What is Justin Welsh's content matrix?
It is a grid of your sub-topics against your post styles that turns into a list of post ideas, plus a log that grades every post and schedules it for reuse. Justin Welsh, a former Chief Revenue Officer who now runs a one-person business and writes an essay every Saturday (per Welsh's about page), set it out in How to Build a 730-Day Content Library, Issue #86, published on 26 August 2023. Welsh's LinkedIn profile showed 887K followers when we checked on 20 October 2026.
The essay has four steps:
- Brain dump sub-topics. List everything you could talk about under your core topic.
- List 5 to 7 content styles. Welsh defines a style as how the content is delivered and its intended outcome.
- Build a matrix with ChatGPT. Explain your business and the outcome your content should deliver, and use five of your sub-topics.
- Create your library. Log each post, grade it, and come back to it at set intervals.
The 730 days in the title are two years, the same span as the essay's last republishing date (our reading; the essay doesn't explain the number). Below we follow each step as Welsh describes it, then build the matrix for our own product, an AI SDR for small B2B teams.
How do sub-topics and post styles produce 75 to 100 ideas?
Each cell of the grid is an idea, and each idea can become several posts. Welsh's example sub-topics, for someone building a marketing agency for small creators, are branding, networking, SEO, analytics and copywriting. The example styles are step-by-step guides, predictions, contrarian takes, inspirational stories and observations.
The essay says each idea can produce three to four pieces of content, which puts you at 75 to 100 content ideas. Our arithmetic on those numbers: five sub-topics times five styles is 25 cells, and 25 cells at three to four posts each is 75 to 100.
Welsh has described more post formats since. In 4 unconventional tactics to skyrocket LinkedIn growth (20 July 2024), two of the four strategies come with a post formula, and either could be a column in your grid:
- Trend translator. Set up Google Alerts for key topics in your industry. When major news breaks, be among the first to post what it means for your niche, using the formula Breaking News + So What? + Now What?
- Contrarian spotlight. Once a week, respectfully challenge a popular opinion in your industry and back it with data or experience: the popular opinion, your unpopular one, three to five bullet points of reasoning, and a question.
Sub-topics are easier to list when they come from buyers. In 4 ways to get instant content ideas (29 June 2024), Welsh points to four places to find the questions your ideal customers ask: Answer Socrates, YouTube comments, Google Autocomplete and competitors' content.
Our example: a content matrix for an AI SDR product
This is our own example, built for our product. We used four sub-topics, one fewer than the five Welsh suggests, and five styles: four from the library essay and the trend translator from the 2024 essay. Each cell is a working headline we wrote for this post; none is a published post.
| Sub-topic | Step-by-step guide | Contrarian take | Prediction | Trend translator | Observation |
|---|---|---|---|---|---|
| AI SDR | How to write a five-line ideal-customer summary an AI SDR can work from | Treat a daily lead goal as a target you're allowed to miss | Which outbound jobs small teams will hand to agents next, and which they won't | LinkedIn's May 2026 note on generic AI content: does it touch outreach? | Why a lead score needs a one-line reason next to it |
| LinkedIn outreach | How to plan a week of connection requests when LinkedIn publishes no number | Most connection requests don't need a note | Where LinkedIn outreach goes once every inbox fills with AI-written notes | LinkedIn updates a help page: what changes for your weekly budget | What we read on a profile before sending a request |
| Cold email | Set up a second mailbox and warm it before the first send | Write the first cold email yourself and let AI handle the follow-ups | Inbox summaries will decide whether your first line gets read | Gmail or Outlook changes a sender rule: what a two-person team does this week | Why follow-ups belong in the same thread |
| Founder-led sales | The first ten sales calls as a founder: what to ask in each | Book 20 meetings yourself before you hire an SDR | Founders will sell for longer before hiring sales | A big AI SDR funding story: what it means for founders who still sell | The objection founders hear most on first calls, and how to answer it |
Twenty cells at Welsh's three to four posts per idea is 60 to 80 posts. At three posts a week, 60 posts is 20 weeks of material (our arithmetic). The trend translator cells are placeholders: write the real post when the news breaks.
