Frequently asked questions
Articles and publishing, keywords and pricing. Agencies, the AI site audit and your data too.
Getting started
What is Probeli AI?
Probeli AI writes and publishes articles that help your buyers find you on Google and in AI search. It finds the keywords your buyers search, writes an article for each one from research, and publishes it to your blog on its date. If you have no blog, we set one up on your own domain.
Do I need to know SEO?
No. Probeli AI does the SEO work: it finds the keywords, plans the articles, writes them from the top search results, and adds the title, the meta description, links and structured data.
Your part is to approve the keywords in setup. Each one shows how many people search it a month and how hard it is to rank for, in plain words. You can read or change any article, but you don't have to.
What is AEO (or GEO)?
AEO (answer engine optimization) and GEO (generative engine optimization) are two names for the same work: getting your brand mentioned and cited when people ask AI assistants for advice. It's close to SEO, but the result is an answer that mentions a few brands, not a page of links.
AI assistants build their answers from pages they can find and trust. Probeli AI writes and publishes researched articles on the topics your buyers ask about, so there are more of your pages for search engines and AI assistants to find.
Who is Probeli AI for?
Probeli AI is for anyone who wants customers from Google and AI search: founders and small businesses (a SaaS product, a shop, a local service), marketing, SEO and brand teams, and agencies that do this work for clients.
Each website gets its own workspace and its own plan, so your brands stay separate. All workspaces shows each one on a single page. You do not need technical skills. Setup starts from your website address, and you add no code to your site.
How long does setup take?
A few minutes, with nothing to install. Enter your website and Probeli AI reads your homepage and suggests keywords your buyers search. You approve the keywords and choose where articles go.
Your first article starts as soon as you finish. After that, Autopilot writes the next ones on your calendar.
How long until I see a change?
Your first article starts as soon as you finish setup, and the next ones follow your calendar.
Results take longer. A new article can take weeks or months to show up on Google, and a change in AI answers can take weeks too. Some pages never rank, and some answers never change. Probeli AI shows you each article's clicks once Search Console is connected, but it cannot promise a result.
How it works
Can Probeli improve the posts and pages I already have?
Yes. Your pages (under Optimize) lists the posts and pages of your own website: the ones your sitemap lists and, once Search Console is connected, the ones Google shows. Articles Probeli wrote for you are left out, since they already have their own refresh. "Scan pages" reads up to 25 of them, the way AI crawlers do.
Each page gets a free report. It has a score for Google and AI answers, with the checks behind it: does the page answer first, are its headings questions people ask, does it use lists and tables, structured data, a visible date, text that shows without JavaScript, a good title and description, and internal links. The report also shows the page's Google clicks and impressions. The article checks apply to articles only: a contact, policy, product or listing page isn't asked for more words, question headings or a date, and checks that don't apply leave its score. The fixes come in order, each saying whether a suggestion can cover it or you do it on your site. An address that now redirects to another page of your site shows where it goes instead: that page has the report.
"Get suggestions" reads the page again and keeps only its article, with no menus, related posts, comments or product grids. It then suggests what to change: a new title or meta description, up to three rewritten sections, two sections to add, and five outdated facts. Each one shows the text before and after. Copy the new text into your site yourself, then mark it done or dismiss it: Probeli never changes your live page. "Read again" spots the suggestions you've applied.
Suggestions for a page use one refresh, never an article: 10 a month on Articles 30. One that finds nothing to change doesn't count. A page can get new suggestions once a day.
Where does the keyword research come from?
From DataForSEO Labs, our data provider for search data. A keyword run reads the keywords your site and your competitors rank for on Google, adds keyword ideas, and groups them into topics with their monthly searches, difficulty and intent. Each keyword shows whether a page of your site already covers it.
Keyword runs a month: 2 on Articles 30.
With Autopilot on, a new keyword run can also start by itself when your topics run out. It uses one of the month's runs.
Which tools does Traffic connect to?
Google Search Console for clicks, impressions, queries and indexing, and your analytics tool, Google Analytics or DataFast, for the visits that come from AI assistants. Each article then shows the clicks and visits it brings in.
Probeli AI only reads Search Console and your analytics. It changes nothing there, unless an owner or admin lets it send your sitemap to Search Console. You can disconnect at any time.
What are content opportunities?
They're the topics your buyers search where your site has nothing that answers them yet. Probeli AI finds them in your keyword research: the keywords other sites rank for that no page of your site covers.
From a keyword you can start an article. Probeli AI researches what already ranks, reads your own pages for facts about your product, and writes the article. It goes live on its date, or waits for your review, as you choose.
Articles and publishing
How do articles get on my site?
Probeli AI publishes them for you. In setup you choose where they go: your WordPress, Webflow, Shopify or Wix site, a webhook to any other system, or a blog we set up for you on your own domain. You connect it once.
After that, each article goes live on its date. You can approve each one first instead, if you prefer. You can also copy any article as HTML or Markdown, or export them all at once.
Why not write the articles with ChatGPT myself?
You can, and the writing may be good. The part a chat cannot do is tell you which article to write, or publish it for you every week. In Probeli AI each topic comes from a keyword your buyers search, with its searches a month and its difficulty.
From there the draft is built from research, not from memory. Probeli AI reads up to 8 of the top Google results for the keyword first, and each claim points to the page it came from. Your products, plans and prices come from a fact sheet built from your own pages, and the draft is checked against it.
Sometimes the search finds no page about the keyword, too few pages to use, or only pages about another meaning of it. Then the article is written from general knowledge. It leaves out prices, statistics and other numbers that no page confirms, and the article page says why.
A second model then checks it and asks for at most 1 rewrite. You get an article with its title, meta, outline, FAQ, structured data and links to your own pages, published on its date or kept for your approval, as you choose. With fact checks on, Probeli AI reads the sources behind its dated claims again about once a month and updates the numbers that changed.
Does AI-written content hurt SEO?
Not by being AI-written. Google's guidance judges content by how helpful it is, not by how it was made, and thin or copied pages do badly whoever writes them.
So each article starts with research: Probeli AI reads up to 8 of the top Google results for its keyword and takes product facts from your own pages. The draft gets a FAQ, structured data, sources and images, and is checked before it reaches you. No tool can promise rankings or AI citations, and we don't.
