Weekly content ideas from calls
Use this skill at the end of the week when next week's content is still a blank page and the only raw material worth using is what clients and prospects actually said on calls. Produces 8 to 12 ranked content ideas, each with the verbatim quote behind it, a hook, and two formats to write it in, plus the questions to answer before drafting. Also use for "Friday content pull", "what should I post next week", "mine my calls for content", "turn client calls into LinkedIn posts".
Garrett JesticeFounder and GTM Advisor, 10x SoloApplies on a fixed day at the end of the week, before next week's content is planned. Produces a ranked idea list built only from things real buyers said out loud.
The play
- Pull every call transcript from the last seven days: sales calls, client calls, partner calls, internal strategy calls where a client problem was discussed. Nothing older. Stale weeks produce generic ideas.
- Read for moments, not topics. A moment is a sentence a buyer said that carries tension, a number, a contradiction, or a phrase in their own words. The types that keep earning, with examples, are in references/moment-types.md. Expect 15 to 30 candidate moments from a normal week and far fewer from a week of status calls.
- Pull each one as a verbatim quote with who said it (role, not name) and what prompted it. Paraphrase at this stage and the idea goes flat by Tuesday.
- Score every candidate against the ranking model in references/ranking-model.md. The metric is comments per 1,000 impressions on past posts, not likes and not raw impressions: it is the only one that tracks whether the audience had something to say back. Build the baseline once from the last 90 days of posts and reuse it.
- Cut to 8 to 12 ideas. Fewer than 8 means the week's calls were too thin, so widen to 14 days once and say so. More than 12 and nothing gets written well.
- For each surviving idea write: the quote, a hook line, and two formats it could be written in. Two formats, not one, because the same moment lands differently as a story than as a list, and the choice is made at drafting time.
- Flag the top 3. These are the ones that get written if the week collapses. Branch by what the week produced:
- A moment repeated by two or more different buyers goes straight to the top 3, whatever it scores.
- A single client's very specific situation scores well but is not a top 3 pick unless the pattern shows up elsewhere.
- A moment that contradicts something already published is always a top 3 pick. Those are the posts that get comments.
- Close with 6 to 8 voice questions, drawn from references/voice-questions.md and pointed at the top 3. These are what the writer answers before drafting, so the post carries a real position instead of a summary of the call.
What good looks like
- First thing I check that most people skip: whether the quote would survive being read aloud to the person who said it. If it would embarrass them, it is not usable, however good the hook is.
- The mediocre version: a list of topics lifted from call summaries. "Clients are struggling with attribution." True, generic, already posted by everyone. The quote is what makes it specific, and the mediocre version drops it.
- I know the output is right when a writer can open the list cold on Monday and start typing without listening to a single call.
- Second test: at least two of the ideas make the writer slightly uncomfortable. Comfortable ideas get no comments.
- Red flag: the top 3 are all from the same call. That is one conversation, not a week of signal.
Rules
- MUST keep every idea anchored to a verbatim quote from an actual call.
- MUST strip client and company names before anything is published, keeping role, size and situation.
- MUST rank on comments per 1,000 impressions, never on likes or follower count.
- MUST give each idea two formats.
- MUST run on the same day every week. The cadence is what makes the backlog compound.
- NEVER publish a quote that identifies the speaker without their explicit yes.
- NEVER carry an unused idea forward more than two weeks. If it has not been written by then, it was not an idea.
references/moment-types.md
What counts as a moment
Seven types earn their place on the list. Everything else is a topic, and topics make generic posts.
1. The buyer's own phrase
They describe their problem in words no marketer would choose. "We're paying for a dashboard nobody opens." Why it works: it is search-and-scroll language. Use it verbatim in the hook.
2. The number they volunteered
Any figure they said without being asked: budget, headcount, hours, close rate, how long something has dragged on. Why it works: a number in the first line stops the scroll and makes the post falsifiable.
3. The contradiction
They say two incompatible things in one call. "We don't have budget for this" followed by a six-figure spend on something adjacent. Why it works: naming the pattern without naming the person is the most commented-on post shape there is.
4. The thing they tried that failed
A previous vendor, tool or internal attempt, and the specific reason it did not stick. Why it works: it is the objection your whole market is carrying, said out loud.
