SEO Autopilot

SEO Autopilot looks at your whole store and tells you what is costing you search traffic. Instead of opening products one by one wondering which need attention, you get a ranked list of what to fix and a route straight to each fix.

You'll find it under Marketing, SEO Autopilot in the sidebar.

You need a storefront connected first. If your store has no sales channel yet, the page shows "Connect a storefront first" with an Add a sales channel button instead of any scores, because the articles and checks are written for one storefront domain. Products, Categories and the site score cover your whole catalog, not one channel. Content, Measurement and Authority each have their own channel picker for the places where a single storefront matters.

Who can do what. Anyone with the View SEO permission can open the page and read everything. Changing the mode, generating, approving or fixing anything needs Manage SEO. By default the Manager role has both, and Staff and Viewer can only look. The mode switch, Generate article now and the main Fix button at the top are hidden for a team member who can only view.

What to fix next: the ranked list

The top of the page is a numbered list of work, highest impact first. Each row says why it sits where it does, in plain numbers:

• How much the fix is worth. Every check carries a weight. A product with no price loses far more than one with a non-Latin URL slug, so the missing price is listed first.
• How many products it affects. A problem on 40 products outranks the same problem on 2.

Rows marked Blocking are different. A blocking problem holds your whole score below Excellent no matter how much else you fix, so it is a ceiling rather than a deduction. Clear those first.

The list shows the first six rows with Show all N for the rest. A row you can fix with AI has a Fix: button, and a row that only you can resolve (a missing barcode, say) is marked Needs your input. The Products, Categories and decaying-articles lists likewise show their five worst rows first, with Show all N and Show less. The list is not ordered by guesswork or by any claimed traffic multiplier. It is ordered by the same scoring the audit itself uses, which is why each row can show you its reasoning.

The seven tabs: Products, Categories, Content, Technical, Authority, Measurement and AI visibility

Under the score and the ranked list, the page is split into seven tabs. The score and the "What to fix next" list stay on screen whichever tab you open, because the list mixes work from every tab.

Content tab: the article board and the three autopilot modes

The Content tab is where SEO Autopilot writes blog articles aimed at searches your customers actually make. Four chips across the top show the stages an article moves through: Planned (keywords picked from your catalog), Writing now, Awaiting review and Published. Click a chip to filter the board to that stage.

The switch beside the title sets how much the engine does without you:

• Off. Nothing is written on a schedule and the Next article time reads Off. You can still press Generate article now and it will write one on demand.
• Review first. The engine writes on your schedule, but every article stops at Awaiting review. Nothing reaches your blog until you press Approve & publish. If a draft is wrong, Reject draft lets you say what was wrong and optionally have it write an improved version for the same keyword.
• Full autopilot. Articles are published to your live blog automatically, with no review, and search engines are told straight away. Switching to it asks you to confirm and shows the credits each article costs. Choose it only if you are happy with articles going live unread.

Switching from Off to Review first or Full autopilot starts the first run about five minutes later, and that run spends credits. After that, runs follow the cadence you set in Settings on this page: daily, twice a week, weekly, every two weeks or monthly. Cadence is open to every plan. Credits are the only limit.

Other things on the board:

• Add keyword lets you type a search phrase (and its intent: informational, commercial, transactional or comparison) so the engine writes about it.
• Plan more keywords asks the AI to look at your catalog and bestsellers and propose about ten new keywords. It costs about 2 credits.
• Generate article now opens a dialog where you pick a planned keyword, or type a new topic with optional instructions for the writer.
• Decaying content appears only when a published article is losing views or is no longer cited by AI answers, or has not been updated for weeks. Refresh drafts a rewritten version. It never replaces the live article by itself, in any mode: you preview it and press Approve & update, and the URL, publish date and cover image stay the same.
• Orphan pages lists products that no category and no article links to. See below.