I don't have a blog. Can I still use Probeli AI?
Yes. We set up a blog for you on your own domain, like blog.yourbrand.com or yourbrand.com/blog. It is your blog, on your site.
Once it is set up, Probeli AI publishes your articles there, on their dates or when you approve them.
Do I approve articles before they go live?
You choose. In setup you pick one of two ways, and you can change it later in the publishing settings:
- Approve each article. Every article arrives as a draft. You read it, edit it in the editor (or ask for an AI edit of a sentence, a section or a part of a section), and publish it when it is ready.
- Publish automatically. Articles that pass the quality check go live on their date. You can still read and edit any of them afterwards.
Autopilot plans and writes articles on a schedule you set, and shows what it will cost before it starts. It takes its topics from your keyword research. Its articles follow the same choice: drafts for your approval, or published automatically.
With automatic publishing you can also set a minimum score. An article that scores lower is not published. Probeli first tries to fix it with your plan's AI edits. If it still scores lower, it stays a draft marked "Held", and you get a notice with what to fix.
Will the articles sound like us?
Yes. Probeli AI reads your own pages first. Until you change anything, each article writes the way your site does: when your site says "we", the article says "we" and "our" too. Facts about your products, prices and plans come from your own pages, not from guesses.
In the writing settings you can set the tone, the form of address, the length and words to avoid, and add instructions for every article. Probeli AI also learns from the edits you publish.
Which sites can Probeli AI publish to?
WordPress (your own site or WordPress.com), Webflow, Shopify and Wix: connect your site once in the workspace settings. Articles publish there on their date, or when you approve them. A webhook sends finished articles to any other system you run.
No blog yet? We set one up for you on your own domain, like blog.yourbrand.com.
Anything else works by hand: copy an article as HTML or Markdown, download it, or export every article at once as a zip file.
What counts as one article?
One article is one finished draft: research, brief, writing, fact check, FAQ, structured data, images and internal links. It counts when it's delivered. One that fails or that you cancel doesn't count, so trying it again costs nothing extra. If our data provider is down while an article is being researched, the article shows "Waiting for the data provider" and tries again by itself for a few hours. It's still the same article, so the wait never counts as another one.
A redo of an article counts as one. Refreshes never do: a refresh of a published article and suggestions for a page you already have use a refresh (10 a month on Articles 30). AI edits in the editor have their own monthly count too (100 a month on Articles 30).
Articles 30 includes 30 articles a month. Unused articles don't carry over to the next month.
Which languages can articles be written in?
English, French, German and Spanish. You pick the language and the country with each keyword, so a workspace can write for several markets, and the research reads the results of that country.
Can Probeli AI keep published articles up to date?
Yes. A refresh reads a published article against fresh results, rewrites its weak sections and adds what's missing, and leaves the parts you edited alone. You see every change before you publish it. A refresh uses one of your monthly refreshes, never an article, and gives it back if nothing needed changing. Articles 30 includes 10 refreshes a month. Suggestions for one of your own pages draw on the same count.
With fact checks on, about once a month Probeli AI also reads the sources behind a published article's dated claims again, up to 8 pages, and updates the numbers that changed. A sentence you edited is flagged, never changed. A fact check doesn't use an article.
Can I go back to an earlier version of an article?
Yes. Open History in the article's ⋯ menu; in the editor, the save note at the top ("Saved") opens it too. Probeli AI keeps the version each generation delivered, and the text before each editing session, AI edit, regeneration, fact check update and restore. Every version shows what changed since, line by line, with the changed words marked.
Owners and admins can restore a version. The version it replaces stays in History, so a restore can be undone too.
What can I add to an article in the editor?
Headings, lists and checklists, an "In short" summary box, FAQ questions, tables, dividers, images and YouTube videos. Type / on an empty line to pick one, or use the toolbar. Each is saved as plain Markdown, so it publishes the same way wherever the article goes, and a FAQ question you add is part of the article's FAQ structured data.
Click an image to change its alt text or remove it. Owners and admins can also replace it from the image library.
Can my team comment on an article?
Yes. Select words in the article and choose Comment. Everyone on your team can comment and reply, and the commented words stay marked in the text until the comment is resolved. Owners and admins resolve comments. Clients you invite as viewers never see them.
Comments are notes beside the article, never part of it: nothing is added to the text you copy or publish.
Does the editor know about my own pages and products?
Yes. The editor tells you at the top when another page of your site, or another of your articles, already targets the article's keyword, shares its topic or has nearly its title. It links to that page, because two pages on one keyword compete with each other. A page from Your pages also links to its report, where you can get refresh suggestions for that page instead.
A new article also links to up to three of your own posts, guides or studies on its subject. Each link sits in the section it fits, such as a packing guide in the packing section. When your article quotes a post, it links to that post. When your site turns our reader away, a Google search of your site finds those pages, and the brief says when one of your posts or guides may already cover the keyword. When your business brief lists no offers, and you chose no call-to- action page in your writing settings, the article's call to action links to your service page on its subject instead of your homepage. That is a page for all your customers, or the page for the article's city when the article targets one. It is never another city's page.
Until you save the Writing settings page, a new article speaks the way your own pages do. When your site writes as "we", the article says "we" and "our" for your products, guides and studies. Accepting a suggestion learned from your published articles does not count as saving. Once you save the page, articles keep to the voice it shows.
The Summary tab lists the offers from your business brief that fit the article's topic and that the text doesn't name yet. Each one comes with the section where it fits, and its link to copy.
A YouTube video in the article shows its channel, length, views and upload date, once Probeli AI has looked it up. It also says when its title is about another platform than the article. Owners and admins can replace it with another video from a YouTube search.
When a section names one of your products that has a page of its own, the article shows that page's picture right after the paragraph naming it. The alt text says what the product is ("Preview of Nova, a managed IT services website theme"). The research may say which thing a name in your keyword means: a web framework, not a game of the same name. A video about the other thing is then skipped, in the article and in the editor's YouTube search. In a how-to or troubleshooting article the video sits right after the steps it shows.
How do AI edits work in the editor?