5. The question they asked you
Especially a question asked twice, or asked by three different buyers this quarter. Why it works: a question asked repeatedly is a content calendar.
6. The push-back
Where they disagreed with you on the call. Keep the disagreement intact. Why it works: the post writes itself as "here is why I still think X", which is a position, not a summary.
7. The aside
The thing said in the first two minutes or after "anyway" that had nothing to do with the agenda. Why it works: it is the unguarded version of their priorities.
Not moments
- A restatement of their job title or company description
- Anything from the agenda slide
- Compliments about your work
- Your own clever line from the call
- Agreement
Capture format
Quote: "..."
Said by: [role, company size, segment]
Prompted by: [what was being discussed]
Type: [1-7]
A normal week of 6 to 10 calls yields 15 to 30 candidates. Under 10 candidates usually means the week was status calls, not discovery. Widen the window to 14 days once, then stop.
references/ranking-model.md
Ranking model
The metric is comments per 1,000 impressions (CP1K) on published posts. Likes are passive and follower-weighted; impressions reward reach, not resonance. Comments cost the reader something, so they are the only cheap signal that a post gave someone a reason to speak.
Build the baseline once
- Export the last 90 days of posts with impressions and comment counts.
- CP1K = comments / (impressions / 1,000).
- Sort and take the median. That is the baseline. Recompute quarterly, not weekly.
Typical ranges for a founder or practitioner account. Use your own numbers once you have 90 days.
| CP1K | Read |
|---|---|
| Under 1.0 | Nobody had anything to say. Topic was agreeable. |
| 1.0 to 2.5 | Working. The normal band. |
| 2.5 to 5.0 | Struck a nerve. Look for what the post did that the others did not. |
| Over 5.0 | Either a genuine position or a fight. Check which before repeating it. |
Score each candidate idea
Five factors, 1 to 5 each. Total out of 25.
| Factor | 1 | 5 |
|---|---|---|
| Specificity | A topic | A sentence a named role actually said, with a number or a concrete situation in it |
| Tension | Everyone agrees | A real practitioner would push back |
| Recurrence | Heard once | Heard from three or more different buyers |
| Proximity to the buyer's money | Interesting | Costs them revenue, headcount or a deadline right now |
| Past CP1K of the nearest matching post | Below baseline | Top quartile |
Thresholds
| Score | Action |
|---|---|
| 19 to 25 | Top 3 candidate |
| 13 to 18 | Keep in the 8 to 12 |
| Under 13 | Cut, unless it is the only moment from a client segment with no coverage this quarter |
Overrides that beat the score
- Repeated by two or more different buyers in the same week: promote to top 3.
- Contradicts something already published: promote to top 3.
- Single client, highly specific, no pattern elsewhere: cap at the 8 to 12, never top 3.
- Any moment that cannot be anonymised: cut, whatever it scores.
references/voice-questions.md
Voice questions
Six to eight questions close the output. They are pointed at the top 3 ideas and answered by the writer before drafting. The point is to get a position on the page, not a summary of the call.
Pick from the bank below. Two or three per top idea, plus one closing question for the set.
Position
- What do you believe about this that most people in your market do not?
- What would you have said if you had been less polite on the call?
- Who is wrong here, and what are they actually doing instead?
Evidence
- What have you seen happen more than three times that makes you sure?
- What is the number you would put behind this, and where did it come from?
- What is the smallest concrete example that proves it?
Cost
- What does this cost the reader if they keep doing it the current way, in money or time?
- What is the deadline or event that makes it urgent for them this quarter?
Counter-case
- When is the opposite true? Who should ignore this advice?
- What would a smart person say back, and what is your answer?
Specificity
- What detail from the call would you be nervous to include, and why?
- What word did the buyer use that you would never have used?
Closing question for the set
- If you could only publish one of these three next week, which one and what changes?
How to use the answers
Answers go straight into the draft, lightly edited. The hook comes from the buyer's phrase or the number. The body comes from the evidence and cost answers. The counter-case answer becomes the second half of the post and is usually what earns the comments.
If a question cannot be answered in two sentences, the idea is not ready. Move it down and promote the next one.