The Settings button at the top of the page holds the schedule and the writing preferences: cadence, article language (one language per store, chosen from your store's languages), whether to generate a cover image, whether to ground articles in web-search sources, topic steering in your own words, an author byline, and the brand voice the writer imitates. The brand voice is learned from your store on the first run, and Regenerate voice relearns it.

Technical tab: what the checks look at

The Technical tab lists the foundations, each with a green or grey status:

• Homepage meta description. Shows whether it is set, and opens the editor.
• Storefront technical check. Press Run check and Brainerce looks at your live storefront: sitemap.xml, robots.txt, whether AI crawlers (ChatGPT, Perplexity, Claude) are named in robots.txt, llms.txt, agents.md, the IndexNow key file, structured data (JSON-LD) on the homepage, Product structured data on a product page, a canonical URL on a product page, and your Google site-verification tag. A failed row says why, for example "serves an HTML page instead of the file, so update your storefront". This check reads your site from the outside. It does not change your storefront, and it does not spend AI credits.
• IndexNow. Instant indexing for search engines that support it. It switches on by itself the first time you publish.
• Google Search Console. Shows whether it is connected and takes you to the Measurement tab to connect it.
• Page speed (Core Web Vitals). Explained in its own section below.

A grey row on this tab means the storefront or the connection is missing something. It is never a promise that Google will rank you lower.

Homepage meta description: two places to edit it

You can edit your homepage meta description in two places, and both open the same editor and save to the same field:

• SEO Autopilot, Technical tab, on the Homepage meta description row.
• Settings, General, on the Homepage SEO card (see General store settings). The Settings card is the same editor, so use whichever page you are already on.

The editor has a Create with AI button if you would rather start from a draft than a blank field. The draft is only a suggestion: it appears in the box, and nothing is saved until you press Save. Writing the draft uses about 1 AI credit. The box stops at 160 characters, and the counter shows how many you have used out of 160. Aim for 50 to 160.

This is one of the few things only you can really write, and it is worth doing by hand or at least editing the draft. It is the sentence Google shows under your store name.

Categories tab: category descriptions

A category page is how a shopper reaches every product under it, and Google has little to rank when the page has no copy of its own. The Categories tab lists your categories, marks the ones with no description ("No description, 12 products") and shows the word count of the rest. Write description drafts 150 to 300 words for one category, and Write all N with AI does every empty one. Rewrite redoes one that already has text. Each description costs a small fraction of a credit (about 0.16). Click a row to open that category and edit it by hand.

Like products, the category scores cover your 100 most recently updated active categories.

Authority tab: trust signals outside your pages

The Authority tab is about signals that tell Google your store is a real business:

• Social profiles linked. Paste your public profile URLs and Brainerce publishes them as sameAs links so Google can connect them to your brand.
• Customer reviews. Shows how many published reviews you have and whether the automated review request, sent after each delivered order, is active.
• Google Business Profile. Connect the Google app, then choose your location. This helps local results such as Maps.
• Backlink and Domain Rating monitoring. Marked Soon and greyed out. It would need an outside SEO data provider, and it is not available today.

None of the Authority actions spend AI credits. They also do not build links for you: SEO Autopilot does not buy, trade or place backlinks.

Measurement tab: Google Search Console

The Measurement tab connects Google Search Console, Google's free first-party report of how your store performs in organic search. Once connected and a verified property is chosen, you see clicks, impressions, average click-through rate and average position, a clicks trend, your top queries and top pages, and how your own articles perform on Google. Data appears within about a day of the first sync, and Sync now refreshes it.

The connection asks Google for read-only Search Console access and nothing else. It is separate from Google Ads. Search Console only knows about your site after you have verified it with Google, which is covered in Get found on Google. Organic Search Console data is also not the same thing as the paid search terms shown in Google Ads campaigns.

AI visibility tab

The last tab measures how AI assistants treat your store: visitors that arrive from them, how often they read your catalog, and what ChatGPT, Gemini and Claude answer when asked shopping questions your customers would ask. It is covered in full in Your store in ChatGPT and AI search. Generating the question list and running a sampling both spend AI credits, and the buttons show the estimate first.

