broad brand keywords. Generic terms like your company name alone pull in too many false positives. Add product-specific modifiers and competitor names to narrow the signal. Broad keywords fill your dashboard with noise that masks real citation patterns. SeaText tracks results by page, keyword, and version, so precise keywords let you attribute citations to specific content.
2. Neglecting source filters. Without defined source filters, the tracker picks up mentions from irrelevant sites. Define which domains and source types matter for your brand category. A filter separates signal from clutter and keeps your reports focused. The ChatGPT Brand Visibility Agent shapes what LLMs recommend about your products, so controlling input sources improves output relevance.
3. Forgetting to activate the ChatGPT Brand Visibility Agent. The agent extracts citation context from raw mention text. If it stays inactive, you get raw mention counts without the structured citation analysis you need. The agent is what turns mentions into actionable citation data. Activation happens in Step 2 of setup: activate the autonomous agents you need.

Each mistake compounds the others. Broad keywords fill your dashboard with noise. Without source filters, you can't separate signal from clutter. Without the visibility agent, you lack the context to judge whether a mention helps or hurts your brand.

Two additional mistakes deserve attention:

4. Skipping language-specific keyword lists. SeaText supports 125 languages. Citation tracking works in each supported language, but you need to configure language-specific keyword lists for accurate results. A single global keyword list misses regional citation patterns.
5. Not enabling the ChatGPT Brand Choice Agent for A/B testing. Auto A/B text testing lets you test different brand messaging variants and see which ones AI platforms cite more often. Without it, you cannot optimize how your brand appears in AI-generated responses.

Decision Framework: Choosing What to Track

Before you configure tracking, decide what matters most for your business goals:

• Total citations per week — shows overall AI visibility volume
• Unique domains citing you — reveals which sites AI trusts for your topic
• Sentiment score of mentions — indicates brand perception in AI outputs
• Citation velocity (change over time) — shows whether your AI presence is growing or declining
• Geographic distribution of mentions — identifies market-specific visibility gaps
• Competitor citation share — benchmarks your AI visibility against rivals
• Citation-to-conversion rate — links AI mentions to downstream actions

These seven metrics give a complete picture. Track all seven from day one. Adding more later is easier than cleaning up incomplete baseline data.

Use the ChatGPT Brand Choice Agent to test messaging variants. The agent includes auto A/B text testing. Deploy variants, measure citation lift, and scale winners. This turns tracking into an optimization loop.

Align metrics with business stage. Early-stage brands should prioritize unique domains and citation velocity. Mature brands should weight sentiment and competitor share. Ecommerce brands should track citation-to-conversion via WebMCP integration.

Practical Scenarios: When Tracking Works and When It Doesn't

Scenario 1: SaaS Brand Launching New Feature

Add feature-specific keywords to your brand list. Activate the ChatGPT Brand Visibility Agent. Set source filters to include tech review sites and developer forums. Enable the ChatGPT Brand Choice Agent to test two positioning statements. Measure citation velocity weekly. Expect initial data within 48 hours. Full baseline reports take 2-4 weeks.

Scenario 2: Ecommerce Brand Expanding to Europe

Configure language-specific keyword lists for German, French, Spanish. Activate 125-language translation agent alongside citation tracking. Set geographic distribution as a primary metric. Use WebMCP to forward purchase signals to Meta and Google CAPI. Track citation-to-conversion by region.

Scenario 3: Established Brand Losing AI Visibility

Audit current keyword list for breadth. Narrow to high-intent product terms. Add competitor names to keyword list for comparison data. Check that ChatGPT Brand Visibility Agent is active. Review source filters for new irrelevant domains. Run A/B tests on brand messaging via Brand Choice Agent. Monitor sentiment score for shifts.

Scenario 4: Agency Managing Multiple Clients

Use SeaText's multi-agent dashboard. Activate ChatGPT Brand Visibility Agent per client. Configure unique keyword lists and source filters per brand. Share read-only reports with clients. Leverage auto A/B testing across client portfolio to identify cross-brand messaging patterns.

Limitations and Edge Cases

SeaText's ChatGPT tracking works best for brands with an established web presence. If your site has low crawl frequency, citation data may lag. The tool focuses on ChatGPT specifically. For Google AI Overviews or Perplexity tracking, check whether your plan includes those platforms.

