Content Marketing Strategy

Anatomy of a strategy document: parts, contents and readers

A strategy is a set of working documents used by whoever writes, whoever approves and whoever is accountable for results: a topic map with priorities, a calendar, briefs, an update plan and KPI definitions. A slide deck of general recommendations isn't a strategy, because nobody knows what to write on Monday.

I rarely write briefs for the whole year upfront. After the first quarter, Search Console data and changes in the results almost always reshape part of the plan, so detailed briefs cover the current and next quarter, while later topics have a scope and a target query. What goes into a brief is described in the content marketing guide.

Inputs: what I collect before choosing the first topic

The best topics come from data your competitors can't see: your Search Console, your CRM and your customers' questions. Keyword tools complete the picture but don't replace it.

• Search Console: queries and pages from the last 16 months (the history the Performance report keeps), to see seasonality and trend. I also use the branded and non-branded query filter Google introduced in November 2025, and Query groups in Search Console Insights, which bundle variants of the same question (available for properties with high query volume).
• GA4 and CRM: which landing pages lead to key events and leads, and which draw traffic with no further action.
• Sales and support: questions from emails, calls and tickets from recent months. I ask for the raw messages, not a summary, because the customer's own words are better material than a salesperson's interpretation.
• SEO tools: demand, difficulty and gaps against competitors (Ahrefs, Semrush; Senuto for the Polish market).
• Questions put to AI assistants: test conversations with ChatGPT, Gemini and Perplexity on topics from the offer, plus Google's People also ask. They show questions with context that tools don't capture.
• The company's expert assets: who can give an interview, what data the company collects, which implementations or tests can be described. Without these there's no first-hand element, and without one a piece won't stand out in the results.

Inventorying existing content and spotting cannibalization

Before planning new pieces, I map every existing URL to a topic and check whether several pages compete for the same queries. For companies with a blog that's a few years old, the first quarters of a strategy are often more about updating and merging than publishing.

Columns in the inventory sheet

• URL, title, publication date and date of the last significant update.
• Clicks, impressions and average position over 16 months, plus year-over-year change.
• Key events and leads from GA4 or CRM attributed to the landing page.
• Main query and topic from the topic database.
• Incoming internal links, and whether the page links to the offer.
• Decision: keep, update, merge, redirect, remove. My criteria are in Content Decay – When to Update, Merge, or Delete Content.

How I spot cannibalization

1. I export query–page pairs from Search Console and look for queries where more than one URL from the site gets impressions.
2. I discard natural cases: a service page and a guide can both show for the same query if they serve different intents.
3. What's left are pairs where two similar articles swap positions over time, or neither climbs. Those are merge candidates: I keep the stronger URL, move the valuable content from the other one into it and set a 301 redirect.
4. I check every new topic in the database the same way before it reaches the calendar. The cheapest cannibalization is the one never published.

Grouping queries into topics: by search results, not word similarity

Two queries belong in one piece when Google shows largely the same pages for both. Word similarity can mislead: "content marketing strategy" and "content marketing examples" sound close but may need completely different formats.

For every query in the database I pull the top ten results and compare them in pairs. My working rule as a starting point: if several of the same URLs appear in both result sets (usually 3–4 out of 10), one page serves both queries. With less overlap I check the results by hand, because the nuances decide: does Google show guides, product pages, forums or video? I automate the grouping with a script, and on large databases part of the work is done by a language model whose output I always verify on a sample. I describe that process in SEO Automation with AI Agents.

What I get from grouping

• A list of topics, each with a main query, variants and side questions.
• Intent and the format that wins in the results (guide, comparison, service page, category).
• Whether the query triggers an AI Overview, video, forum results or a local pack.
• A target URL: an existing page to expand or a new piece.

Questions from AI conversations go into topics as side questions, even with no volume in tools. According to Google's documentation, AI features in Search may break a question into many related searches (query fan-out), so a piece that answers them has more chances to be cited. More in Keyword Research in the AI Era.

How I prioritize topics: scoring on four criteria

I set the order of topics with a score in which demand is only one of four criteria. The others are the link to the offer, a realistic chance of visibility, and whether the company has something of its own to say. That way the start of the calendar holds pieces that can bring enquiries, not just traffic.

I weigh the total against effort: updating a page that already has impressions just below the top five is usually cheaper than a new piece with the same score. I adjust the weights to the situation. For a new domain without authority, chance of visibility weighs more; for a B2B company with a long sales cycle, the link to the offer does.

Competitor analysis: the topic gap and the information gap

The topic gap is the set of topics where competitors are visible and you aren't. The information gap is what's missing from the pieces already ranking. Tools show the first; you find the second by reading the results.

I collect the topic gap with a competitor analysis tool, but only count competitors who sell the same thing, not every publisher visible for those queries. I check the information gap by hand for the highest-scoring topics:

• Are the results current? Pieces from a few years ago on a topic that has changed (tools, regulations, prices) are the easiest opening.
• Does anyone give concrete numbers, price ranges, steps or selection criteria, or does everyone stop at "it depends"?
• Is there a first-hand element: a test, a screenshot, an implementation example, a practitioner's opinion?
• Which side questions go unanswered in the top ten but appear in People also ask and in AI conversations?
• Who gets cited in AI Overviews and assistant answers? If it's always the same site, I look at how its passage differs from the rest.
• Are there forums, Reddit or video in the results? That signals people want experience, not a definition.