Page speed (Core Web Vitals)

Under Technical, you'll find your store's page-speed score. This is Google's Core Web Vitals: how fast your homepage loads and how stable it feels while loading, measured the same way Google measures it.

Two kinds of data show up here. Real visitor data (from Chrome) only appears once your store has enough traffic. A new or small store won't have it yet, and that's normal, not a problem. Until then, you'll see a simulated test instead, run the same way every time, with a list of the biggest concrete opportunities to fix (like "properly size your images" or "remove render-blocking scripts") ranked by how much time each one would save.

Page speed is a real Google ranking signal, but a minor one: Google itself describes it as a tie-breaker between pages that are otherwise similarly relevant. A slow page won't outrank thin content just because it's fast. Click Check now to run a fresh check any time; otherwise it refreshes on its own about once a week.

How often it runs, and what it costs

Scheduling is available on every plan, including Free. You can set it to run daily, twice a week, weekly, every two weeks, monthly, or not at all (the Off mode).

What limits it is AI credits, not your plan tier. A Free store gets 30 credits a month and an article costs roughly 1.5, so a daily cadence will run out partway through the month.

What spends credits, approximately:

• An article: about 1.45 credits, plus about 0.2 if cover images are on in Settings. In Review first or Full autopilot, this is charged on every scheduled run.
• Refreshing a decaying article: about the same as a new article.
• Plan more keywords: about 2 credits.
• Regenerate voice (the brand voice in Settings): about 1 credit.
• A category description: about 0.16 credits each.
• Homepage meta description, Create with AI: about 1 credit.
• Fix with AI on products: each option in the dialog lists its cost before you confirm, and nothing runs until you press the confirm button.
• AI visibility: generating questions and each sampling run spend credits. The buttons show the estimate.

Free of AI credits: the ranked list and scores, the storefront technical check, the page-speed check, the weekly orphan-page scan, the Authority tab and everything on the Measurement tab.

If you run out of credits, runs fail with a credit error rather than silently stopping. Top up or lower the cadence.

Large catalogs: the scores cover a sample

The products section scores your 100 most recently updated active products, and the footer says how many it scored out of how many you have (for example "scored 100 of 840 products", marked sampled). The category scores likewise cover your 100 most recently updated active categories. If you have more than that, the store score and averages describe that sample, not every item. Products you edited a long time ago can be missing from the ranked list until you touch them; use the SEO column in the products table to check any single product.

Per-product SEO scores in the products table

Separately from this page, the products table carries an SEO column. Click a product's score to open that product's own audit: what is missing, what is weak, what is fine.

The score and the colours mean exactly the same thing in both places. Green is 80 and above, amber is 60 to 79, red is below 60.

The audit also tells you how much fixing one product moves your store's average, and links back here.

Use the products table when you are already working on a product. Use this page when you want to know which products to work on.

Does the SEO audit penalize HTML in my product description or a Hebrew URL slug?

No to both, with one small condition each.

• A description with formatting (HTML). HTML is a supported description format, and the audit never asks you to remove formatting. It only asks that the description be marked as HTML. That happens automatically when you save a description that contains HTML tags, and if the audit still flags one saved earlier ("Description format matches its content"), one click on Mark as HTML fixes it. The text itself does not change. Plain text that merely contains a "<" or ">" sign, such as "fits sizes < 5", is not treated as HTML.
• A URL slug in your own language. If your store's language is not written in Latin letters (Hebrew, Arabic, Russian, Chinese and others), the slug check is skipped and listed under Not applicable to this product. A slug in your own language works in Google Search, so it costs you nothing. In a store whose language is written in Latin letters, a slug with non-Latin characters still loses 2 points.

Does the SEO audit penalize missing translations if Multi-Language is off?

