AI reads your page. And everything said about you.
Lumar content evals read every page of your crawl the way a demanding editor would. Then they research what independent sources say about you, and where competitors win off your site.
An AI answer has many sources. Your site is one of them.
Ask an assistant which bag to buy, which bank to use or which software to choose, and it does not just read your website. It reads reviews, roundups, comparisons, forums and the news, and builds its answer from all of them. A site with perfect pages can still be missing from every list the assistant reads. That is why what is said about you deserves as much care as what you say.
What your pages say
Whether an AI can find the content, trust each section, see what the page answers, and tell it apart from the competition.
Scored on every crawlWhat everyone else says
What independent sources say about you, where competitors are named and you are not, and what to do about it.
Researched every monthNot a trick for AI. The basics of being worth reading.
Every check is founded in E-E-A-T: the experience, expertise, authority and trust that Google asks its own quality raters to look for. A specific, complete, current page, and a brand that independent sources speak well of, is what every reader looks for, whether it is a person, a search engine or an AI assistant. So this is a long-term investment in everything that is read about your brand, whatever reads it next.
What only you have, set against the pages that rank beside yours.
Sections that are specific and accurate, and pages that cover what they set out to answer.
What independent sources say about you, counted by voice, and where you are missing.
Answers to the questions people really ask, on pages kept as current as they expect.
You can’t retrain the model. You can fix what it reads.
There is little evidence that tricks aimed at a model change what it believes. What a business does control is the evidence an AI reads: whether the right page is found, whether the answer can be lifted out without guesswork, whether it is current, and whether others confirm it. That is what these checks measure.
Your pages, read the way an editor would.
From the smallest part of a page to the page as a whole, then against the pages it competes with. Scored on every crawl.
First, find the page inside the page.
- 01It breaks the page into its parts: menus, banners, footers and the content itself.
- 02The clutter is set aside. Nothing is rewritten, so what stays is your exact text.
- 03Checked against a person on 90 sites: 92% of the clutter dropped, 92% of the content kept.
Which sections earn their place?
- 01Every section is judged alongside the rest of the page, never on its own.
- 02Five tests: is it specific, complete, accurate and clear, and does it have a job?
- 03Each section gets a grade, and the weak ones get a fix. On 150 sections a person graded, it called 48% essential. They said 52%.
Does the page answer what it sets out to?
- 01It works out what the page is trying to be the answer to, and for whom. A product page is not judged like an article.
- 02It scores five things: breadth, depth, coherence, completeness and authority.
- 03It lists what is missing, so the fix is specific.
Is the page different from the ones that rank?
- 01It searches Google for the topic and your real competitors’ pages on it, and reads the closest ten, with where each one ranks.
- 02Each is compared with yours: what you both cover, what only you have, and what it has that you lack.
- 03All ten are scored together on one scale, with a suggestion for every gap.
Does it answer what people actually ask?
- 01It takes the searches that bring people to the page, from Search Console, plus your AI prompts.
- 02Each becomes a plain question, scored against the page out of ten and tagged fully answered, partially answered or not answered.
- 03A lost point has to name the missing fact.
Does Google’s AI cite you, or just name you?
- 01It asks AI Mode and AI Overviews the page’s own question, with your brand taken out, so a mention has to be earned.
- 02Named: the answer says your brand. Cited: it links to your site.
- 03Every page lands in one of four: named and cited, named only, cited only, or absent.
Update this page now, or leave it alone?
- 01It finds the page’s real date, from its markup, its headers and search data.
- 02Google Trends and the news say whether the topic is rising. Query deserves freshness (QDF) scores how fresh people expect each search’s answer to be.
- 03Both go into one answer: a priority, from critical to skip, with the reason.
The half that is not on your site.
AI answers draw on reviews, forums, lists and news as much as on your pages. This research runs once a month, whatever your crawl schedule.
What the web says about you, counted.
- 01It reads reviews, forums, news, video and app stores. Never your own site.
- 02Every quote is checked against its page. One that is not there is dropped.
- 03Opinions are counted by independent voices, not by pages.
Where competitors win off your site.
- 01It finds the lists, reviews and comparisons where your competitors appear.
- 02It checks whether you are already there before suggesting anything.
- 03Each gap becomes an action on someone else’s site, with an owner and a done-when.
Did the work land?
- 01When an action is due, it goes back to check.
- 02It searches for the page the action was meant to earn.
- 03It counts as landed only with a link to prove it.
Your pages, and everything said about you.
The whole method is on this page, and you can run it by hand on a few pages. What Lumar adds is scale: every page of every crawl, and the web around you every month.