A content marketing plan is the document that sets out which content you publish, which questions it answers, which business goal it serves and what yardstick you judge it by. The publishing calendar is not the plan: it is the last page of the plan, and the easiest one to fill in.
Between 1 May and 20 August 2026 the Visilay website collected 53,723 impressions on Google and 106 clicks. Seventy-one pages out of ninety-six did not get a single click in three and a half months. These are our own numbers, taken from Search Console, and they sit here at the top because they show clearly what a plan built on article counts, rather than on the questions the articles answer, produces.
Below is the method we use to redo that plan, the data from two real sites and the points where we got it wrong first.
What a content marketing plan is and what it is not
In practice three different things get called by the same name, and the confusion costs months.
The content marketing plan decides what you want to achieve and who you are writing for. It contains the commercial goal, the people you need to reach, the topics you cover and the metrics.
The editorial plan turns those topics into a list of pieces: title, format, target query, channel.
The content calendar puts dates and owners on that list.
If you start from the third, you get regular publishing with no direction at all. It is the mistake we find in most of the company blogs we analyse: schedule kept, forty articles a year, zero enquiries.
The number to start from: how much of what you publish produces nothing
Before writing the new plan it pays to look at what the old one delivered. This is the raw data from two Search Console properties we work on, same period, 1 May to 20 August 2026.
| Site | URLs with impressions | Impressions | Clicks | CTR | URLs with zero clicks |
|---|---|---|---|---|---|
| B2B services site (visilay.com) | 96 | 53,723 | 106 | 0.20% | 71 (74%) |
| Italian outdoor e-commerce (anonymised) | 23 | 231,744 | 4,503 | 1.94% | 6 (26%) |
The B2B site is us. The most useful detail is not the overall 0.20% CTR, it is the distribution: the SEO service page collected 8,780 impressions and 6 clicks, an English article on B2B marketing strategies collected 9,335 and no clicks. Ranking twentieth for a query with thousands of searches produces impressions that look like a result and are not one.
The outdoor e-commerce site is a client, so the numbers stay anonymous: 23 URLs with impressions, 4,503 clicks, 93.8% of them on the two versions of the home page. The site's only editorial piece, a page on what trail running is, totalled 256 impressions and zero clicks in three and a half months. It is an article published without a plan: nobody ever gave it a role, so it plays none.
The first exercise, before any outline: export the pages from Search Console, sort them by clicks and count how many are at zero. That number is the real starting point.
Step 1. A goal counted in enquiries
Generic goals ("increase brand awareness", "grow organic") do not let you rule anything out, and a plan exists mainly to rule things out.
A useful goal has three parts: a quantity, a unit that touches revenue, a date. Twenty qualified quote requests a month within twelve months. Thirty demo sign-ups per quarter. Eight form enquiries on a specific product line.
It changes everything downstream. If the unit is quote requests, an article on the history of the sector does not make it into the plan even if it has search volume. If the unit is visits, it does, and a year later you have a rising graph and a sales team with an empty diary. We have written a dedicated guide on the parameters to keep an eye on, the SEO KPIs, and the economic reasoning is in the article on ROI.
Step 2. The questions come from quotes, not from tools
Keyword research tools are there to estimate volumes. They do not know what customers ask you on the phone.
Three sources to work through before opening Semrush or Ahrefs:
- The last fifty enquiries received through the site or by email. Note the words the customer used to describe the problem. Those words are queries, written by people who buy.
- The sales team. Which questions do they get asked in the first two minutes of a call? Almost always they are questions the site does not answer anywhere.
- Customer support. Recurring tickets are content that has already been validated: someone had that problem badly enough to write in.
Only then do you move on to volumes, to find out which of those questions also have search demand and how they are phrased. In the UK, for instance, "content plan" gets about 720 searches a month and "content marketing plan" about 170, while "how to create a content marketing plan" barely registers (DataForSEO, Google UK, September 2026): the same need, worded three ways, with very different demand. The working method is in the guide to keyword research; checking what the person typing that query really wants is covered in the article on search intent.
Step 3. Decide what not to write
It is the part of the plan nobody puts in writing, and the one that saves the most money.
The Pew Research Center tracked the browsing of 900 US adults across 68,879 Google searches made in March 2025. When an AI-generated summary appears on the page, the share of visits that end with a click on a result drops from 15% to 8%; only 1% click a link inside the summary. The data is from the US, but AI Overviews are live in the UK too and the direction is the same.
Turned into a planning rule: short, definitional questions should be covered in order to be cited, not to bring traffic. An 1,800-word article on "what is content marketing" is a losing investment today, unless you put something in it that AI cannot summarise because nobody else has published it. We have written two pieces on how the citation mechanism works: one on AI Overviews and one on how to appear in ChatGPT answers.
The questions that still get clicks are the ones where the user has to evaluate, compare, buy or decide whom to trust. The plan should lean that way.
Step 4. A role for every piece of content
In B2B a buying decision involves several people, and each one arrives on the site with a different question. In the industrial projects we work on there are four recurring figures: the technical lead looks for constraints and specifications, procurement looks for suppliers and comparisons, finance looks for total cost, senior management looks for evidence that the company is reliable.