How does Welsh grade and log each post?
Every post gets a row and a score out of 10. In the library essay, Welsh fills in a formula-based Notion sheet every day with three things for each post: the date it went out, the URL of the tweet or LinkedIn post, and the style chosen that day. At the end of the day, Welsh grades each piece from 1 to 10 by how well or poorly it resonated with the audience. The essay notes the same sheet can be built in Google Sheets.
| Column | What goes in it | Whose idea |
|---|---|---|
| Date posted | The day the post went out | Welsh |
| Post URL | Link to the LinkedIn post or tweet | Welsh |
| Style | The style chosen that day, such as contrarian take | Welsh |
| Grade | 1 to 10, at the end of the day | Welsh |
| Republish dates | 6, 12, 18 and 24 months after posting, calculated by the sheet | Welsh |
| Sub-topic | The matrix row the post came from | Ours |
| Buyer replies | Comments or messages from people who fit your ideal customer | Ours |
A 1 to 10 grade is a judgement call. Welsh describes a more data-led review in Reverse engineering LinkedIn success (12 October 2024): split each post into opener, body, call-to-conversation and image, then sort past posts by a different number for each of the first three parts and study what the best ones share. Openers are ranked by impressions, bodies by likes, and calls-to-conversation by the share of the audience who saw the post and commented. For that quarter Welsh reports a 28.7% rise in average engagement rate and growth from 586,000 to over 645,000 followers in 90 days. Those are self-reported results for one large account, and the essay itself tells readers to run their own tests.
When does Welsh republish a LinkedIn post?
At 6, 12, 18 and 24 months after it first goes out. The library essay says the sheet calculates those four dates automatically because Welsh is comfortable repurposing high-performing content every six months. When a date comes round, Welsh recycles or edits the posts that resonated, and tweaks the ones that didn't to see if the score improves.
For a post published on 20 October 2026, the four dates are (our calculation):
| Interval | Republish date | If it graded well | If it graded poorly |
|---|---|---|---|
| 6 months | 20 April 2027 | Recycle or edit it | Tweak it and see if the score improves |
| 12 months | 20 October 2027 | Recycle or edit it | Tweak it and see if the score improves |
| 18 months | 20 April 2028 | Recycle or edit it | Tweak it and see if the score improves |
| 24 months | 20 October 2028 | Recycle or edit it | Tweak it and see if the score improves |
In Google Sheets, =EDATE(A2, 6) returns the date six months after the date in A2, according to Google's EDATE help page. Change the 6 to 12, 18 and 24 for the other three columns.
Why repeat at all? In The Power of Repeatable Content (30 September 2023), Welsh describes posting two near-identical tweets 12 days apart, on 7 and 19 September, and reports that they averaged 3,220 interactions each against a normal of around 900. That was one self-reported test on X, not LinkedIn, so treat it as an anecdote. The action step in the same essay is practical: pick your 3 to 5 best-performing posts and turn them into templates you can reuse on other topics.
How to adapt the matrix when the goal is pipeline
Grade on conversations with buyers as well as reach. Welsh grades by how well a post resonated with the audience. For a small B2B team, the post that matters is often the one that brings a single buyer into your messages. These changes are our adaptation, not Welsh's:
- Keep the 1 to 10 grade, with a reason. Score each post at the end of the day, as Welsh does, and write one line on why it earned that score.
- Count buyers separately. A post with 40 likes and two comments from ideal customers can matter more than one with 400 likes and none.
- Sort by sub-topic after 30 posts. See which rows of the matrix bring buyers and which only bring reach, and cut or replace the weakest row.
- Put your own material in every post. Welsh's Step 3 uses ChatGPT to produce ideas. Whoever drafts the post, it should carry something only your team knows. LinkedIn said on 20 May 2026 that content which appears AI-generated and lacks a clear perspective is less likely to be widely distributed beyond a person's immediate network (LinkedIn's announcement).