Citation tracking measures mentions, not citations. A mention is when your brand appears. A citation is when ChatGPT references your content as a source. The distinction matters for interpreting your reports.

The source pack does not specify pricing for the ChatGPT visibility features. Contact SeaText sales for plan details. The 87% client report acceptance rate cited in SeaText materials applies to bot refund reports, not citation tracking reports.

SeaText's ChatGPT tracking depends on ChatGPT's indexing frequency, which SeaText cannot control. If ChatGPT doesn't index your site frequently, citation data will be sparse regardless of setup quality.

Auto A/B text testing requires sufficient citation volume to reach statistical significance. Low-citation brands may need longer test durations or should prioritize visibility growth before testing.

WebMCP integration requires technical setup. Not all hosting environments support MCP server configuration. Check with your development team before enabling.

Terminology Reference

Citation: When an AI platform references your brand or content in a generated response. Citations carry authority signal because they indicate the AI treats your source as relevant and trustworthy.

ChatGPT Brand Visibility Agent: The core tracking agent that monitors how ChatGPT references your brand, products, and competitors. Tracks which sources get cited, how often, and in what context.

ChatGPT Influence Agent: An agent that adapts your site's signals in real time to influence citation outcomes. Tracks results by page, keyword, and version.

ChatGPT Brand Choice Agent: An agent that includes auto A/B text testing. Lets you test different brand messaging variants and see which ones AI platforms cite more often.

Source Filter: A defined list of domains or source types the tracker should monitor. Without a filter, the system captures all mentions, including irrelevant ones that dilute your data.

WebMCP: A protocol that turns your website into an MCP server for Claude and ChatGPT, enabling direct AI platform integration and signal forwarding.

Citation Velocity: Rate of change in citation frequency over time. Shows whether your AI presence is growing or declining.

FAQ

How long until I see citation data after setup?

Initial data appears within 48 hours. Full baseline reports take 2-4 weeks to build. Citation patterns shift slowly, so check weekly, not daily.

Does this track competitor citations too?

Yes. The visibility agent tracks competitor mentions alongside your own. Add competitor brand names to your keyword list for comparison data.

What if ChatGPT doesn't cite me at all?

Zero citations is a valid data point. It means AI platforms don't currently see your content as authoritative on the tracked topics. Use that signal to adjust your content strategy.

Can I track citations across multiple languages?

SeaText supports 125 languages. Citation tracking works in each supported language, but you need to configure language-specific keyword lists for accurate results.

How is this different from Google Search Console?

Google Search Console tracks search clicks and impressions. Citation tracking measures AI platform references. They measure different discovery channels. Use both for complete visibility.

What setup effort is required?

Add SeaText to your website in under 1 minute. Activate the ChatGPT Brand Visibility agent. Configure your brand keywords and source filters. Most users complete setup in under 30 minutes.

Can I export citation data for external analysis?

SeaText tracks results by page, keyword, and version. Export capabilities depend on your plan. Check with the vendor for API access and data export formats.

Does the ChatGPT Influence Agent guarantee more citations?

No. The agent adapts your site's signals to influence outcomes, but ChatGPT's citation decisions remain opaque. Track citation velocity to measure impact.

What happens if I change my brand keywords mid-campaign?

Changing keywords resets baseline tracking for those terms. Historical data remains for previous keywords. Plan keyword changes during low-traffic periods to minimize disruption.

Yes, Seatext can help a Notion site appear in ChatGPT recommendations. Seatext runs a dedicated ChatGPT Brand Choice Agent that builds the proof, product details, and comparison answers AI tools need to understand why buyers should choose you. Once your Notion site is published on a real domain, Seatext treats it like any other web property and adds the structured, crawlable content that helps large language models recommend it.

This guide explains exactly what that looks like, the steps a Notion site owner should follow, where the limits are, and how to judge whether Seatext is the right tool for your specific situation.

What "appearing in ChatGPT recommendations" actually means

When someone asks ChatGPT "what is the best tool for X" or "which brand should I choose for Y," the model does not browse the web live for every question. It draws on a mix of training data, retrieval-augmented sources, and plugins or browsing tools. For a brand to be named in those answers, three things usually need to be true:

• The brand has a public website that crawlers can read.
• The site contains clear, specific answers to the kinds of questions buyers ask.
• Third-party sources mention the brand in similar contexts.