Select text and choose AI edit: rewrite, expand, shorten, add a sourced fact, fix it for the market, or write a custom instruction in your own words. The toolbar's AI menu also regenerates a whole section or one of its subsections, with instructions if you like. Owners and admins make AI edits, on articles Probeli wrote.
The proposal shows in place: the old text struck through, the new text under it with Accept, Discard, Try again and Edit instruction. Nothing changes until you accept, and one Undo takes it back. History keeps the text from before the edit.
Each request, Try again included, uses one AI edit of your plan's monthly count, never an article. A subsection too long to regenerate is dimmed in the menu: select text in it instead.
Can I find and replace words in an article?
Yes. Press Ctrl+F (⌘F on a Mac) in the article's text, or choose Find and replace at the end of the toolbar. The matches are counted and marked in paragraphs, headings, lists, tables and the sources, never inside an image or a video. Match case and Whole words only narrow the search.
Show replace adds the second field. Replace changes the current match and keeps its bold or its link; Replace all changes every match at once, and one Undo takes them all back. A replacement saves like any other edit. Find and replace is in the editor, where owners and admins edit.
Can I hide an article's sources and bring them back?
Yes. Turn off Show sources in the article, in the editor's Sources card: the Sources list and the [1], [2] markers in the text are removed. Turning it on again puts both back, each marker after the sentence it cited, except in a sentence you rewrote since. An earlier version in History still has them too.
What does the video card in an article show?
The video's title, channel, length, views and upload date, as Probeli stored them when it picked or checked the video. A video added by its link alone shows its title only. The player loads on your site, not in the editor, so the card has no thumbnail.
When the video's title names another platform than the article's topic, the card says so, for example "This video is about React, not Astro." Owners and admins can replace it. Replace opens the YouTube search with the focus keyword filled in, and the video you pick takes the old one's place.
Why does the editor say my page already targets this keyword?
Because one of your own pages or articles is about the same keyword, and two pages on one keyword compete with each other. The editor compares the article's focus keyword and title with your own pages and your other articles. It compares Your pages and the pages your business brief read or links to. It then names the closest ones, with a link to each.
Update that page instead, or give the article another keyword and angle. Dismiss hides a match on this article in your browser. The same check runs before writing. When one of your pages already has the keyword in its title, the brief waits for your approval with Update this page instead, and Autopilot skips the keyword.
What does the Your offers card show?
The offers from your business brief that fit the article's topic and that the text doesn't name yet. An offer fits when it shares a word that sets it apart from your other offers, such as "agency" for "best agency templates"; words like "best", "top", "free" or a year never count.
Each offer comes with its page, the section it fits under with a Go to, and Copy link, which copies a link to paste into the text. The offers the text already names are listed on one line. Without a business brief, the card links to it.
Why does the editor say editing would change the text?
The saved text holds something the editor can't keep, such as checkboxes in a numbered list, so the editor stops before it changes anything. The notice names what would be lost. Nothing was saved, and the title, summary and meta fields still save as usual.
Copy saved text copies the article's Markdown untouched. Edit text anyway makes the text editable and saves it as the editor shows it, without the parts the notice named.
How are an article's title, URL slug and meta description written?
When an article is written, one set of rules checks the title, the URL slug, the SEO title and the
meta description. They are checked against the finished text, and against the top 5 Google results
the research found. The slug is short: the keyword's own words, without small words, years or list
counts, and never a whole question ("what is exit rate" becomes exit-rate). It follows how the
top results name their pages: "best" is left out when they leave it out, and when their addresses
are longer, the slug adds what the finished article is about (drano-alternatives-clogged- drains). It never ends on a filler word such as "explained" or "overview"; when another of your
articles already has the slug, another angle of the title is used before a number. The SEO title
keeps the keyword. A title never doubles a superlative ("Top Best"). It never puts "Best" in front
of statistics or another keyword that ranks nothing. It names the year only when the article is
about a year's data or events. The meta description the writer wrote is kept as written, unless
something is wrong with it. It is replaced when it is shorter than 110 or longer than 160
characters, stops mid-sentence, names nothing the article is about, or gives a count the article
contradicts. It is also replaced when it writes a name in the wrong case, opens with stock
phrases, has the keyword pasted in front of a sentence, or says mostly what one of the top
results' descriptions says. A count the article gives otherwise is corrected; any other
replacement must pass every one of these checks. The keyword is never added to the front of a
description. For "best" lists, the title keeps "best".
Once an article is written, the SEO title and description panel above the title shows its URL slug, SEO title and meta description even while it is closed; open it to edit them. When you change the title, the focus keyword or a good part of the text, the panel shows a suggested slug, SEO title or meta description. Choose Use to take one; it saves like any other edit, and nothing changes until you do.
A published article keeps its slug: changing it would change the article's address and break links to it, so its suggested slug is shown for reference only. A content refresh keeps the title and the address. It updates the SEO title or meta description only when Probeli wrote them and they no longer fit the refreshed text. The ones you wrote stay, and any suggestion is shown in the editor.
Do you tell search engines about my new pages?
Some of them. When Probeli publishes a page, changes it or takes it down, it can tell Bing, Yandex, Naver, Seznam and Yep. It uses IndexNow, an open protocol those engines read. It is free and needs no account. the Internet Archive and Amazon's crawler take part in IndexNow too, so they hear about the page as well.
Those engines still choose when to crawl the page. What changes is that they hear about it at once.
Your workspace shows whether this is on for you, in Settings · Publishing. It needs one small file on your site first (below).
Your Indexing page puts it all in one place: every page Probeli knows about, whether Google has indexed it, and what the engines above were told about it.
Google is not one of them. Google has no way for a tool like Probeli to ask it to index a page. Its Indexing API is only for job postings and live broadcasts. The old sitemap ping was switched off in 2024. So we don't claim to submit your pages to Google.
For Google we do the part we can. For every page you publish, Probeli says whether Google has
indexed it. When it hasn't, Probeli says why, in plain words: a noindex tag, a robots.txt rule, a
canonical pointing at another address, a page Google read as empty, a page it has read and left
out, or one it hasn't read yet. Each reason comes with the fix.
Then Probeli links you into Search Console, where you ask Google to index the page. That step is yours: we can take you to the right place, not press the button.