For first lines that suit each style, see our 60 LinkedIn hook examples. For full post structures, see our 12 LinkedIn post templates.
Build your matrix in one hour this week
One sitting gets you a grid, a log and a month of posts. Times are our estimates.
List your sub-topics
Write everything that sits under your core topic, then keep four or five. Start from the questions buyers asked on your last ten calls.
Pick five styles
Start with the five in Welsh's essay: step-by-step guides, predictions, contrarian takes, inspirational stories and observations. Swap one for trend translator if your industry has news most weeks.
Fill the grid
Write one working headline per cell. An AI tool can suggest headlines, as ChatGPT does in Welsh's Step 3; rewrite each one so it names something only your team knows.
Set up the log
Columns for date, URL, style, sub-topic, grade and buyer replies, plus four EDATE columns for the republish dates.
Schedule a month
Pick 12 cells for three posts a week, mixing rows and styles. If you use Siela's LinkedIn Presence agent, set the pace to 3 a week and choose "show me every post first", so you see each draft before it goes out.
Grade daily, review monthly
Score each post on the day it goes out. At the end of the month, sort by sub-topic and by style, and note the first republish date for your best three.
Frequently asked questions
What is a content matrix for LinkedIn?
A content matrix is a grid with your sub-topics on one side and your post styles on the other, where each cell is a post idea. Justin Welsh describes it in a 26 August 2023 essay: list the sub-topics under your core topic, pick 5 to 7 styles such as step-by-step guides or contrarian takes, and cross them. Welsh puts the result at 75 to 100 content ideas.
How often does Justin Welsh repurpose content?
Every six months. In the 2023 content library essay, Welsh says a Notion sheet calculates the dates 6, 12, 18 and 24 months after each post is published. On those dates Welsh recycles or edits the posts that resonated, and tweaks the weaker ones to see if their score improves.
How does Justin Welsh grade LinkedIn posts?
At the end of each day, Welsh grades the content published that day from 1 to 10, based on how well it resonated with the audience. The grade sits in the same row as the post's date, link and style, which makes strong posts easy to find when their republish date comes round.
How many sub-topics should a LinkedIn content matrix have?
Welsh's essay says to brain dump every sub-topic under your core topic, then use five of them when building the matrix with 5 to 7 styles. Five sub-topics against five styles gives 25 cells. Our own example for an AI SDR product uses four sub-topics, which gives 20 cells and 60 to 80 posts at three to four posts per idea.
Which post styles does Justin Welsh suggest?
The 2023 library essay lists step-by-step guides, predictions, contrarian takes, inspirational stories and observations as example styles. A July 2024 essay describes two LinkedIn formats: the trend translator, built on Breaking News + So What? + Now What?, and the contrarian spotlight, which challenges a popular opinion with three to five reasons and ends with a question.
Sources
- Justin Welsh: How to Build a 730-Day Content Library (Issue #86) (published 26 Aug 2023)
- Justin Welsh: 4 unconventional tactics to skyrocket LinkedIn growth (Issue #133) (published 20 Jul 2024)
- Justin Welsh: 4 ways to get instant content ideas (Issue #130) (published 29 Jun 2024)
- Justin Welsh: Reverse engineering LinkedIn success (Issue #145) (published 12 Oct 2024)
- Justin Welsh: The Power of Repeatable Content (Issue #91) (published 30 Sep 2023)
- Justin Welsh: About (checked 20 Oct 2026)
- Justin Welsh: LinkedIn profile (checked 20 Oct 2026)
- Google Docs Editors Help: EDATE (checked 20 Oct 2026)
- Laura Lorenzetti, LinkedIn: Keeping conversations real on LinkedIn (published 20 May 2026)
- Siela: 60 LinkedIn hook examples (checked 20 Oct 2026)
- Siela: 12 LinkedIn post templates (checked 20 Oct 2026)