- A substantial 1,200-word article takes four to eight hours end to end, so the available hours set the cadence, not the ambition.
- The best-performing client articles came from questions asked in sales calls, not from a keyword tool.
- Decision content earns enquiries, original data earns links, explainers earn credibility and fewer clicks than they used to.
- After year one, updating the top twenty pages usually beats publishing new ones.
- Content takes three to six months to settle on an established site, so a six-week verdict is a verdict on noise.
We have been handed a lot of content strategies. They run to forty pages, include a persona document, and propose two articles a week. By month four publishing has stopped, the client believes content does not work, and the agency has moved on.
The problem is almost never the idea. It is the sizing.
Start from hours, not ambitions
Ask one question: who is writing this, and how many hours a month do they genuinely have? Not hope to have. Have, in a normal month with a client emergency in it.
A good 1,200-word article with real substance takes between four and eight hours end to end: outline, draft, edit, internal links, publishing, promotion. If the honest answer is eight hours a month, the plan is one article a month. Not four.
One article a month sustained for two years is twenty-four assets. Four a month sustained for three months is twelve and then nothing, plus a client who now believes the channel is broken.
Take the topics from sales, not from a tool
The best-performing articles we have published for clients came from questions asked in sales calls. They convert because the reader is at the same point as the person who asked.
The method: sit with whoever answers enquiries and list the twenty questions they answer most. Then check search volume, not to choose but to prioritise among questions you already know matter. A question with 40 searches a month that every prospect asks is worth more than one with 4,000 that none of them do.
Keyword tools are good at telling you what exists. They are poor at telling you what your buyers need in order to decide.
Book an hour with whoever answers enquiries. Write down every question they are asked more than once a month. That list is your content plan for the next two years, in priority order.
Pick formats by what they can earn
Not all content earns the same things.
Explainers (what is X, how does Y work) earn visibility and, increasingly, citations in AI answers. They earn fewer clicks than they used to. Publish them for credibility and internal linking, not traffic.
Decision content (X vs Y, how much does X cost, what to ask before hiring) earns enquiries. It is the highest-value type for a service business and the least written, because it requires having opinions in public.
Original data, a survey of your own customers, results across your own projects, a benchmark you can publish, earns links. It is the only format that reliably does. One good data piece a year beats twelve explainers for authority.
Client work written up earns trust and closes deals. Ask permission, use real numbers, name the constraint you were working under.
Give every article a destination
An article with no commercial destination is a hobby. Each one links, in the body, to one service or product page, in a sentence the reader benefits from. Not a sidebar banner. Not six links in a footer block.
That single habit is the difference between a blog that supports the business and a blog that costs money.
Update before you write
After the first year, the highest-return content work is usually maintenance. An article already sitting at position six that gets a genuine update, new figures, a section answering a question the original missed, current examples, will typically outperform a new article on a fresh topic, because it starts with history and links.
We review the top twenty pages every six months and update three or four. That is often worth more than the new pieces published in the same period.
Measure on the right timescale
Content is the slowest channel there is. A new article on an established site takes three to six months to settle; on a young domain, longer. Judging at six weeks is judging noise.
Report monthly: clicks by page group, enquiries attributed to content-first sessions, and the position band of the target queries. Review at six months: whether the pieces published are reaching anyone, and which format earned most.
The plan on one page
A content strategy that survives fits on one page. Who writes. How many hours. One article a month, or whatever the hours permit. The twenty questions in priority order. One data piece a year. A six-monthly update cycle. The destination each piece links to.
Everything else, personas, tone documents, editorial calendars stretching eighteen months, is optional and mostly gets abandoned. We have never seen a client fail for lack of a persona document. We have seen plenty fail for publishing four times in month one and never again.
We plan, write and publish content for clients who do not have the hours, sized to a cadence we can actually sustain, with the internal linking done properly.
What we would cut first
Social-first content that does not live on your own site. Round-up posts. Anything written because a competitor has one. And any piece that could have been written without knowing anything about your business: those read like everyone else's, and increasingly like a machine's.
The month-four conversation
Every content programme has a difficult month, and it is usually the fourth. Nothing ranks yet, three articles are published, and someone senior asks what the money bought.
Have the answer ready before it is asked. Ours is a one-page note: what was published, what queries each piece is now visible for even at position 20, what the sales team has used it for, and what the six-month expectation was when we started. Content programmes rarely die of poor results. They die because nobody could describe progress at the point results did not exist yet.
When to bring someone in
Writing is the part clients most often keep, and they should, because the subject knowledge is in the building. Where an outside team earns its fee is the sizing, the question list, the update cycle and the internal linking, all the parts that quietly stop happening when everyone is busy. That is how our content marketing engagements are shaped, usually alongside SEO rather than instead of it. If you want the argument for publishing less, read what changed when AI answers arrived.
Questions
How often should a small business publish?
As often as it can sustain for two years. For most of our clients that is once or twice a month. A consistent monthly article outperforms a burst of weekly posts that stops in March.
How long should articles be?
As long as the question needs. Decision content is often 800 to 1,200 words; a genuine guide may run to 2,500. Padding to hit a word count is visible to readers and does not help rankings.
Should I use AI to write articles?
For outlines, research summaries and first-draft structure, it saves real time. For the substance, the opinion, the numbers, the client example, it cannot help. It does not know your business. Published straight, it produces the exact content that no longer earns anything.
How do I know if content marketing is working?
At six months: are the articles being found, are people reaching the service pages from them, and has anyone enquired mentioning something they read? If all three are no, the topics are wrong, not the channel.
Is it better to update old posts or write new ones?
After the first year, usually update. A page with history and links responds faster to improvement than a new URL does to publication. We aim for roughly one update for every two new pieces.