Notion sites can clear the first bar easily once they are published under a custom domain. The second and third bars are where most Notion sites struggle, because Notion pages are often short, image-led, and not written in the Q&A format that language models parse well.

What Seatext's ChatGPT Brand Choice Agent does

Seatext's ChatGPT Brand Choice Agent is one of more than 20 agents on the platform. Its job is to shape what LLMs understand and recommend about your products. According to Seatext's product pages, the agent "creates the proof, product details, and comparison answers AI tools need to understand why buyers should choose you." It also ships an "AI Agent Actions (WebMCP)" capability that can turn your website into an MCP server for Claude and ChatGPT, which gives those assistants a structured way to read your offering.

For a Notion site, the practical effect is that Seatext:

• Adds crawlable, indexable Q&A pages on top of your existing Notion content.
• Surfaces the proof points, product specs, and side-by-side comparisons that language models look for when picking a brand to recommend.
• Makes the comparison context explicit, so your brand appears in answers to questions like "which X is best for Y" instead of only on branded searches.

Why Notion sites often miss ChatGPT recommendations

Notion is a fast way to publish a website, and Super.so, Fruition, and similar tools make it easy to put Notion pages on a custom domain. The trade-off is that the underlying structure is optimized for reading, not for being parsed by AI.

Common gaps Seatext addresses:

• Few Q&A pages. Buyers ask ChatGPT questions. Notion pages are usually essays or feature lists.
• Thin comparison content. ChatGPT recommends by comparing. Notion sites rarely publish "you vs. competitor" tables.
• Limited structured proof. Testimonials, integrations, and pricing live in scattered places instead of in formats an LLM can quote.
• Weak topical depth. Seatext's own research notes that "most websites cover only 1-5% of search demand in their industry," which is the same gap that hides brands from ChatGPT.

Step-by-step: putting a Notion site in front of ChatGPT with Seatext

1. Publish Notion on a custom domain. Use Super.so, Fruition, or Notion's built-in site publishing. Search engines and LLM crawlers need a stable, crawlable URL.
2. Install the Seatext snippet. Seatext says you can "add Seatext to your site in under 1 minute." The script works on any HTML-rendered page, including Notion sites served through Super.so.
3. Activate the ChatGPT Brand Choice Agent. In the Seatext dashboard, turn on the agent that "shapes what LLMs recommend about your products."
4. Activate the AI SEO Content Factory. This agent publishes indexed Q&A pages on your domain. These pages answer the long-tail questions your buyers ask ChatGPT.
5. Activate WebMCP / AI Agent Actions. If you want ChatGPT and Claude to read your site as a live MCP source instead of only through retrieval, switch on the AI Agent Actions capability.
6. Review the published pages. Seatext adds new pages on top of your existing Notion content rather than replacing it. Check that the questions match what your buyers actually ask.
7. Monitor mentions. Ask ChatGPT the questions your buyers ask and watch whether your brand starts to appear in answers over the following weeks.

Key facts about Seatext for Notion + ChatGPT

Common mistakes when trying to get a Notion site into ChatGPT

• Skipping the custom domain. Notion's default notion.site URLs are sometimes throttled for crawlers. A custom domain is required for reliable indexing.
• Writing only about your brand. ChatGPT recommends by comparison. Without "you vs. alternative A vs. alternative B" pages, the model has nothing to compare against.
• Hiding proof in PDFs. Reviews, certifications, and case studies need to be on the live HTML page so they can be quoted.
• Treating ChatGPT recommendations like Google rankings. They overlap but are not the same. ChatGPT pulls from many sources, and a single high-ranking page is not enough.
• Ignoring freshness. Models weight recent mentions. A burst of new Q&A pages today helps more than a single evergreen page from a year ago.

Limitations and when Seatext is not the right fit

Seatext shapes what language models can find and quote about your brand. It cannot guarantee a specific ChatGPT answer, because OpenAI controls ranking, retrieval, and safety behavior inside ChatGPT. If your Notion site is still on a default notion.site URL with no custom domain, fix that first; Seatext works on the rendered HTML, not on Notion's internal editor.

