Your first fifty posts should not come from a keyword tool

A keyword tool can only show you questions that enough people have already typed, and in most B2B markets those questions already have an owner. For a young archive, that is the wrong place to start. The first fifty posts should answer what your buyers ask your salespeople, most of which no keyword tool will list, and the tool earns its place later, as a check on vocabulary rather than a source of subjects.

A keyword tool can only count questions that have already been asked

Every volume figure is a report on the past: queries that were typed often enough, for long enough, to be counted. That is useful, and it is also a hard limit. Google's own Keyword Planner documentation says that keywords with very low search volumes aren't discoverable or forecastable. In a narrow B2B market, that covers most of the questions that actually precede a purchase.

It also misses what is new. When Google introduced BERT in 2019, Pandu Nayak wrote that 15 percent of the queries Google sees each day are ones it has never seen before. A question first asked this morning has no volume to report. The buyer who types a long, specific, slightly awkward question about your exact category is, by construction, invisible to the tool you were going to plan the year with.

The demand it does show you already has an owner

The queries that do appear with healthy numbers are the ones established competitors have been answering for years. We have argued that when worse writing outranks you, the reasons are usually links, age and coverage rather than prose. A domain with none of those, entering a head term with ten years of incumbents, is not competing. It is queueing.

So a keyword-led first year fails in a particular way. You publish forty competent posts on the terms with the biggest numbers, each one lands on page three, and the conclusion drawn at month eight is that the blog does not work. The blog was fine. It was aimed at the one part of the market where a newcomer has no chance yet.

There is a second cost, and it is the one Google names. Its guidance on helpful, people-first content asks whether content is made primarily to attract visits from search engines. A backlog assembled from a volume column answers yes, whoever or whatever writes it, which is the whole case in our cornerstone on AI content: the problem is the page nobody needed, not the author.

Your sales calls hear the questions before search does

The people who hear your market's real questions first are the ones on your calls. The objection that stalls a deal, the comparison every buyer makes unprompted, the integration question that comes up on every second demo. None of those is likely to show a volume, and all of them are asked by someone with a budget.

This is where the earlier half of the argument pays off. A question heard on your calls is one you can answer better than anyone, because you have answered it out loud a dozen times, and it is one your salesperson will send to the next buyer whatever Google does with it. A post that works as a sales document on day one and a search result in month fourteen beats a post that only works if it ranks.

Worth conceding: a question one buyer asked once can be an oddity. The discipline is to write about questions you have heard at least twice, from different people. That filter is cruder than a volume number, and much better aimed.

What keyword research is still good for

Keyword tools are right about one thing that matters a great deal: the words people use. If your buyers search for "customer portal" and your product team says "client workspace", a post titled in your internal vocabulary will not be found by the people it was written for. Checking the phrasing of a subject you have already chosen is a good use of a keyword tool. Choosing the subject from it is not.

The better source arrives once you have published. Search Console's Performance report groups your traffic by the queries people actually typed, and after a few months of posts it will show you questions you did not know you were answering. That is keyword research built from your own archive, which is the only kind a newcomer can win.

Fifty is not a magic number. It is roughly a year at one post a week, and about the point where an archive has earned some links and some history of its own. After that, targeting a volume figure stops being queueing and starts being a fair fight.

Start with the questions nobody has typed yet

If you accept this, the first brief looks different. It opens with a list of the questions your salespeople answered last month, not a spreadsheet of terms. The archive grows narrower and deeper than a traffic target would make it, and each post is useful to a buyer before it is findable by one.

The keyword tool is still on the desk. It is just no longer the one handing out the assignments.