No. Three checks only count when Multi-Language is switched on in Settings: Translated to all store languages, Description language matches store locale and Alt text matches store language. With it off, your website shows one language, has no pages in other languages and sends Google no language tags, so a translation could not help your ranking. The three checks are listed under Not applicable to this product with the reason "Multi-language is turned off".

Listing a second language does not switch Multi-Language on. A store can list extra languages (they also decide which languages your emails can be written in) and still show only one language on the website. Translations saved in that state are kept, but shoppers and Google do not see them until you turn on Enable Multi-Language. Once it is on, the three checks start counting again.

Orphan pages: products nothing links to

A product with no category and no article pointing to it can be genuinely hard for a shopper (or Google) to ever find, even though the page itself is live. Under Content, SEO Autopilot lists these as orphan pages.

If a listed product is already indexed by Google despite being unlinked, that row is flagged as the most urgent kind: Google found it some other way, but nothing on your own site currently leads a visitor there. The fix is usually simple: add the product to a relevant category, or link it from a related article.

If your storefront is a fully custom, JavaScript-rendered site, the scan may not be able to verify links by visiting your live pages, and will say so rather than guessing. It still checks your catalog data (category assignment and article links), so you'll always see something. Just know that a live-verified result isn't guaranteed on every storefront setup. Click Scan now to run a fresh scan any time; otherwise it refreshes on its own about once a week.

Fixing things with AI

Where AI can genuinely help, there is a Fix with AI button. It can write a meta description, a product description, a product title, alt text for an image, and product questions and answers. It can also enlarge a photo that is too small for Google Shopping.

It enlarges your real photo. It never generates a product image. A generated photo of a product you actually ship is a returns problem, not an SEO win.

It will not invent a barcode. GTIN, EAN, UPC and manufacturer part numbers are real numbers printed on real packaging. If a product is missing one, the audit will tell you, and you have to type it in yourself.

Every AI fix spends credits. On this page, Fix with AI asks you to confirm the credit cost first, and then writes and saves the result straight away, with no separate preview step. Review what it wrote afterwards on the product or category itself, where you can edit it by hand. The AI assistant in chat works differently: it shows you the change for approval before saving.

Ask the AI assistant about your SEO

You do not have to open this page to use it. The AI assistant reads the same scores, so you can ask "which products are hurting my SEO?" or "write descriptions for the worst five" in chat and it works from this exact data, not from general SEO advice.

Ask BOBO to do the same on a schedule. There is no separate SEO agent to hire, because BOBO is the only AI assistant. See BOBO, your AI agent for how its approval flow works.

The assistant spends the same AI credits the buttons on this page do, and in chat it shows you the change for approval before saving it. It is held to the same limits: it will not invent a barcode, a brand or a product photo either.

What SEO Autopilot does not do

It will not publish changes behind your back. In Review first mode, articles wait for your approval. Full autopilot is opt-in: it publishes articles without review and asks you to confirm, with the credit cost, before it switches on. Even in Full autopilot, a refreshed version of an existing article always waits for your approval.

Off means no scheduled writing. Generate article now still works when the mode is Off, because you pressed it yourself.

It does not build backlinks. The Authority tab shows trust signals you can set up yourself. Backlink and Domain Rating monitoring is marked Soon and does not exist yet.

It is not a rank tracker. It will not tell you your position in Google for a keyword, track your rankings over time, or compare you to a competitor. If you connect Google Search Console, you get the real clicks and queries Google reports, which is a different and more reliable thing.

It does not submit your store to Google. Getting indexed is covered in Get found on Google.

It does not touch your storefront's code. If your storefront is missing structured data or canonical URLs, the Technical section will tell you, but fixing it means changing the storefront itself.

Page speed is measured on your homepage, not every page. The Core Web Vitals check reports on your storefront's homepage speed, used as a representative sample. It does not run a separate speed test for every product or category page.

What's next?

• Add a product has the slug and meta description fields this page checks.
• Get found on Google covers indexing and Search Console.
• Your store in ChatGPT and AI search is the AI tab on this page.
• Content pages and blog is the other side of what search engines read.