That file is a plain text file Probeli makes for you. You put it at the root of your site, so the engines can check the key is really yours. Settings · Publishing shows you the file, where it goes, and checks it for you. On a blog we set up for you there is nothing to do: we add that file ourselves.
Pricing and plans
Do I need a credit card to start?
Yes. You add your website, then pick a plan and pay through Paddle. Your first article starts right after you pay.
7-day money-back guarantee: if it's not for you, we refund 100% within 7 days of your first payment. No questions asked.
Already on a free trial? What happens when it ends?
The plan you picked starts and your card is charged, unless you canceled before the end. The day before, a notice in the app shows the amount and the date, and we email you once the plan has started.
If you canceled, new articles stop and your data is kept. You can still open your articles and your workspace, and the AI site audit keeps working. Choose a plan in your workspace settings to start again.
How much does Probeli AI cost?
Articles 30 costs $39 a month for 30 articles. Paid yearly, it costs $300, which saves $168.
It finds your keywords, writes the articles and publishes them. Prices are in US dollars, and the plan covers one website (see the plans).
Is there a launch offer?
Articles 30 sells at a launch price. The price you start on stays yours for as long as you stay subscribed; if we raise prices later, only new customers pay more.
For agencies, our first 50 founding customers get 40% off the first 6 months of Agency on monthly billing. On yearly billing they get 30% off the first year, whenever the plan cards show that price. It's applied at checkout, and runs until 31 December 2026 or until the spots run out.
I'm already a customer. Does my plan change?
You keep your plan. The new article plan, Articles 30, is for new sign-ups.
If you're on Grow or Pro, your plan moves to a lower price at your next renewal. You keep everything it includes now, your team and your monthly allowances too. We email you before the new price applies, and you don't need to do anything.
Do all plans include the same features?
Yes. Every plan includes keyword research, the content calendar and Autopilot, articles with AI edits and refreshes, publishing, Your pages reports, Traffic and the AI site audit.
Plans differ in how much they include: articles, AI edits, refreshes and keyword runs a month, team members and client viewers.
Is there a plan for agencies?
Yes. Agency ($249 a month) covers 5 client workspaces and Agency Pro ($599 a month) 15, with one shared pool of articles, white-label client reports and client viewer logins. See Probeli AI for agencies for how billing and clients work.
Can I use Probeli AI for several websites?
Yes. Each website is its own workspace, with its own keywords, articles and plan. Add one from the switcher at the top of the sidebar and switch between them there. The agency plans, which cover several workspaces, are for agencies managing client websites, not for a business's own websites.
Once you have two or more, All workspaces in the switcher puts every workspace you belong to on one page. For each one you see what needs attention, with the next step. You see which website needs you without opening each one.
Can my team use Probeli AI?
Yes, on every paid plan. Articles 30 includes 5 team members.
Owners and admins manage articles, keywords and publishing; members see everything but change nothing. On Articles 30, you can also invite viewers, who only see the dashboards and don't take a member seat. Plans are priced per workspace, not per seat, so adding a member never changes your bill.
Owners and admins can invite several people at once by pasting a list of email addresses. If you own other workspaces, you can also invite the same people to several of them in the same step. Each workspace's plan and member limit apply, and each person gets one email for all the invitations from that step rather than one per workspace.
What can owners, admins, members and viewers do?
Each person has one role in each workspace. Owners and admins run it: they change keywords, articles and settings, start runs and audits, invite people and manage the plan. Only an owner can change the website, open the billing portal or delete the workspace.
Members see everything the team works on, including the Team and Billing pages, but can't spend or change anything.
Viewers are for people who only need to look. They see the dashboards and can't change anything, invite anyone or see billing. Viewers don't take a member seat: Articles 30 includes 25 viewer seats per workspace.
Anyone but an owner can leave a workspace from the workspace menu. Owners and admins change a person's role or remove them in the workspace's Team settings.
Can I change or cancel my plan?
Yes. To move to another plan or switch between monthly and yearly billing, open Billing in your workspace settings (owners and admins). The confirmation shows what the change costs when Paddle can work it out, and the new plan applies right away. The change is prorated: you're charged the difference now (for example for an upgrade), or it's credited to your Paddle balance and pays toward your next bills (for example for a downgrade). Credits aren't paid out. If the payment for a change fails, your plan stays as it was. Have a discount code? Enter it in the confirmation ("Have a discount code?"): the cost shown then includes the discount, and it applies to the new plan right away.
A downgrade never deletes anything: your articles and settings stay, and the new plan's monthly limits apply from then on.
To cancel, the workspace owner opens Billing in the workspace settings and clicks "Cancel plan". The first step asks why you are leaving: pick one reason, and a note is optional. The second shows when the plan ends and what stops, with the "Cancel plan" button right there. "Manage billing" in Paddle's portal works too, without any question. Your plan runs until the end of the billing period, nothing more is charged, and your data stays. If you only need a break, the second step offers to pause the plan instead, for one to three months from the end of the paid period. An agency plan cannot be paused. Nothing is charged and new articles stop while it is paused. It starts again by itself on the date shown, when the next period is charged. You can resume early or cancel at any time. Checkout and invoices are handled by Paddle, our reseller and Merchant of Record.
What happens if a payment fails?
New articles pause until the payment goes through, and nothing is deleted. There's no grace period: the app shows a "Payment failed" notice right away and asks you to update your billing details.
Once the payment succeeds, your plan carries on by itself. Your articles and settings stay in your workspace the whole time.
Can I cancel in the app?
Yes. The workspace owner opens Billing in the workspace settings and clicks "Cancel plan". Your plan runs to the end of the period you paid for, and your data stays.
I paid, but the app asks me to pay again. What do I do?
You are probably signed in with a different email than the one you paid with. Open Find my subscription and type the email on your Paddle receipt. If an account uses that email, we send it a sign-in link, and your workspace with its plan is there.
If you paid with an email that has no Probeli AI account, contact us with your receipt. We link the payment to your account by hand.
What happens once I sign up?
You add your website and pay. Then you approve your keywords, and Probeli AI starts writing and publishing for you.
7-day money-back guarantee: if it's not for you, we refund 100% within 7 days of your first payment. No questions asked.
