The content strategy framework that survives contact with a real business

Why most content strategies die in month three

There is a particular kind of document that gets produced at the start of a content engagement. It is eighty pages long. It has a persona section with stock photography, a competitor matrix, a colour-coded calendar stretching four quarters into the future, and a slide near the end with the words thought leadership on it.

It is presented. Everybody nods. It is saved to a shared drive. And by the middle of the second quarter, nobody has opened it in six weeks, because a product launch moved, two writers left, and the head of demand generation needs three landing pages by Friday.

This is not a discipline problem. It is a design problem. Strategies that die share a structural flaw: they describe a destination in detail and say almost nothing about the mechanics of getting there when conditions change. They assume a business that stands still.

A content strategy is not a plan for what you will publish. It is a set of decisions that make future publishing decisions faster.The working definition we use internally

What follows is the seven-part structure we build for clients. It is deliberately shorter than the eighty-page version. The test we apply to every section: could a new writer, on their first day, use this to make a judgement call without asking anyone? If not, the section is decoration.

Part one: decide what you are actually claiming

Before any topic research, before a single keyword, you need one sentence that a competitor could not honestly say about themselves. Not a mission statement — a claim.

Most companies cannot produce this on request. What they produce instead is a category description: “we are a leading provider of workforce management software.” Every competitor in that category could sign their name to it. It gives a writer nothing to work with.

The exercise we run takes about ninety minutes with three or four people who know the product and at least one who talks to customers daily. Three questions:

1. What do we believe about this problem that most of our market does not? This is the seed of every strong opinion piece you will ever publish.
2. What can we prove that others can only assert? Proprietary data, unusual scale, an unusual customer base, an operational detail nobody else has. This determines what research you can credibly publish.
3. Who are we deliberately not for? The answer sharpens everything downstream. Content that tries to be useful to everyone reads as useful to nobody.

Write the answers into a single paragraph. That paragraph goes at the top of every brief you ever write. If a piece of content does not ladder up to it, it is somebody else's content.

Part two: gather buyer evidence, not opinions

Personas assembled in a workshop are a record of what a marketing team assumes. Personas assembled from evidence are a record of what buyers actually said. The two rarely match.

Four evidence sources, in rough order of value per hour invested:

1. Sales call recordings

If your company records calls, you are sitting on the single richest source of content input in the business and almost certainly not using it. Listen to twenty calls: ten that closed, ten that did not. Log every question asked by the prospect, verbatim. Do not paraphrase — the exact wording matters, because that wording is what people type into search engines and into AI assistants.

2. Support tickets and onboarding notes

Where do new customers get stuck in the first thirty days? Those friction points are the best mid-funnel content in the business, and they have the useful property of being provably real.

3. Ten customer conversations

Thirty minutes each, no pitch. The two questions that consistently pay for the whole exercise: “What were you doing about this before us?” and “What nearly stopped you from buying?” The second one is where your objection-handling content comes from.

4. Search and community language

Query data, community threads, review sites, and the comment sections of competitor content. You are looking for vocabulary mismatches — the gap between what your company calls something and what buyers call it. Publish in the buyer's vocabulary; define your own terms inside the piece.

Part three: pick territory you can own

The instinct is to map every topic your buyers care about. The result is a plan spread so thin that no individual area ever reaches enough depth to rank, be remembered, or be cited.

Pick three to five territories. That is the whole list. Each one gets a plain-language definition, a reason you have the right to speak on it, and a rough sense of the demand behind it.

A territory is not a keyword. It is a body of related questions that a single team could plausibly become the best public source on within twelve to eighteen months. Under each, you build a spine: one substantial cornerstone piece, six to ten supporting pieces answering the specific questions underneath it, and a steady trickle of shorter reactive work.

The prioritisation matrix is unglamorous and works. Score each candidate territory one to five on:

• Commercial proximity — how close is somebody researching this to needing what you sell?
• Right to speak — do you have data, practitioners, or unusual experience here?
• Competitive slack — is the existing published material genuinely good, or merely present?
• Sustainable supply — can you produce credible material here every month for a year without burning out an expert?

That last criterion is the one teams skip, and it is the one that most often kills a territory. An area where only your overloaded CTO can speak credibly is not a territory. It is a bottleneck with a content plan attached.

Part four: choose formats by job, not by fashion

Format decisions get made for the wrong reasons — because a competitor launched a podcast, because somebody read that short video is up, because a tool the company already pays for makes interactive quizzes.

Start from the job instead. There are really only five jobs content does, and each has formats that suit it:

Two rules we apply almost without exception. First, every territory needs at least one piece doing the “remove friction” job — these are consistently the highest-converting pages on any site and consistently the last ones anybody writes. Second, do not launch a format you cannot sustain for six months. A podcast with nine episodes and an eighteen-month gap actively damages credibility.

Part five: build a calendar with slack in it

A calendar planned to one hundred per cent of capacity is a calendar that breaks the first time somebody is ill. Plan to roughly seventy per cent and hold the rest for two things: reactive work, and rework on pieces that are nearly good.

The rhythm that has held up best across our clients, adapted to team size:

• One cornerstone per quarter per territory. The substantial piece. Three to six weeks of elapsed time, involving a subject expert.
• Two to four supporting pieces per month. Focused, answering one question well.
• One update cycle per month. Revisiting existing work that is decaying. Almost always cheaper per unit of traffic than new work.
• Reactive capacity, unallocated. Roughly thirty per cent of the total.

Note the update cycle. Teams under-invest here badly. A piece that ranked well eighteen months ago and has slipped to position nine will often return to the top three after a serious rewrite — for a fraction of the cost of a new piece and with none of the ramp-up delay.

Part six: governance is the strategy

This is the section that gets three lines in most strategy documents and determines whether any of the other six sections happen.

Governance means answering, in writing, before work starts:

1. Who approves a topic? One named person. Not a committee.
2. Who is the subject expert for each territory, and how much of their time is committed? Get an actual number of hours per month and get their manager to agree to it. This single step prevents more programme failures than any other.
3. What is the review path? Ideally one subject review and one editorial review. Legal only where legally required. Every additional reviewer adds delay and removes voice.
4. What is the maximum review turnaround? Three working days, and after that the piece proceeds. Without a default, review becomes an indefinite hold.
5. Who can kill a piece, and on what grounds? Being able to stop bad work is as important as being able to start good work.

Write these on one page. Circulate it. When a programme starts slipping — and it will — this page is what you return to, because the cause is almost always here rather than in the creative work.

Part seven: decide how you will be wrong

The measurement section should be written before publishing starts, and its most important content is not the list of metrics. It is the list of decisions those metrics will trigger.

For each territory, write down: if this is working at six months, we will see X; if it is not, we will see Y, and we will then do Z. Committing to Z in advance is what stops a programme drifting for two years on the strength of vanity numbers.

Sensible six-month checkpoints for a new territory:

• Cornerstone piece indexed and ranking in the top twenty for its primary theme
• At least three supporting pieces bringing consistent non-branded entrances
• Measurable assisted contribution to at least a handful of pipeline conversations
• At least one external citation, mention or inbound link earned rather than bought

Miss all four and the honest conclusion is usually that the territory was wrong, not that the writing was. Territory errors are cheap to fix at six months and expensive to fix at two years.

What the first ninety days look like

Compressed, this is the sequence we run:

Thirteen weeks, one cornerstone, a running calendar and a governance page that people actually follow. That is a working strategy. The eighty-page version can wait — or, better, never get written at all.