AI visibility: do it yourself, hire an agency, or use a platform?

What are the four jobs behind AI visibility?

AI visibility work splits into four jobs, and the options differ mainly in who does each one:

1. Measuring what ChatGPT, Gemini, Perplexity, Copilot and Google’s AI features say when buyers ask about your category, repeated often enough to trust the result.
2. Technical changes to your site: crawler access, structured data, metadata, page templates, and content hidden in sliders, tabs, popups or images that AI reads poorly. This is development work in your site’s code or CMS.
3. Content: answer-first rewrites, corrections where AI repeats something wrong about you, and the missing pages buyers ask about.
4. Authority: mentions on the third-party sites AI answers already cite, such as publications, review sites and forums.

None of the four is a one-off. Google says its generative AI features in Search are “rooted in our core Search ranking and quality systems”, so a lot of the technical job is good SEO practice. But AI answers change between runs, competitors publish, and every new measurement finds new gaps, so all four jobs come back every month.

DIY vs agency vs platform: quick comparison

“Platform” covers two different products. A tracking tool measures and hands you a report. A platform that ships also makes the changes. The table separates them because the difference decides who does the work.

The core difference: doing it yourself costs your hours, an agency costs a retainer for someone else’s hours, a tracking tool tells you what to do, and a platform that ships does it while you keep the final say.

How many hours does DIY AI visibility take?

About 20–25 hours in the first month, then about four hours a week. These are our estimates from adding up the free starter routine further down this page, not survey data. Here is where the time goes:

Three things this table still leaves out:

• The measurement stays rough. Three answers per question is far below the “at least 60-100X” that SparkToro advises. At that level, 20 questions on 6 platforms is over 7,000 answers a month, which nobody logs by hand. A platform does it on a schedule.
• It assumes you can edit your site’s code. Schema, metadata, crawler rules and revealing content hidden in sliders or tabs are development changes. If you cannot make them yourself, add a developer’s hours.
• It covers one new page a month. A platform that ships can produce several: Far & Wide’s Fix plan includes up to 10 articles, 5 design fixes and 20 authority opportunities a month, plus the technical changes each measurement finds. Writing 10 articles alone would take most founders 30 hours or more.

Should you rewrite your content yourself or have a service handle the AEO fixes?

Rewrite your most important pages yourself if you know the product better than any outsider and write clearly. Google tells site owners to “create the content yourself based on what you know about the topic” and says “you don’t need to write in a specific way just for generative AI search.” Hand the work to a service when rewrites keep slipping, when you have dozens of high-intent pages, when the fixes are code changes you cannot make, or when the real gap is off-site authority. A good service starts from your knowledge: it drafts, and you correct the facts before anything goes live.

Ranking on Google but missing from the AI answer

A page can rank on page one and still miss the AI answer, because that answer is built from several sources and can lean on third-party pages such as comparison articles, review sites and forums. Google’s entry bar is plain: a page “must be indexed and eligible to be shown in Google Search with a snippet”, and “there are no additional requirements to appear in AI Overviews or AI Mode.” Meeting that bar makes you eligible, not chosen. Assistants such as ChatGPT, Perplexity and Copilot can draw on their own search, so a competitor named in one popular comparison article can become the answer instead of you.

To check, ask a buyer question in the ChatGPT, Gemini, Perplexity and Copilot apps, note the brands and sources named, and compare them with your Google Search Console queries. If you rank but don’t appear, the fix is usually content shape (a direct answer near the top), content AI cannot see (hidden in a slider, tab or image), or authority (being on the pages the AI cites).

Entity and schema markup you can do yourself

Entity markup tells search engines which company you are. If you can edit your homepage’s code or your CMS’s header settings, you can add it yourself:

1. Add Organization JSON-LD to your homepage with name, url, logo, founder and foundingDate.
2. Fill sameAs with your LinkedIn company page, review-site profiles and a Wikidata entry if one exists. Schema.org defines sameAs as the “URL of a reference Web page that unambiguously indicates the item’s identity.”
3. Use Product or Service markup on product pages, and FAQPage only where the questions are visible on the page. Google asks that “your structured data matches the visible text on the page.”
4. Validate in Google’s Rich Results Test, which Google calls “an easy and useful tool”, and in the Schema Markup Validator, which checks any schema.org type.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Example Co",
  "url": "https://www.example.com",
  "logo": "https://www.example.com/logo.png",
  "foundingDate": "2023",
  "founder": { "@type": "Person", "name": "Jane Doe" },
  "sameAs": ["https://www.linkedin.com/company/example-co"]
}

Google states that “structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add.” Schema is identity hygiene, not a switch that gets you cited, and it is one change among the many a measurement turns up. See our guides to schema markup for AEO and brand entity optimization.

Should you hire an agency or use a platform to improve your AI visibility?

Use a platform that ships if you want the four jobs done without hiring a developer, a writer or an agency: it covers the code changes, the content and the outreach drafts, and you approve each one. Hire an agency if you need people in your meetings, brand strategy, or PR campaigns run across several markets, and the budget for a retainer. A tracking tool only makes sense if your own team already has the developers and writers to act on its reports. For costs, see how much AEO costs.

What an AI visibility agency actually does

A typical engagement runs in six steps: benchmark what the AI engines say about you and competitors; fix crawl access, indexing and schema; rewrite the highest-intent pages answer-first; write missing pages such as comparisons and alternatives; earn mentions on review sites, directories and media; then re-measure and report. An SEO agency CEO quoted by Digiday put it bluntly: “80 percent of GEO is good, fundamental SEO.” Ask any agency which parts of its plan are specific to AI answers (AEO vs SEO).