Is it really all automatic?
Almost. The setup takes a few minutes: your website, your keywords and where to publish. Then Autopilot plans, writes, adds images and links, publishes and tells search engines. You can also approve each article first.
What if it's not right for my business?
7-day money-back guarantee. If it's not for you, we refund 100% within 7 days of your first payment. No questions asked. Articles are written for your trade, your market and your language.
Does it work with my website?
Yes, in most cases. Probeli AI publishes straight to WordPress, WordPress.com, Shopify, Webflow and Wix, so each article becomes a real post on your own site. Any other site can receive articles through a webhook, which your developer sets up once. You connect your site once in the app, and no code is needed for the platforms above. No blog yet? Then we set one up for you on your own domain.
How do the articles get onto my site?
You connect your platform once in the app. After that, each article goes live as a normal post on your site, with its images. You can also copy an article, download it or export them all.
Will my articles show on Google and in AI answers?
Nobody can honestly promise a ranking. Each article gets an SEO title, links to your pages and its sources, and an IndexNow notice the day it goes live. Search Console then shows when Google starts showing it.
Can I cancel any time?
Yes. The workspace owner cancels under Billing in the app. Your plan runs to the end of the period you paid for.
7-day money-back guarantee: if it's not for you, we refund 100% within 7 days of your first payment. No questions asked.
Agencies
Do you have plans for agencies?
Yes. Agency costs $249 a month for 5 client workspaces, your own included, and Agency Pro $599 a month for 15. On Agency Pro each client after that is $39 a month.
Every client workspace gets every feature. Articles come from one pool the clients share: 100 a month on Agency, 300 on Agency Pro.
The agency plans are for agencies and consultants who manage websites for client businesses. A business with several websites of its own takes a plan for each website instead.
How does billing work for my clients?
One subscription on your agency's workspace pays for every client it covers, with one invoice from Paddle. Billing shows each client's share of the article pool, so you can bill your clients as you like.
Clients never need an account with us or a card of their own.
How do I add a client?
Click "Add client" and enter the client's website. While your plan has a free slot, the client is covered right away, with no checkout. When Agency is full, the app offers Agency Pro; when Agency Pro is full, one more client at $39 a month, prorated.
Each client gets its own workspace: its keywords, articles, publishing and reports, kept apart from the others.
Can I send clients reports with my own brand?
Yes. Each workspace has a monthly client report: the articles published, the results they brought in, and next month's plan. On any paid plan you add your logo and name and turn off "Made with Probeli".
Share it as a read-only link or a PDF, or have it emailed to your client every month. It can be in English, French, German or Spanish.
Can my clients log in?
Yes, as viewers. A viewer sees the dashboards and reports of the workspaces you invite them to, and can't change anything, spend or see billing. Viewers don't take a member seat.
Invite them from the workspace's Team settings. Your team works as owners, admins and members.
AI site audit
What does the AI site audit check?
The AI site audit runs 18 checks of how easily AI search engines and agents can crawl, read and use your website.
The checks come in three groups. Content and structure covers semantic HTML, titles and structured data. Discovery and crawlability covers robots.txt access for crawlers such as OAI-SearchBot and PerplexityBot, sitemaps and llms.txt, bot challenges, and content that only appears with JavaScript. The third group is developer and agent readiness.
The audit also works out what kind of site you have: an online shop, a local business, a SaaS product, a publisher or a blog. It also looks for a sensitive topic, such as money, health or the law. It uses your business brief and the pages it reads. It then adds the checks that matter for that kind of site. For a shop, that means product markup and visible prices. For a local business, your address, phone and opening hours. For software, a pricing page and docs. For a publisher, named authors and dates. For a sensitive topic, expert authors, sources and disclaimers. Technical checks look at page weight and render-blocking files, the mobile viewport, language versions (hreflang), redirect chains and conflicting canonical links. An owner or admin can correct the detected category at any time.
You get a score out of 100 for the core checks. You also get one list of fixes, ranked by how much each one helps your kind of site, each with why it matters and step-by-step instructions. The extra checks read only the pages the audit already fetches, so they don't add requests to your site.
Is the site audit free?
Yes. The free site audit needs no account. Type your website and your email, and you see the results on the page.
We email you the audit, a free sample article written for your website, and one short follow-up. There is no newsletter. Every email has a link to unsubscribe in one click.
Inside a Probeli AI workspace, the AI site audit is included at no extra cost. Owners and admins can run it: one re-run right after the first audit, then one every 24 hours.
Does Probeli AI crawl my website?
Yes, lightly, and only public pages. Probeli AI reads a site in four cases. First, its homepage, when you add it as your website or as a competitor; you can also ask to read your own homepage again in its settings, which shows whether that read worked, and why not. Second, up to 40 of your pages, plus robots.txt, sitemaps and llms.txt, when you run the AI site audit. Third, a few pages of your own website, for its business brief. Fourth, the pages that AI answers link to, to find the brands they name.
The free site audit reads only a website's homepage and one sitemap file, after its robots.txt.
For the brief, Probeli AI reads your homepage and up to 4 key pages, such as about, products or pricing, one at a time. One workspace's brief does not read your site again within 10 minutes.
Every request identifies itself as ProbeliAIBot. The audit, the brief and the cited-page reader
follow robots.txt (the token is probeliaibot) for every request, sitemaps, llms.txt and
redirects included, and limit how fast they send requests to each site. Only robots.txt itself and
the homepage read of your own website when you enter it or ask to read it again aren't subject to
it. If your robots.txt blocks ProbeliAIBot, the AI site audit runs as a limited audit. Our
ProbeliAIBot page lists those requests and has the details.
Privacy and security
What data does Probeli AI store?
For each workspace: your website and its business brief, brands, competitors and prompts, the AI answers with their sources, and the metrics calculated from them. For your account: your name, email, sign-in methods and active sessions, including the device and IP address of each.
The brief is written from a few pages of your website: we keep the address and title of each page read and the short quotes the brief uses, not the full page text.
Card details go straight to Paddle, which handles our checkout, so Probeli AI never sees them. Our privacy policy lists everything, including the kinds of providers we use.
Can I delete my data?