Seatext is also a paid platform (the site shows pricing starting around $59/month, with an enterprise tier for managed rollout). If you only want one or two Q&A pages, a plain blog post may be enough. Seatext is most useful when you want a steady stream of new crawlable pages, comparison content, and a structured way to feed ChatGPT and Claude.

Practical scenarios

Solo founder with a Notion landing page. Turn on the ChatGPT Brand Choice Agent and the AI SEO Content Factory. Let Seatext publish 30 to 50 Q&A pages over the first two months.

Small SaaS team with a Notion docs site. Add Seatext on top of the docs domain. Use WebMCP so ChatGPT can read your docs as a live source. Compare this with hand-writing docs Q&A pages, which is slow and rarely keeps up with product changes.

Agency managing several Notion client sites. Activate the ChatGPT agent and the Website Translation Agent together. You get LLM-friendly content in 125 languages from the same source pages.

FAQ

Does Seatext work on a Notion site specifically?

Seatext runs a JavaScript snippet on your site. It works on any HTML-rendered page, including Notion sites served through Super.so, Fruition, or Notion's native site publishing. It does not work inside the Notion app itself.

How long before my Notion site shows up in ChatGPT answers?

Seatext can publish new pages within days, but ChatGPT's update cycle is slower and not fully transparent. Plan on 4 to 12 weeks of consistent publishing before you judge the result, and ask ChatGPT the same questions your buyers ask to track change over time.

Do I still need SEO if I want ChatGPT recommendations?

Yes. ChatGPT pulls from indexed web pages and from third-party mentions. Strong SEO is a leading indicator of ChatGPT visibility, and Seatext's own AI SEO Content Factory is built to publish the indexed Q&A pages both Google and ChatGPT use.

What does WebMCP actually do for a Notion site?

WebMCP turns your site into an MCP server that Claude and ChatGPT can read directly. For a Notion site, this means the assistants can pull your product specs, pricing, and proof points on demand instead of relying on cached retrieval.

Can Seatext get my brand into ChatGPT for queries I have never ranked for?

Seatext can publish comparison and Q&A pages that target those queries, which gives the model new sources to draw from. Whether ChatGPT chooses to recommend your brand depends on the model's own logic and on the strength of your proof compared with named alternatives.

Is Seatext only for ecommerce?

No. The ChatGPT Brand Choice Agent works for any brand with a public website. Notion sites for SaaS, agencies, personal brands, and consultants fit the same model.

What happens to my existing Notion pages when I add Seatext?

Seatext adds new pages and content on top of your domain. It does not rewrite or replace your existing Notion pages. You keep editorial control of the core site while Seatext expands the surface area that crawlers and LLMs can read.

Direct answer

SeaText charges per AI agent you enable and the scope of its deployment. Each agent performs a single growth task, and pricing varies with the number of agents, the languages you translate into, and the volume of content.

Cost drivers

• Agent selection – Every AI agent (e.g., Google Ads Landing Page Agent, Translation Agent, Bot Refund Agent) has its own price. Adding more agents increases the total cost.
• Enterprise controls – Advanced permissions and multi‑region deployment add a premium.
• Translation specifics – For translation agents, cost depends on text length, language pair, content complexity, urgency, formatting, and revision needs.

Variables that affect price

1. Number of agents you activate.
2. Number of languages and word count for translation.
3. Complexity of the content (technical, marketing, etc.).
4. Required enterprise features such as multi‑site management and compliance reporting.

Scoping the work

1. Identify the growth metrics you need to improve (conversion, traffic, bot protection, international reach).
2. Choose the corresponding AI agents that address those metrics.
3. Estimate the volume of content or traffic each agent will handle.
4. Add any enterprise controls you require.
5. Review the pricing page to see the total cost for the selected configuration.

What is AI powered SEO?

AI powered SEO is the use of artificial‑intelligence agents to automatically discover search intent, create or rewrite website copy, and publish optimized pages without manual writing or spreadsheets. The AI reads campaign keywords, visitor intent, and real‑world questions, then produces content that matches what users and search engines are looking for.

