What this playbook does
Pages Google wants is a list of pages to add to your site, taken from searches Google already connects with you. When someone searches for a service you offer and Google can only show your homepage or a loosely related page, that search is asking for a page you haven't written. The agent finds those searches and proposes a page for each topic.
Use it when you know you should add pages but aren't sure which, or before you pay for a list of blog ideas. Each suggestion starts from demand Google has already seen on your own site.
How the agent works through it
- Reads the last 90 days of Search Console: every search your site appeared for and the page Google showed for it. Searches for your own brand name are left out, as are searches with fewer than 50 impressions and those where your best page already sits in the top three.
- Flags a search when the page Google shows covers less than half its meaningful words (judged from that page's title, main heading and address), or when Google spreads the impressions over three or more of your pages with none getting 60%. Related searches, including Hindi and Hinglish versions of the same need, are grouped into one topic.
- Leaves out topics whose page is in your Qelvra change log from the last 35 days (cards you marked done, posts published through Qelvra, changes you told the agent about): it hasn't had time to rank.
- For each topic that needs a new page, searches your sitemap to confirm nothing already covers it, then checks the live Google results for the main search to see which type of page wins (a guide, tool, template, product page or local page), and proposes an address, a title of up to 60 characters and a main heading.
- For each topic where you already have a suitable page that Google isn't choosing, it doesn't propose a new one. It suggests a sharper title for that page and two or three internal links to it from your other pages.
- Saves each page to write as a content card in your Action plan, with a rough monthly click estimate, and the fixes to existing pages as on-page cards. It then offers to write a brief for the top one.
On a paid plan or trial with Search Console connected, the weekly site check also does a shorter version of this on its own: it adds the strongest topic with no page to your Action plan as a "Write a page about …" card, at most 1 a week, with the searches, their impressions and the page Google shows today. It only adds the card; it doesn't write or publish anything. Run this playbook when you want the full list, the type of page that wins for each topic and the fixes to pages you already have.
What you get back
A list of topics, each with its evidence: the searches in it, their impressions and average position, and the page Google shows today. New pages come with an address, title and heading; existing pages with a better title and the pages that should link to them.
An example of the shape of the answer, for a made-up site (not real data), a physiotherapy clinic in Hyderabad:
- Topic: frozen shoulder treatment (new page)
- Searches: "frozen shoulder physiotherapy hyderabad", "frozen shoulder physiotherapy near me", "physiotherapy for frozen shoulder"
- 1,900 impressions in 90 days, average position 18. Google shows the homepage.
- Proposed: /treatments/frozen-shoulder, titled "Frozen Shoulder Physiotherapy in Hyderabad". Rough estimate: 25 clicks a month.
What it needs
- Your site in Qelvra, entered by its address.
- Search Console, connected for that site. This playbook is built on your own search data, so it needs it. Without it, the agent says so, gives you the one step to connect Google, and offers a fallback instead: keywords your competitors rank for that you don't, clearly labelled as competitor data, not yours.
- A sitemap Qelvra can read. Without one, the agent can't be sure you lack a page, so it checks your site before recommending one.
Where it stops
- It only finds searches Google already shows you for. For topics where you have no visibility yet, run keyword research.
- Pages are judged on their title, main heading and address, not every word of the body, and only a limited number of pages are read in full (up to 25 in each of two passes); the rest are judged by their address alone.
- The click estimate is rough. It assumes the new page reaches position 6 and gets a typical click rate there.
- On very large sites, only the top 25,000 search-and-page rows are read, so the smallest searches can be missed.
- A small or new site may have too few impressions for any topic to qualify; the agent then points to pages you can improve.
Frequently asked questions
Why does Google show my homepage for searches about one service?
Usually because no other page on your site is clearly about that service, so Google falls back to the page it trusts most. A dedicated page tends to answer the search better.
Will it suggest a page I already have?
It checks your sitemap and page titles first. If a page on the topic exists but Google isn't picking it, the advice is to improve that page, not write a duplicate.
I published a page on this topic last week. Why isn't it listed?
If the page is in your Qelvra change log from the last 35 days, its topic is skipped on purpose: a new page needs a few weeks before Google's numbers say anything about it. A page you published elsewhere and never told the agent about is not skipped.
Can it write the page for me?
It offers a brief for the top topic; the content writing playbook turns a brief into a full draft.