Yes. An owner can delete a workspace in its settings. The workspace is locked right away and all work on it stops, and 3 days later it is deleted for good: its answers, prompts, competitors and insights are removed and its plan ends. Every member gets an email when this happens, and again when the workspace is restored or deleted. Until then, any owner can restore it from the workspace. After that it can't be undone. The workspace logo stays in our file storage until you ask us to delete it.
To close your account, delete it in your account settings. This also deletes every workspace you're the only member of right away (without the 3-day wait) and ends its plan. If you're the only owner of a workspace with other members, make someone else an owner or remove the other members first.
Deleting a workspace or your account needs a sign-in within the last day: if yours is older, sign out and sign in again first.
To stop repeated sign-ups, we keep a small record of each new workspace: your account ID, a one- way hash of your email, a one-way hash of your IP address, and the website's domain. Our Privacy Policy has the details. If you need a hand, contact us.
How do you keep accounts and data secure?
You can sign in with a password, a magic link or a passkey, and turn on two-factor authentication. Adding a passkey, deleting a workspace and deleting your account need a sign-in within the last day: if yours is older, sign out and sign in again first. Only a workspace's members can see its data, and only owners and admins can change it; the server checks this on every request.
Probeli AI staff can access accounts for support, security and abuse handling. Every connection uses HTTPS, you can see and end your active sessions in your account settings, and our error reports never include your prompts, answers, name or email address.
Free SEO tools
Are the free tools really free?
Yes. Every tool on this page is free, and you don't need an account or a card to use it.
Each tool allows 20 uses a day from one network. When someone asked the same thing a short time ago, you may see a saved copy of that result.
What do you store when I use a tool?
We store only the result for what you typed, so the same input can be answered again. Most results are kept for up to 7 days. The AI answer check keeps its result for 1 day, and the meta tag and sitemap checkers for 10 minutes.
To count the daily uses, we also keep a scrambled key of the network you came from for 3 days. It is not the address itself. The tools that run in your browser send nothing to us.
Why is there a "checking you're a person" step?
The tools that call our server use Cloudflare Turnstile to check that a person is asking. This stops scripts from using up the tools for everyone. Most of the time it needs no click.
The tools that run only in your browser have no such step, because they send nothing.
Will these tools get me to the top of Google?
No tool can promise that. The tools can help you write better titles and descriptions, pick better topics and fix simple mistakes on a page.
Your rankings depend on many things, such as the quality of your pages, the other pages on the same topic and the links to your site.
Where do the search volumes come from?
The search volume, keyword difficulty and cost per click come from DataForSEO Labs, for the country you choose. Search volume is the number of searches a month.
When there is no data for a keyword, the field stays blank. We never guess a number.
Why do ChatGPT's answers change?
Each answer you see is one ask, and AI answers change from one ask to the next. Ask the same question twice and the brands, their order and the sources can differ.
So one answer is a snapshot, not a rule: check again another day before you read much into it.
Does the checker follow robots.txt?
Yes. The meta tag checker, the sitemap checker and the blog post ideas tool read a page as ProbeliAIBot. They read the site's robots.txt first and never read a page it blocks.
Each check reads only the address you typed, once. It never fills in forms or runs the page's scripts.
How is the readability score worked out?
The score is the Flesch reading ease. It uses the average number of words in a sentence and the average number of syllables in a word. Higher scores mean easier text.
The grade is the school level that the same numbers point to. The check runs in your browser, so your text is never sent to us.
Which rules check the titles and descriptions?
The same rules that Probeli's article engine uses for every article it writes. A title must name the topic and fit the length. It must not hold a past year, a model's label or stock words such as "Explore" or "Ultimate".
A description must be 110 to 160 characters, name the topic and end with a whole sentence. Each result that fails a rule shows which rule it broke.
Should I block AI crawlers?
It depends on the crawler. Some crawlers collect pages to train AI models. Others read pages to answer a person's question in an AI search engine, often with a link to the source.
Blocking a training crawler keeps your pages out of future training. Blocking a search or answer crawler can keep your site out of that engine's answers.
What is an llms.txt file?
An llms.txt file is a short text file at the root of your site, such as yoursite.com/llms.txt. It tells AI tools in plain words what your site is and links to your most useful pages.
It follows the format described at llmstxt.org. It is a new idea, and not every AI engine reads it yet.
Where do I put the file or code I made?
Put robots.txt and llms.txt at the root of your site, so they open at yoursite.com/robots.txt and yoursite.com/llms.txt.
Paste the schema markup into the HTML of the page it describes, in the head or the body. Most website builders have a field for custom code in each page's settings.
What is keyword density, and what is a good number?
Keyword density is how often a word or phrase is used, as a share of all the words in the text. There is no number that search engines ask for.
Write for the reader and use the words people search for where they fit. When one phrase is a large share of the text, the text can read as written for search engines, not for people. The checker warns you about that.
How does the robots.txt validator decide if a path is allowed?
It follows the robots.txt standard, RFC 9309, which Google also follows. A crawler uses the group
that names it, or the * group when none does. Then the rule with the longest matching path
decides. When an Allow and a Disallow rule match with the same length, Allow wins.
The check runs in your browser. We don't fetch your file, so paste all of it.
What does the meta tag checker check?
It reads the page as ProbeliAIBot and checks the title and its length, the description and its length, the canonical link, noindex (in the page and in its headers), the language, one H1, the mobile viewport, Open Graph and Twitter tags and structured data. Each check says what it found.
What is a URL slug?
The slug is the last part of a page's web address. It comes after your domain and any folder. In example.com/blog/bread-flour-guide, the slug is bread-flour-guide. A clear slug tells readers and search engines what the page is about before they open it.
How long should a URL slug be?
Three to six words is a good length. Short slugs are easier to read, share and type, and search results cut off long addresses. Leave out small words such as "the", "and" and "of" when the slug still makes sense without them.
Should I use hyphens or underscores?
Use hyphens. Google's own advice is to put hyphens between the words of an address. With underscores, the words can run together and read as one long word. All three slugs from this tool use hyphens.
Can I change a slug after the page is live?
You can, but add a 301 redirect from the old address to the new one. Without it, old links and bookmarks lead to an error page, and the page can lose the visits it had from search. This tool doesn't change your page: it only suggests slugs.
