Which Google searches are likely to bring customers?

Most link budgets are allocated by volume: find the biggest keyword a page could rank for, then buy links until it does. It feels rational, because volume is the number every SEO tool puts first. But volume measures how many people search, not how many of them will ever pay you. When you spend on links, only the second number matters.

This guide shows how to check whether a search is worth link money, with a worked example where the biggest search on the page gets nothing.

Volume and value are different numbers

Every search has two properties that decide what ranking for it is worth:

• Demand: how many people search it each month.
• Intent: what they want when they do. Some are ready to buy or comparing options. Others want to learn something and will leave once they have.

A click from someone learning can be worth a tenth of a click from someone buying, or less. That's why the same keyword can look huge in a keyword tool and be nearly worthless on a revenue report.

Links are the most expensive way to move a ranking. So the question for a link budget isn't "which search is biggest?" It is "which search, if we moved up, would pay back what the links cost?"

Four kinds of search

A simple, transparent way to sort searches is by the words in them. It isn't perfect, but it's right most of the time, and you can see why it decided what it did.

Two rules make this work better:

1. Buying and comparing words beat question words. "How much does invoicing software cost" starts like a question, but the person is pricing a purchase.
2. A bare category phrase is usually commercial. Someone searching "invoicing software" is shopping, even with no buying word in the query.

When you have measured intent from a data provider, use it, and fall back to the word rules only when you don't.

A worked example

Take an invoicing software company whose /invoicing page shows up in Search Console for these searches. The figures are illustrative but typical of this market.

Sorted by volume, the plan writes itself: "invoice template" has eleven times the demand of anything else, so buy links until the page ranks for it.

Now put a value on each click. Say 4% of people comparing invoicing software go on to become customers worth $600 in their first year: a click from a comparing or buying search is worth about $24. People downloading a template rarely buy software; if one in 250 does, a template click is worth about $2.

Then estimate what the realistic next step is worth, and what it costs:

The search with eleven times the volume earns less, and costs five times as much to win. Even after discounting the second column by half for uncertainty, it pays back in under ten months. The first never makes sense.

The rules we use

Before a page gets any link budget, its main search has to pass three checks:

1. It's a buying or comparing search. Pages that mainly win learning searches still get better content, never link money.
2. It has real demand. Below about 100 searches a month, even a big ranking gain is too few clicks to repay links. The content can still be improved.
3. The page has room to climb. A page already in the top three with nobody above worth chasing is protected, not paid for.

Then the payback test decides: if the extra revenue from the next realistic step up, discounted for uncertainty, doesn't repay the links within about a year, don't buy them. The guide to backlink purchases that can't pay for themselves goes through that calculation.

What to do with the big learning search

Declining to buy links for it doesn't mean ignoring it.

• Improve the content. A clear, direct answer near the top of the page can win clicks and AI citations without any links. This is cheap and it compounds.
• Link from it. A page that ranks for a learning search can pass readers, and some authority, to the pages that sell. An internal link from the explainer to the product page is free.
• Split the page if it serves both. If one page is trying to be a glossary entry and a product page, it will usually do both badly. A separate explainer that links to the product page often lifts both.

Where this goes wrong

• Word rules misread some searches. "Apply for a corporate card" is clearly buying, but "corporate card for dentists" has no buying word and is still commercial. Check the searches that carry the most money by hand.
• Some learning searches do convert. In some markets, such as B2B software with long sales cycles, readers of a good explainer become customers months later. If your analytics show that, weight those searches higher. Don't assume it.
• Volume estimates are estimates. Keyword tools disagree, sometimes by a factor of two. Use Search Console impressions for the searches you already show up for; they are measured, not estimated.

The key points

Rank searches by what a click is worth multiplied by how many clicks you could realistically gain, not by volume. Buy links only for buying searches with real demand, on pages with room to climb, when the payback is under a year. Give everything else better content and internal links.