pSEO ideas

Programmatic SEO is one page template filled from a dataset of real things: one page per place, product, company or account. Enter a seed keyword and a country, and the pSEO idea finder loads real search data, finds the sets of real things people search for in the same way, such as plumber {place}, and ranks each one as a set of pages you could build: how many pages, how many searches, what a click is worth, how hard it is and who ranks now. Question phrasings and short lists of ordinary words are not pSEO, so they are never an idea.

Add your site to see which ideas you already have: your pages that match are checked live, and only a page that answers and can be indexed is shown. Your own pages never count as a site to beat. Or start from your own Search Console searches instead of a seed.

Sign in to run it. Your first seed each month is free, then 1 credit ($1) buys 2 seeds, up to 4 seeds a run. Surprise me is always free. A run on your own Search Console searches costs 1 credit. A run that fails gives its credits back, and I’ll email you when the report is ready (you can turn that off in your account settings).

What you get

• Up to 30 page patterns, best first, each with real example searches.
• For each pattern: pages found, monthly searches, the median and top pages, how much of the demand is in the long tail, cost per click, keyword difficulty, main intent and the trend over the last year.
• For place patterns: how many of the country’s places the searches already cover, and an estimate of the searches the places not found would get.
• Who ranks now for the top patterns, from 5 sample searches each: forums, thin listings, low Domain Rating sites and titles that do not match the search, and the site to beat.
• A plan for each pattern: the URL pattern, the page sections in order, where the data comes from, when to publish a page, and how it can make money.
• With your site: your pages like each idea, checked live, and which ideas you have already built.
• For each idea: the dataset (one page per what, and how many found), the page template, an index rule, the internal links and the evidence.
• The evidence for every idea: where each number comes from and its date, the results pages checked and when.
• CSV and JSON downloads, and Copy for agent to plan the build with your AI agent.

How it works

• It loads up to 1,600 keywords for each seed from DataForSEO: suggestions, related searches and wider ideas, with volumes, cost per click, difficulty and intent.
• It finds place names with a list of every town and city with 1,000 or more people in the country, from GeoNames. Years, numbers and postcodes are found too.
• Handles, web addresses, codes, numbers and place names are marked first. Typos join the word they misspell.
• A part of a search that changes while the words around it stay the same, in any order, is a slot. Frames that share their fillers are one slot: {company} revenue and how many employees does {company} have are two sections of one page per company.
• A slot counts only with 20 or more different fillers, a long tail (no filler with 30% of the searches, most fillers searched only a few times) and fewer than 40% common words. Under 50 fillers it is shown as information, marked Low confidence. The same searches always give the same slots.
• Jev, an AI model, can only remove a slot, never add one. It checks that the fillers are one kind of named thing (people, places, organisations, products, titles, accounts, codes or dates) with a list you can get, checks the biggest fillers one by one, and checks place names that are also ordinary words. Its answers are kept, so the same slot always gets the same answer.
• For the top 3 patterns it checks 5 real results pages each, in a fixed order from the biggest page to the smallest, and looks up each ranking site’s Domain Rating.

How ideas are ranked

Score = log10(monthly searches + 1) × money × results × data ÷ difficulty.

• Money: 1 + cost per click ÷ 10, up to $20 a click.
• Results: 0.5 + how weak the pages that rank now are (0 to 1). Forums, user posts and sites with a Domain Rating under 30 count fully, thin listings mostly and titles that do not match the search by half.
• Data: 1 when a full list of the things in the slot is public, 0.7 when it has to be collected, 0.4 when there is no list.
• Difficulty: 1 + the median keyword difficulty ÷ 50.
• Marked and ranked lower: searches down 20% or more in 6 months (× 0.7), searches for one site rather than a type of page (× 0.5), under 200 searches a month in all (low confidence, × 0.5), and patterns whose results pages were not checked (results count as 0.75, below any checked pattern with weak results).

Search volumes are rounded by Google and small countries have thin data, so treat the numbers, and the estimates most of all, as a guide.

Pricing and privacy

• One seed free each month. Then 1 credit ($1) for 2 seeds, rounded up. Credits are charged when the run starts and given back if it fails.
• The same price whether the data is new or was loaded for anyone in the last 30 days. The report says which month the search data is from.
• A run on your own Search Console searches costs 1 credit. It reads the last 28 days of your searches, then looks up your 3 biggest search patterns in DataForSEO’s keyword data for your country, so it finds datasets your own searches are too few to show. Each idea says whether it came from your searches or the keyword data. For the biggest dataset it finds, it also looks for the facts people search about its records, such as {account} follower count or {place} population, and keeps the ones on your site’s topic. A fact counts when people search it for several different records: most of those searches are too small to show in keyword data, so the report says how many records it was seen for and gives the clicks it has seen as a minimum. For the top ideas it checks who ranks for 3 of their searches and what the searchers want. When most searchers want one site, such as the platform itself, the idea is information, not a task.
• Your site, when you add it, is read from its sitemap with ianbot, which honours robots.txt.
• Your seeds are kept only with your run’s report, which only you can open, until you delete your account. They are never logged, shown to anyone else or used to pick anyone’s Surprise me.