How the process works

1. Intent discovery: The AI scans paid and organic traffic to identify high‑value keywords and long‑tail questions.
2. Content generation: Using large language models, it writes answers, headlines, product descriptions, and localized pages that directly address the discovered intent.
3. Automatic publishing: New pages are uploaded to your site and indexed by Google, creating a growing library of searchable Q&A.
4. Continuous optimization: The AI monitors performance, rewrites copy, and runs A/B tests to keep the pages aligned with evolving search trends.

Common mistake to avoid

Relying solely on AI‑generated copy without reviewing brand tone can lead to off‑message pages. Always set enterprise controls that let you approve variants before they go live.

Next step

Deploy Seatext’s AI SEO agents to start generating intent‑driven pages in minutes.

What is AI‑powered personalization?

It is an AI agent that analyzes a visitor’s source data (keyword, ad campaign, account information, or referral) and dynamically rewrites on‑page copy—headlines, product blocks, CTAs—to mirror that intent.

The process

1. Capture intent. The agent reads the campaign, keyword, or CRM data that brought the visitor to the page.
2. Generate tailored copy. Using large‑language‑model prompts, it creates headlines, offers, and calls‑to‑action that align with the detected intent.
3. Rewrite in real time. The page is updated instantly, so the visitor sees a version that feels built for their search.
4. Test and verify. Variants are A/B‑tested automatically; winning versions are kept and rolled out.

Common mistake

Deploying the agent without a clear data source (e.g., missing keyword or CRM fields) leads to generic rewrites that can confuse users rather than personalize them.

How to verify success

Monitor conversion‑rate changes per source and compare against a control page that isn’t rewritten; a lift indicates the personalization is effective.

How AI-Powered Multilingual Growth Works

What AI-Powered Multilingual Growth Means

AI-powered multilingual growth is the practice of using machine learning models to translate, adapt, and optimize every page of a website for visitors who speak different languages. Instead of hiring translators, managing translation memories, or maintaining separate language sites, a single AI agent detects each visitor's language, translates the page in milliseconds, and serves a version that reads like it was written for that market.

The scope covers three connected layers: translation (converting text), localization (adapting cultural references, currencies, units, and legal disclaimers), and conversion optimization (testing which phrasing drives the most sign-ups, purchases, or leads in each language). When these layers run continuously, the site improves in every market without additional headcount.

How the Process Works End to End

1. Language detection. The agent reads the visitor's browser headers, IP geography, and on-site behavior to decide which language to serve before the page renders.
2. Instant translation. The page content — headlines, product descriptions, buttons, navigation, schema markup — is translated in roughly 3 ms. New pages, posts, product updates, and headline changes are picked up automatically and translated in the background.
3. Brand-context preservation. A glossary and style guide keep product names, tone, legal terms, and formatting consistent across all 125 supported languages.
4. Localized rendering. The translated HTML is served to the visitor with correct hreflang tags, localized URLs, and translated meta data so search engines index each language version.
5. Performance tracking. Conversions, bounce rates, and revenue are attributed per language and market, giving a clear view of which regions grow and which need attention.
6. Continuous optimization. Optional A/B testing creates variant translations for the same language, measures lift, and promotes the winner automatically.

Key Components of an AI Multilingual System

• Translation engine. Neural models trained on web content, not just generic corpora, so marketing copy, CTAs, and product specs sound natural.
• Glossary and brand memory. A centralized repository of approved terms, banned words, and formatting rules that the model consults on every request.
• SEO layer. Automatic hreflang injection, translated meta titles and descriptions, localized sitemaps, and structured data so each language version ranks independently.
• Analytics dashboard. Language-level reporting on traffic, conversions, revenue, and test results. Teams can see which markets deliver ROI and where to invest next.
• Testing framework. Built-in A/B testing that splits traffic per language, measures statistical significance, and rolls out winning variants without developer involvement.

Why It Matters: Business Impact

Most companies treat translation as a one-time project. They launch a few languages, then stop because the workflow is slow and expensive. AI-powered multilingual growth changes the economics: the marginal cost of adding a language drops to near zero, and the time to launch goes from months to minutes.

