A content strategy useful to an SMB fits on one page: three to five broad subjects matching what your customers search for, one reference piece per subject surrounded by three or four shorter ones, and a sustainable cadence — two articles a month is enough, and beats eight for one quarter then nothing. Allow three to five hours per article for a piece that holds up. Everything else — illustrated personas, a sixty-cell calendar, a fifteen-page editorial charter — is work on the plan rather than work on the content.
This piece covers how to build that plan and keep it running. It assumes you already know why you would produce content; if not, our piece on content marketing deals with that question.
Start from the questions, not the subjects
Most editorial plans begin with a session listing what you would like to talk about. Wrong end. The right starting point is the list of questions your customers actually ask you — on the phone, during quoting, on site. They already come in their words, they have already proved somebody cares, and they overlap from one customer to the next.
What remains is checking those questions are also typed into a search engine, which is not automatic: you can have a great deal to say about something nobody looks for. The checking method is in our piece on choosing keywords. That step turns a list of wishes into a plan.
Organise in clusters, not in lists
A cluster is a reference piece covering a broad subject, surrounded by three or four shorter pieces each covering one precise aspect of it. The short pieces link to the reference piece, which links to the matching service page. The point is not theoretical: this structure concentrates authority on one page instead of scattering it across fifteen, and it stops two of your pieces from competing on the same search.
Three to five clusters are enough for an SMB, and they have to match what you sell. A cluster on a subject where you have nothing to sell attracts readers who will never become customers — pleasant for the statistics, irrelevant to revenue. And one rule that saves a lot of damage: fix the addresses of your future pieces from the start, because each piece links to its neighbours and you otherwise end up publishing links to pages that do not exist.
Two articles a month for two years beats twelve published in January and nothing after.
Cadence is the variable that really counts
Almost every company blog dies the same way: an enthusiastic start, six pieces in two months, then a busy season, then nothing for a year. The problem is not lack of motivation, it is a cadence chosen for good days instead of bad ones. Set it for the worst week of your year, not the best.
In practice, two pieces a month is sustainable for an SMB, and one beats none. Two useful guardrails: write in batches — three pieces in a row cost less than three spread out, because the research is already done — and always keep one piece in reserve, which absorbs the week when nothing goes to plan.
Write to answer, not to introduce
A piece has to answer the question asked within the first three lines, with a number or a clear position. No scene-setting, no in a world where everything is digital. Nuance comes afterwards. That rule serves two purposes at once: it stops the reader going back to the results page, and it makes the paragraph quotable by the engines that summarise the web.
The rest follows the same logic. One subheading per sub-question, phrased the way a reader would ask it. Length dictated by what it takes to answer, not by a word target. And one practical box per piece giving a concrete action: it is often the block assistants pick up, and the one readers remember. Being cited by those systems comes down to the same editorial habits as classic search.
Who writes
Three arrangements, and only one wrong answer. You write: the content is accurate and specific, but the cadence collapses the moment the season starts. An outside writer writes: the cadence holds, but the pieces ring hollow if nobody feeds them material. The arrangement that works best in an SMB is mixed — twenty minutes of your time on the phone to hand over what you know, the writing done by somebody else, and your review before publication.
The wrong answer is production handed entirely to a tool with no review by somebody who knows the trade. This is not a question of technology but of substance: those pieces contain no information the reader cannot find elsewhere, and that is precisely what rankings punish. What cannot be automated is what you know and your competitors do not say.
What to measure, and from when
Do not judge a cluster before six months: before that, you are mostly measuring your own patience. The two indicators that count are positions per cluster in Search Console and click-through rate on your titles. A piece well positioned but rarely clicked is a title to rewrite, not content to throw away — the cheapest fix there is, and the most often forgotten.
What measures nothing: the number of articles published, total traffic, time on page taken in isolation. The final verdict remains the number of requests attributable to content, and it arrives late. The five numbers a business owner should track apply here too, and the pillar on costs places this line item against the others.
If you would rather we built this plan with you than start from a blank page, the work is part of our SEO and GEO offer, and the thirty-minute call exists to check whether content really is your next job — sometimes it is not.
Defining your ideal customer: the eight-line sheet — the document the whole editorial calendar depends on.