What does a canonical link do?
It tells search engines which address is the main one for a page. The same page can open from
several addresses, for example with and without www. Without a canonical link, they may split
the page's value between them. Point it to the page itself.
Does this tool keep what I type?
No. This tool runs in your browser. What you type is never sent to us, and we keep no record of it. You don't need an account, and there is no daily limit.
What is a title tag?
It is the page title in your page's HTML. Google often shows it as the blue link of your result, and browsers show it on the tab. It is one of the first things a searcher reads, so it helps decide who clicks.
How long should a title tag be?
Our rules ask for 30 to 60 characters. Google cuts longer titles in its results, and a very short title leaves out words that could tell the searcher more. Put the main words near the start, in case the end gets cut.
Is there a best keyword density?
No search engine asks for a number. Use your main phrase where it fits: in the title, near the start and in a heading or two. When your most used word is over 3% of a long text, it can read as written for search engines, not people. Then say some of those uses in other words.
Should the title tag name my brand?
On your homepage, yes. On a blog post, the topic usually matters more to the searcher than your name. Our rules leave the site name off the end, and many website builders add it for you anyway.
Why does it show two- and three-word phrases?
People search in phrases, not single words. Someone types "dog grooming prices", not "dog". The phrase lists show if the words your buyers search for really appear in your text. One word can look fine while the phrase you care about is almost missing.
Why does the old card still show after I changed my tags?
The platforms keep a copy of each card for a while. Facebook's Sharing Debugger and LinkedIn's Post Inspector read the page again on request. Paste your address there after you change the tags.
Does it count small words like "the"?
Not as single words. The word list leaves out small common words such as "the", "and", "of" and "for". They are in every text and say nothing about the topic. The phrase lists keep them inside a phrase, as in "the dough", but leave out a phrase made only of small words, such as "of the".
Should the title tag match the H1?
They should be about the same topic, but the words can differ. The title tag is for the result page, where it has to earn the click. The H1 is for the reader who already arrived.
What is robots.txt?
A plain text file at the root of your site, such as example.com/robots.txt. It tells crawlers which parts of the site they may read. A mistake in it can stop Google from reading your pages, however good they are.
How do I know which title works best?
Use one title for a few weeks, then compare clicks in Google Search Console. Change one thing at a time. When you connect Search Console, Probeli's Traffic page shows the clicks for each page.
What are the most common robots.txt mistakes?
Three come up most. Disallow: / under User-agent: *, which blocks your whole site. A Sitemap
line without the full https:// address. And a rule placed before any User-agent line, which
crawlers ignore. The validator flags all three, and blocks of AI crawlers too.
Why not count the words myself?
You could, but it takes a long time. This tool counts every word and every two- and three-word phrase at once. It shows each one's share of the text and warns you when one is used too often.
Does robots.txt keep a page out of Google?
No. It only stops crawling. A blocked page can still show in Google if other sites link to it. To keep a page out, let Google read it and add a noindex tag, or remove the page.
What is a meta description?
It is a short summary of your page in its HTML. Google often shows it under your title in the results. It tells searchers what they will find, which can help them decide to click.
What is the Flesch reading ease score?
It is a score from 0 to 100, and higher means easier. It uses the average sentence length and the average number of syllables in a word. A score of 60 to 70 is plain English for most adults. Below 50 is hard going for many readers.
How long should a meta description be?
Our rules ask for 110 to 160 characters. Google shortens longer ones, and a short one wastes space that could answer the searcher. Each description should end with a whole sentence, not mid-word.
What is the Flesch–Kincaid grade?
It turns the same two numbers into a US school grade. Grade 8 means a typical 13- or 14-year-old can follow the text. For a blog or a shop page, grade 7 to 9 suits most readers. Above grade 12, many readers start to struggle.
Why do I need a sitemap?
It lists the pages you want search engines to find, with the date each one last changed. Without one, Google finds pages only by following links. That is slower, and on a new site it often misses pages.
Does the meta description change my ranking?
Google says it doesn't use it to rank pages. A clear description can still help more people click your result. Google may also show other text from your page in its place.
How does the checker find my sitemap?
It reads your robots.txt and uses the first Sitemap line it finds. If there is none, it tries /sitemap.xml. It reads up to 1 MB of the file, counts what is in it and lists the first 10 entries.
Why do the five scores not agree?
Each score counts different things. Flesch uses sentence length and syllables. Gunning fog and SMOG count long words of three syllables or more. ARI counts letters, not syllables. Small differences are normal, so look at what most of the scores say.
How do I tell Google about my sitemap?
Two ways, and you can use both. In Google Search Console, open Sitemaps and add the address. And
add a line Sitemap: https://yoursite.com/sitemap.xml to the end of your robots.txt, so every
crawler can find it.
Should my score be high or low?
For reading ease, higher is easier. For the grade scores, lower is easier. Most web pages should be easy: almost every reader prefers it, whatever their schooling. A guide for experts can be harder, as long as its readers can follow it.
Why do I get several meta descriptions?
You get up to 5, so you can pick the one that sounds most like you. Each one is checked by our rules first. The ones that pass are shown first, and each failed rule is named.
Can I paste the meta descriptions straight into my site?
Yes, they are yours to use. Read your page once first and check the description matches it. When a description doesn't match the page, Google often shows other text instead.
Does Google use a readability score?
Not as a score: there is no known reading-level number in how Google ranks pages. But easy text helps people find their answer fast, and search engines want to show pages like that. Write for your reader first.
What if my site has no sitemap?
Most platforms make one for you: WordPress, Shopify, Webflow and Wix all do. Look for /sitemap.xml or /sitemap_index.xml on your site. A blog we set up for you has its own sitemap from the first post.
How do I know if my text is easy enough?
Paste it above and read the first line of the result. A reading ease of 70 or more is easy to read. 50 to 69 is fine for most readers but heavy in places. Below 50 is hard to read. Then shorten the long sentences the checker lists.
What is keyword difficulty?
It is a score from 0 to 100 for how hard it is to reach the top results for a search. A low number means fewer strong pages compete. Start with ideas that have searches and a lower difficulty.
Do I need a robots.txt if I don't block anything?