This shifts the strategic question from "which languages can we afford?" to "which markets show demand?" Teams can test 20 languages in a quarter, double down on the three that convert, and pause the rest. The same engine that translates also optimizes, so the Spanish version that converts 15% better than the initial draft is the one that stays live.

Main Options and Trade-offs

Choose human agencies if you have highly regulated content (medical, legal) where liability requires a certified translator's signature on every page.

Choose traditional SaaS if you need a visual editor for non-technical marketers to tweak translations daily and you accept per-word pricing.

Choose an AI agent with optimization if you want to launch many languages fast, measure real conversion impact, and iterate without adding headcount.

Step-by-Step Implementation Framework

1. Audit current international traffic. Check analytics for top non-English countries, languages in browser settings, and search console impressions by country.
2. Define the glossary. Export 200–500 key terms: product names, feature terms, legal phrases, brand voice adjectives. Upload to the AI agent before launch.
3. Install the agent. Add a single JavaScript snippet or CMS plugin. The agent begins translating immediately for detected languages.
4. Verify hreflang and indexing. Use Search Console's International Targeting report to confirm each language version is crawled and indexed.
5. Set up conversion goals per language. Map existing GA4 or Matomo events (purchase, lead, sign-up) to language dimensions.
6. Run a 30-day baseline. Collect traffic and conversion data without testing. Identify the top 3–5 languages by revenue potential.
7. Enable A/B testing on priority languages. The agent generates variant headlines, CTAs, and product copy. Review winners weekly; approve rollouts.
8. Expand or pause. Add languages showing organic demand. Pause languages with zero traffic after 60 days to keep the dashboard clean.

Common Mistakes and How to Avoid Them

Limitations and When This Advice Does Not Apply

• Highly regulated copy. Medical device claims, financial disclosures, and government-mandated labeling often require certified human translation with legal sign-off. AI can draft, but a qualified reviewer must approve.
• Creative brand campaigns. Taglines, humor, and cultural wordplay rarely survive machine translation intact. Keep flagship campaign pages in human hands.
• Right-to-left languages with complex layouts. Arabic, Hebrew, and Farsi may need CSS adjustments (flex direction, icon mirroring) that the translation layer doesn't handle.
• Sites with heavy client-side rendering. If content loads via React/Vue after the initial HTML, the agent must be configured to translate dynamic fragments. Test thoroughly in staging.
• Zero existing international traffic. If analytics show no demand from non-English countries, the ROI case is weak. Run a paid test campaign first to validate interest.

Key Facts

Terminology Quick Reference

hreflang

HTML attribute telling search engines which language and regional version of a page to serve to users in a given locale.

Glossary (translation memory)

Structured list of approved term translations, do-not-translate rules, and formatting exceptions that the AI consults on every request.

Locale

Language plus region code (e.g., es-MX, fr-CA) that determines currency, date format, number format, and idiomatic preferences.

Variant testing

Controlled experiment where two or more translations of the same element are shown to split traffic; the version with statistically significant higher conversion wins.

Bot refund agent

Separate AI agent that detects fraudulent paid clicks, documents sessions, and prepares evidence for ad-platform refund claims (Google, Meta, TikTok, Reddit).

FAQ

How long does it take to launch a new language?

Minutes. The 125 languages are pre-enabled. After the glossary is uploaded, the agent serves translated pages on the first visit from a user with that language preference.

What happens when we publish a new product page in English?

The agent detects the new URL, translates it automatically, and serves the localized version to visitors in their detected language. No manual trigger required.

Can we review translations before they go live?

Yes. Enterprise plans include review gates: winning A/B variants and new language rollouts can be set to require approval before publishing.

Does the free tier include A/B testing?

No. The free tier covers unlimited translation, languages, pages, and traffic. A/B testing to find the highest-converting translation per language is a premium feature.

How do we know which languages are worth keeping?

The dashboard shows traffic, conversions, and revenue per language. Pause languages with zero sessions after 60 days; double down on languages with positive ROAS.

Will AI translation hurt our SEO?

No, if hreflang tags, translated meta data, and localized structured data are implemented correctly. The agent handles all three automatically.

What if we need certified translation for legal pages?

Use the AI draft as a starting point, then have a certified translator review and sign off. The glossary ensures the certified version stays consistent with the rest of the site.