A plan that only produces content for the first figure collects technical visits that never reach a negotiation. One that only produces company stories collects compliments.
In the plan's table, next to each piece, you need one column for the person it speaks to and one for the action it should trigger. If the column stays empty, the piece goes. The full reasoning is in the guide to B2B SEO and, in a sector version, in the one on SEO for manufacturing. We have collected the rest on the formats that work in business to business in the article on B2B content marketing.
Step 5. The calendar, and the slot for updates
There is only one rule here: frequency is set by real production capacity, not by ambition. Two articles a month published for three years beat eight a month published for five months.
The part almost no content calendar allows for is time to rework what is already there. In our plan one slot in three is an update: an article that has lost positions, one that gets impressions without clicks, one about a tool that has changed in the meantime. It is less satisfying work than the blank page and it pays back far more.
The candidates take two minutes to find: in Search Console, pages with stable impressions and falling clicks. Anyone with a large, messy archive should start with a round of content pruning before adding anything. For setting up regular production, the method is in the blog SEO strategies and in the article on the ideal blog post length.
Step 6. The three things to measure
The Content Marketing Institute's research B2B Content Marketing: Insights for 2026, carried out with MarketingProfs and Storyblok among 1,015 B2B marketers between 24 June and 14 August 2025 and mainly covering North America, puts measuring effectiveness third among the stated challenges, at 33%. Above it are creating content that prompts action (40%) and resource constraints (39%).
Three metrics are enough, and they are listed in the order you should read them.
- Enquiries attributed to content. Not visits: completed forms, calls, quote requests that have a content page in their path.
- Queries in positions 8-20. These are pages one step from page one. It is the indicator that tells you where to work next month.
- Clicks per page, not per site. The site average hides everything. The distribution tells you whether you are building an archive or a graveyard.
Since June 2026 Search Console has included reports dedicated to AI features, with impressions broken down by page, country and device: visibility in generated answers has become a readable metric and belongs in the monthly report. We have a dedicated guide on using Google Search Console.
Again from the Content Marketing Institute research: 95% of the organisations surveyed use AI tools, 87% of those using them to create content report productivity gains, but only 39% report better content performance. More content produced faster is not more content that works.
How long it takes before it works
Longer than quotes suggest. Macropix, a Milan-based LED screen manufacturer, did not appear for any of its sector's category keywords in 2020: for "monitor pubblicitario" (advertising monitor) it sat in position 88, which is page nine. In 2025 the same query shows it in second place, and in July 2026 the macropix.it domain holds a 25.55% share of voice on the ledwall cluster, ahead of amazon.it at 18.08% (source: Semrush, Google Italy).
Five years. A content marketing plan is not the lever for this quarter: in the meantime you need channels that bring enquiries from day one, and the most direct route is SEO consultancy alongside paid search. More cases with numbers are on the projects page.
A point of context on how uncrowded the field still is: having a website is now the norm for companies of almost any size; having a plan that decides what goes on it is not.
The mistake we see most often
It is not the writing, and it is not technical SEO. It is that the plan does not say who answers the questions.
Useful B2B content contains information that lives in the head of an engineer, a salesperson, whoever has handled the complaints. If the plan does not set aside half an hour a month of those people's time, the copywriter rewrites what is already on page one and the piece comes out identical to everyone else's. It is also the practical difference between content Google recognises as first-hand experience and generic content: we covered this when writing about E-E-A-T and how to create SEO content.
In the plan, next to every row, goes an internal name and a date. Not the author of the text: the person who supplies the substance.
Frequently asked questions
The content marketing plan defines the commercial goal, audience, topics covered and metrics. The editorial plan turns those topics into individual pieces with title, format and target query. The content calendar adds dates and owners. Anyone who starts from the calendar publishes regularly without knowing why.
You get the number by dividing the hours the company can genuinely give to content by the cost of a piece with original data, which usually runs to six to ten hours across research, internal interview, writing and review. Two pieces a month kept up for years are worth more than eight a month for five months.
It changes which content is worth producing. According to the Pew Research Center, in a US sample of 900 adults and 68,879 searches, when an AI summary appears the share of visits with a click on a result drops from 15% to 8%. Definitional questions lose clicks; questions about evaluating, comparing and buying keep them.
In the B2B projects we work on, the first enquiries from content arrive after six to twelve months, and a stable position on category keywords takes years: for Macropix, the Italian query monitor pubblicitario (advertising monitor) went from position 88 in 2020 to second place in 2025.
With enquiries attributed to content, not with visits. Then read the queries in positions 8-20, which show where to work the following month, and the distribution of clicks per page, which the site average hides.
The piece missing from almost every plan
The column most often missing is not the keyword or the date: it is the cost. A 2,000-word article with original data takes six to ten hours across research, internal interview, writing and review. With that number in hand, the question "how many articles a month do we publish" stops being an editorial choice and becomes a division: how many hours the company can genuinely put in, divided by the cost of a piece worth reading.
Most plans break there, not on strategy. Whoever writes it down in black and white before starting publishes less and stops redoing the calendar every two months. If you need a hand setting it up, let's talk: the first step, usually, is opening Search Console together and counting the pages with zero clicks.