You don't have to have one. Without the file, crawlers may read every page. A short file is still useful: a Sitemap line in it helps every crawler find your sitemap.
How is this different from asking ChatGPT?
A chatbot can suggest titles, but it can't tell you how many people search for them. Each idea here comes with real monthly searches and a difficulty score for the country you pick. With a website, it reads your page first.
Can I save the blog post ideas?
Copy them from the table. In Probeli, the ideas you keep go into your keyword library and your content calendar, so you can plan what to write each week.
Do you keep the address I check?
We keep the result for 10 minutes. We also keep a scrambled key of your network for 3 days, to count your daily uses.
Can Probeli write these posts for me?
Yes, that is what Probeli does. It finds the searches your buyers make, writes an article for each one and publishes it to your blog. You choose how much it does on its own.
Do you keep the file I paste?
No. The validator runs in your browser. Nothing you paste is sent to us or saved.
What does the SEO report check?
Three things. Your Google numbers: estimated visits a month, how many keywords you rank for and your top ones. Searches you could write for, with how often people search them. And checks of your homepage's title, description and tags, and of your sitemap.
What makes a good headline?
It names the topic in the first few words and makes one clear promise. It is specific, with a number or a real outcome, and it fits in a Google result. It also avoids words that sound like an advert.
How long should a headline be?
Aim for 30 to 60 characters. Google shows about 600 pixels of a title on a computer, which is around 55 to 60 normal characters. Wide letters such as W and M take more room, so the quick score measures the width.
Why are some numbers missing from my report?
The report shows only what it measured. A new or very small site may have no Google numbers yet, and then the report says so. It never shows a guess.
Why does the report list searches I could write for?
Each useful article is another search your business can show up for, on Google and in AI answers. The report finds searches about your topics that your site doesn't answer yet. They are the best places to start.
How do I get this report every month?
With a Probeli AI plan, you don't need to run it again. Probeli writes and publishes articles for those searches. The Traffic page shows the Google clicks each article brings, once you connect Search Console.
What are stock words?
They are words that appear in so many headlines that readers skip them, such as "Ultimate", "Unlock" or "Explore". Our rules fail them. Say what the reader gets instead.
How is this different from other headline analyzers?
Many analyzers only match your words against a list. The full check here uses the title rules Probeli's article engine uses for every article it writes. Its 2 rewrites pass the same rules before you see them.
Should I always pick the headline with the highest score?
Not always. A headline that promises more than the page gives may get the click, but the reader leaves. Pick the best headline that is still true to your page.
Can I use the headline rewrites?
Yes. Anything the tool writes for you is yours to use, with no credit needed.
What does it mean when AI recommends a brand?
When someone asks ChatGPT or Google AI Mode which product to choose, the answer often names a few brands. Being one of those names is a recommendation. Buyers who trust the answer often look at those brands first, and may skip the others.
How does this checker work?
We send your question to ChatGPT and to Google AI Mode once each, for the country you chose. We read each answer and list the brands it names, in order. Then we show whether your brand is one of them, in which place, and whether your site is a source.
Why doesn't the answer name my brand?
AI engines name the brands they find often on pages they trust for that topic. A new or small brand may not be on enough of those pages yet. The answer can also change with the country, the exact words of the question and the day.
How can I get named more often?
Publish clear pages that answer the questions your buyers ask, and make sure AI crawlers may read your site. No one can promise a mention. Probeli AI helps: it writes and publishes articles for your buyers' questions, and checks every week whether the engines your plan includes name you.
Why add an llms.txt file to my site?
Without it, an AI tool has to read many pages to work out what your site is about. With it, you tell the tool directly, in a few lines, and point it to your best pages. It takes a few minutes to make.
How is llms.txt different from robots.txt and a sitemap?
robots.txt tells crawlers which pages they may read. A sitemap lists every page you want found. An llms.txt is a short guide: what the site is, and the few pages that matter most, written in plain words. Many sites use all three.
Will an llms.txt get me into AI answers?
Not on its own. It is a new idea, and not every AI engine reads it yet. What AI answers quote are useful pages about your topic. Probeli AI writes and publishes those pages for you, and your llms.txt can point to them.
What is the difference between GPTBot and ChatGPT-User?
GPTBot collects pages that OpenAI may use to train its models. ChatGPT-User reads a page when a person asks ChatGPT about it. OAI-SearchBot builds the index that ChatGPT search uses. You can allow or block each one separately.
Does blocking Google-Extended remove me from Google Search?
No. Google-Extended only controls whether Google may use your pages to train its AI models, such as Gemini. Googlebot, which ranks your pages in Google Search, is a different crawler, and this file doesn't block it.
Do AI crawlers follow robots.txt?
The large AI companies say their crawlers follow it. But robots.txt is a request, not a lock: a crawler chooses to follow it. To block a crawler for sure, your host or firewall has to refuse it.
What is schema markup?
It is code in a fixed format, schema.org, that describes what a page is: a business, an article, a product or a set of questions. Search engines and AI tools read it. It can make the page eligible for richer results in Google.
How do I check that my schema markup works?
Open Google's Rich Results Test, paste your code or your page's address, and run the test. It shows the types it found and any errors. Google can take a few days to show a change in its results.
Does schema markup improve my rankings?
Not directly. Google says it is not a ranking factor. But it can make your page eligible for richer results, such as stars or questions, which can help more people click. Google decides when to show them.
Can one page have more than one type?
Yes. A blog post can have Article and FAQ markup, for example. Make each one with this tool and paste them one after the other. Probeli AI does this for you on every article it writes.
Still have questions?
Can't find what you're looking for? Send us a message and we'll reply by email.
7-day money-back guarantee · Cancel anytime
The content calendar of Webflow as the example brand: this month in month view with an article planned every Monday, Wednesday and Friday, the earlier ones published to blog.webflow.com, the next one being written and the rest scheduled, and the Autopilot card on, publishing the articles that pass the quality check on their date.
A sample article's first days: on the day it went live, IndexNow told five search engines about it, one after another, and on day three Google showed it as indexed.
Get found on Google and in AI search.
Add your website and Probeli AI writes and publishes your first article.
7-day money-back guarantee · Cancel anytime