How AI models pick sources and how to write passages that can be quoted is covered in How to Write Content AI Cites. In a strategy, that knowledge becomes requirements in the brief, not a separate content type.

From article to enquiry: mapping content to decision stages

Every topic in the strategy has a customer decision stage, the next step it leads to, and a metric that tells me whether it works. Without that, educational articles collect traffic but nobody moves on from them.

In the strategy table every topic has a specific target URL. A cluster that doesn't end on any offer page I label as brand building and measure with different metrics, instead of pretending it will bring leads.

How I build a 12-month publishing calendar

I start the calendar with quick wins, updating pages that already have impressions, and then complete the highest-scoring clusters one by one instead of starting all of them at once. Half of five clusters usually delivers less than one complete cluster.

1. First weeks: updates and merges. Pages with impressions but weak positions, outdated pieces with a traffic history, and the merge pairs from the inventory.
2. The first cluster. The pillar page and supporting articles for the highest-scoring service. The order inside a cluster depends on the situation: on a new domain I start with narrow supporting articles, because they gain visibility faster, and publish the pillar once it has something to link to.
3. Seasonality. Seasonal topics are planned ahead, so the piece has time to get indexed and settle before the peak. I read the peaks from 16 months of Search Console data and from Google Trends.
4. Update slots. Every month I reserve part of the capacity for refreshing earlier pieces, because maintaining them is part of the strategy, not an extra.
5. Buffer. I leave room for topics nobody can predict: Google changes, new regulations, new tools in the industry.
6. Quarterly review. After each quarter I move topics based on the data and write briefs for the next quarter.

Publishing pace follows the team's capacity and the number of clusters, not an assumption that you must publish weekly. If I take over the writing, it runs as SEO copywriting or blog management.

Distribution and repurposing

A strategy also covers where a piece goes after it's published. In 2026 a brand's visibility in AI answers also depends on mentions outside your own site, so one data-backed article should feed several channels.

• Newsletter: the cheapest way to bring readers back, independent of Google changes.
• The expert's LinkedIn: short versions of the article's points, posted by the person who wrote it or is the source of the knowledge. More in Expert Personal Brand in the AI Era.
• Video and YouTube: for topics where results show video, and as an extra source for AI models. I write about it in YouTube and Video in AI Visibility Strategy.
• Digital PR: pieces with original data or research go to industry media, since that's what most often earns links and mentions. See Digital PR and Brand Mentions.
• Repurposing: one piece becomes posts, short videos and newsletter snippets. The AI-assisted process is in Social Media Automation with AI.

For each cluster the strategy records which channels are mandatory and which optional. I don't add channels just in case; only where the customers are and where the company has the resources to publish regularly.

Who does what: roles, approval and rules for using AI

A strategy works only when it's clear who writes, who checks the substance, who approves and who publishes. The most common reason a plan stalls after two months is that nobody on the company side approves pieces within the agreed time.

The working rules also say how the team uses AI: yes for research, structure and editing; no for publishing text without fact-checking and without the expert's input. Google judges content by quality, not by how it was produced, but mass-publishing pages without value violates its scaled content abuse policy. Where the lines are is covered in AI Content and Google Policy.

Content strategy KPIs and quarterly reviews

I measure the strategy per cluster, not per article, and separate leading indicators (impressions, positions, citations) from business metrics (leads, sales, branded searches). Leading indicators show whether a cluster is heading the right way before conversions appear.

Decisions at the quarterly review

• Grow the cluster when leading indicators rise and the first pieces send traffic to offer pages.
• Fix the cluster when there are impressions but weak positions or no onward clicks: usually linking, a first-hand element or intent match is missing.
• Pause the cluster when, after a few quarters, complete pieces still get no impressions. Then I check whether domain authority is the problem and consider earning mentions instead of publishing more.

I measure visibility in AI answers separately, because AI traffic is only part of the effect. The method is described under the AI visibility audit, and measuring assistant traffic in How to Measure Traffic from ChatGPT, Perplexity and AI Mode. The wider context of falling clicks is in Zero-Click and Declining Organic Traffic.

Strategy in special situations

The same method works for any company, but criteria weights, pace and order change with the starting point.

• A new domain with no history. I start with narrow topics that have a clear gap and a strong first-hand element, not broad queries held by big publishers. In parallel I plan mentions and links, because content alone rarely builds authority. My approach to links is described under link building strategy.
• YMYL industries (finance, health, law). Every piece has a qualified reviewer, a visible author and sources. That stretches the calendar, so I plan fewer pieces with better documentation. The principles are in E-E-A-T in 2026.
• B2B with a long sales cycle. More comparisons, calculators and case studies, less general content. Leads arrive with a delay, so reviews focus on intermediate metrics.
• Several languages or markets. Each market gets its own topic database, because demand and competition differ. The offer map and working rules are shared.
• An online store. The strategy covers category descriptions and buying guides tied to categories, not just the blog.