What a platform that ships does

It runs the same six steps with software, and you approve each change instead of doing it. Taking Far & Wide as the example:

• Measures the buyer questions on the 4 AI platforms you choose, in the real consumer apps, asking each question several times per check as different buyer types (how AI visibility tracking works, real apps vs API).
• Makes the technical changes: crawler rules, schema, metadata and llms.txt, written as the exact code change and shipped as a GitHub pull request you merge or as a draft in WordPress, Webflow or HubSpot. On Shopify and other sites it arrives as ready-to-paste code.
• Fixes page design that hides content from AI, such as FAQs in accordions, reviews in carousels or text in images, without changing how the page looks.
• Writes content: corrections where AI repeats something wrong about you, and the missing pages buyers ask about.
• Prepares authority work: the publications, forums and threads AI already cites, with outreach drafted.
• Proves it: every change is checked live on your site, can be undone, and the affected questions are re-measured.

Questions to ask an agency or platform before you sign

• Which buyer questions will you track, and can I add my own?
• On which AI platforms, and in the real apps or through an API?
• How many times is each question asked per check?
• Do you make the changes on my site, or do I get a list of recommendations?
• How do you confirm a change is live, and can it be undone?
• Who owns the content, accounts and data if we stop?
• Does the report show the share of answers that mention us, or a “rank”? SparkToro concluded that “any tool that gives a ‘ranking position in AI’ is full of baloney.”

Which option fits your team?

The hours are our estimates, not survey data.

What does a good AI visibility audit check?

A useful audit shows what AI says about you, why, and which fixes come first. You can run a basic version yourself:

1. Questions. Collect 10–20 real buyer questions from sales calls, support tickets and reviews (how to find them).
2. Answers. Ask each in the ChatGPT, Gemini, Perplexity and Copilot apps, plus Google AI Overviews and AI Mode, three times each.
3. Log. Record whether you are mentioned or linked, whether what it says about you is accurate, which competitors appear and which sources are cited.
4. Crawler access. Check yourdomain.com/robots.txt allows OAI-SearchBot, Googlebot and Bingbot. OpenAI says: “we recommend allowing OAI-SearchBot in your site’s robots.txt file.”
5. Indexing. Confirm your important pages are indexed in Google Search Console and Bing Webmaster Tools.
6. Copilot citations. Open the AI Performance report in Bing Webmaster Tools, which shows how content “appears across Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations”, with the grounding queries behind each citation.
7. Schema. Validate your Organization and product markup.
8. Output. List fixes by job (technical, design, content, authority), ranked by how many buyer questions each affects.

An audit only tells you what to do. Each item on that list is then a change someone has to make, check and re-measure, which is where most of the hours go. The AEO audit guide covers the full method.

I’m a solo founder with no budget. Can I improve my AI visibility myself?

Yes, you can start, as long as you plan for about 20–25 hours in the first month and about four hours a week after that. Your first-hand product knowledge is exactly the “in-depth experience” Google asks content to bring. The limits are real: hand-checked measurement stays rough, code changes need someone who can edit the site, and earning third-party mentions is slow. Many founders use the routine below to learn where they stand, then hand the ongoing work to a platform.

Free tools for each AI visibility job

The free starter routine for the first 30 days

1. List 10–20 buyer questions from sales calls, support tickets and reviews (1 hour).
2. Ask each three times in the ChatGPT, Gemini, Perplexity and Copilot apps, plus Google AI Overviews and AI Mode, and log every answer (about 6 hours for 20 questions).
3. Check crawler access and indexing in robots.txt, Search Console and Bing Webmaster Tools (30 minutes).
4. Add Organization schema with sameAs and validate it (1–2 hours if you can edit the site’s code).
5. Rewrite your 3–5 highest-intent pages answer-first, with a direct answer in the first two sentences. Bing recommends “clear headings, tables, and FAQ sections” and “examples, data, and cited sources” (about 5–8 hours for the set).
6. Publish one missing page buyers ask about, such as a comparison or alternatives page (3–4 hours).
7. Get onto one or two third-party pages the AI already cites, such as review sites or community threads (about 1 hour a week).
8. Re-run the same questions monthly and track the share of answers that mention you, not position (about 6 hours a month).

Why can a weekly manual check mislead you?

One manual check a week mostly measures chance. In SparkToro’s January 2026 study, 600 volunteers ran 12 prompts through three AI tools, including ChatGPT and Google’s AI, a combined 2,961 times. The authors found “a <1 in 100 chance that ChatGPT or Google’s AI, if asked 100X, will give you the same list of brands in any two responses” and advised asking “at least 60-100X” before averaging. The authors say they are not professional researchers; Digiday makes the same point: “LLMs are probabilistic, not deterministic.”

For DIY reporting, keep the question set fixed, run each question several times, and report the monthly share of answers that mention you. Know what that number is: a rough trend from a few answers per question. Reliable measurement means thousands of answers a month, which is software’s job, not a spreadsheet’s.

Which option should you choose?

• Start yourself if you have no budget yet and can give it 20–25 hours in the first month. You will learn where you stand, and how much of the work keeps coming back.
• Choose a platform that ships if you want the measuring, the code changes, the content and the outreach drafts done for you, with your approval on each change and about an hour or two of your time a week.
• Choose an agency if you need people in your meetings, brand strategy or PR campaigns across several markets, and have the budget for a retainer.
• Choose a tracking tool only if your own developers and writers will act on every report.

The simple rule: doing it yourself is free but costs about half a day a week once you count everything, and the measurement stays rough. A platform that ships gives you those hours back for a few hundred euros a month. An agency is the choice when you need people more than output.