A topic backlog fails in two ways. One is empty, and the team searches chat history the day before publishing. The other is full, with two hundred ideas and no answer to what comes next. The first is a replenishment problem. The second is usually a scoring and pruning problem.
The parent guide on the social media content system already lays out the framework, pillars, topic sources, four scoring dimensions, five statuses, a two-to-four-week buffer, platform adaptation, and the production flow. This article does not repeat the framework. It fills in the four pieces that stall in practice, scoring criteria, operating metrics, quarterly pruning, and handoff fields. If the wider landscape is new, start with the global social media guide.
Four Dimensions with Written Criteria
The parent guide names four dimensions, business relevance, audience fit, producibility, and timeliness, each scored one to five. Without criteria, two people can differ by three points. Make the score comparable by writing down what one, three, and five mean.
| Dimension | 1 | 3 | 5 | Evidence |
|---|---|---|---|---|
| Business relevance | No direct connection to product, service, or revenue | Indirect, builds awareness or trust | Directly maps to a product, offer, or sales scenario | Can you name the product page, quote, or sales question |
| Audience fit | Not a question the target audience cares about | Broad industry topic, possibly interesting | Core audience is asking, searching, or discussing it | Support logs, comments, search terms, community questions |
| Producibility | Needs assets, data, rights, or people we do not have | Doable with existing assets but needs a shoot or interview | Can be produced with current assets and capacity | Asset library, interview subject, legal and schedule confirmed |
| Timeliness | Window has passed, or the content ages badly | Evergreen, or the window is still ahead | Window is open, publishing within two to four weeks pays off | Policy dates, event calendar, platform discussion volume |
The total is 20. The default scheduling line is 14, with no dimension at 1. A high total with a 1 means a clear weak spot, so fix that first. The threshold is a team rule, not an industry standard. Adjust it at the quarterly review, not per topic.
A few details.
Business relevance is not brand exposure. It means naming the product, page, or sales scenario the content eventually supports. If you cannot name it, the score is at most 3.
Audience fit depends on evidence. A question support answers five times a week deserves a 5 more than a topic the meeting believes users should care about. Without evidence, the score is at most 3. For finding search demand, see the keyword research and search intent guide.
Producibility needs an early check. Do the assets exist, does it need a shoot, does legal need to review it. Many topics die after scheduling, and the cause is usually assets or capacity, not the idea itself.
Timeliness has two traps. News with a closed window scores 1. Evergreen tutorials should not be marked down for lacking a hot topic, so they score 3. A 5 goes only to a topic whose window is open and whose team can deliver inside it. Scoring a 5 and scheduling three weeks later is self-deception.
When Two People Disagree
One scorer brings bias. Two scorers bring arguments. Set the rules before disagreement appears.
First, score independently, without comparing notes. Each person writes the score and one reason tied to evidence, not a feeling.
Second, check the total gap. A gap of 2 or less, with no single dimension differing by 2 or more, averages out and goes into the library. A gap of 3 or more, or any single dimension differing by 2 or more, goes to discussion.
Third, discuss only the dimension in dispute, with evidence. If business relevance gets a 2 from one person and a 5 from another, return to the question of which product page the content supports. Name it and the score goes up. Fail and it goes down. Do not reopen all four dimensions.
Fourth, if disagreement remains, the content lead decides and records the reason. The decider is the person accountable for business results, not the person with the higher score.
Fifth, calibrate monthly. Rescore five published topics and compare the original scores with actual performance. Look for a dimension that is consistently overrated or underrated. Calibrate the criteria, not the people.
The process sounds heavy. Once it runs, each score takes a few extra minutes. The real change is a shared language for what is worth making. Precision matters less.
Four Operating Metrics
Backlog health is not the number of rows. Track four metrics, each with a clear numerator and denominator.
| Metric | Definition | Question it answers | Dashboard fields |
|---|---|---|---|
| Buffer weeks | (Scheduled + publishable in review) divided by average weekly output | How many weeks are covered | Weekly output, publishable count |
| Intake-to-publish rate | Topics published in the last 90 days divided by topics added in the same period | How many ideas become content | Intake, published, rate |
| Reuse rate | Topics reused in the last 90 days divided by topics published in the same period | Does content live only once | Reuse count, reuse format |
| Pillar balance | Deviation between actual and target share of published topics by pillar | Is one pillar eating all capacity | Published count and target share per pillar |
Buffer weeks follows the parent guide's two-to-four-week principle, but the definition must be fixed. Counting only scheduled topics understates. Counting every idea overstates. Scheduled plus publishable items in review is closer to real capacity.
Intake-to-publish needs definitions. Intake means topics that passed scoring and entered the to-write status, not every note. Published means actually live, not scheduled. If the denominator is fuzzy, the rate improves every year and explains nothing.
Reuse needs a definition. Cross-platform adaptation, splitting into short clips, updating and republishing after six months, and compiling into a collection all count. Reposting the same content with a new cover does not. A high reuse rate suggests long-tail value. A low one suggests content is consumed once.
Pillar balance does not require equal shares. Targets can differ, for example 40 percent education, 30 percent product evidence, 20 percent brand story, and 10 percent industry opinion. Watch the deviation from target. A gap above ten percentage points means one type is overproduced.
Put all four in one weekly report, one row per week, and read eight weeks in a row. One odd week is noise. Four consecutive weeks moving the same way deserves action.
Quarterly Pruning Rules
Bigger is not better. An oversized library raises scoring and maintenance costs, eats capacity, and makes the team slow to schedule because there is always more. Prune quarterly with three rules.
| Type | Test | Action |
|---|---|---|
| Long idle | 90 days since intake, still in idea or to-write, never scheduled | Downgrade to watch list. Delete at 180 days with no movement |
| Repeatedly postponed | Postponed three or more times, or moved out of three consecutive scheduling cycles | Check dependencies first. Missing evidence gets fixed, missing owner gets reassigned. If still stuck, split or cut |
| No response after publishing | Eight weeks live, engagement, clicks, and conversions all below the pillar median | Mark as do not reuse. If the proposition is still core, rewrite and retest once, not repost as is |
Keep a record for all three. Do not erase the row. Set the status to retired and write the reason. When someone raises the same idea six months later, the history explains why it was dropped.
Write the merge rule too. If two topics share a pillar, an audience, and a problem, and neither reaches the scheduling line, merge them and keep the stronger evidence. The goal is less duplication, not making both sides happy.
Pruning meets psychological resistance. A simple test helps. If a topic has not been scheduled for two quarters, it is unlikely to find a better moment. Keeping it only makes the topics that matter look less urgent.
Run the quarterly meeting in half an hour. Review the four metrics, then the three pruning lists, then refill to the target buffer. When refilling, favor underweight pillars instead of adding more of the type already in surplus.
Topic Cards and Handoff
A topic moves from idea to scheduled through a card. If the fields are incomplete, production will rework it.
| Field | What to write | Why it matters |
|---|---|---|
| Proposition | One sentence a reader understands | Prevents discovering halfway that there is no point |
| Evidence | Data, cases, customer quotes, source links | Without evidence, content relies on adjectives |
| Pillar | Which content pillar it belongs to | Keeps the structure balanced |
| Funnel stage | Awareness, consideration, decision | Sets depth and CTA |
| Audience and motive | Who watches, in what situation | Sets the opening and examples |
| Primary format | Short video, carousel, long post, live | Sets production resources |
| Search demand | Query or topic tag | Gives distribution an entry point |
| Score | Four dimensions and total | Scheduling basis |
| Owner | Who writes, reviews, publishes | No owner means no deadline |
| Deadline | Publish date and review date | Works production backward |
| Dependencies | Who and what assets are needed | Surfaces blockers early |
Five conditions gate the move from to-write to scheduled. The score reaches the line with no 1s. Evidence has at least two independent sources. Owner and deadline are set. The primary format and assets are confirmed. Sensitive categories or customer data have been reviewed. Miss one and it does not enter the schedule.
Handoff to production is more than a card. Attach asset links, reference examples, three must-have points, and the reviewer. The writer needs to know which sentence cannot change and which section is open.
Handoff to publishing adds three things, the scheduling window, copy and hashtags, and tracking links. After publishing, put the URL back into the card, change the status to published, and start the reuse record. For paid amplification, see the paid social guide.
Common Misconceptions
Scoring without criteria. Two people differ by three points, and the meeting decides nothing.
Averaging over disagreement. A clear split on one dimension disappears in the average and reappears in execution.
Treating backlog size as health. Two hundred unscheduled rows are worse than twenty that can run consecutively.
A fuzzy conversion rate. If intake and published definitions change, the metric is not comparable.
Refilling without pruning. The library grows until maintenance costs crush output.
Topic cards without evidence. Halfway through writing, there is no data, and the piece turns generic.
Related Reading
- Social media content system, the parent guide to pillars, backlog, adaptation, and production.
- Social audience research, turning audience and motives into topic sources.
- Social media measurement and review, post-publish metrics and review cadence.
- Organic social growth, distribution and negative feedback after publishing.