Content Scoring: How to Evaluate Quality Before Publishing

Most weak content isn't published because someone decided it was good. It's published because nobody stopped to ask whether it was. Content scoring is the habit of asking — running each piece through a consistent check before it goes out, so "is this good enough?" stops being a gut call made under deadline and becomes a decision you can make the same way every time. It's the difference between publishing on purpose and publishing by default.

Why score content at all

The case for scoring is simple: content quality is a decision, and undecided decisions default to "publish." When there's no standard, the piece that's merely finished gets treated the same as the piece that's genuinely good, because the only question anyone asked was "is it done?" A score forces the better question — "is it worth it?" — at the one moment it still matters, before the thing is live under your name.

This matters more now than it used to, because the cost of thin content has gone up. Search rewards depth and consistency; a pile of forgettable posts doesn't just fail to help, it dilutes the authority of the good pages around it. AI answer engines simply skip content that says nothing specific. So the weak piece isn't neutral — it's a small tax on everything else you've published. Scoring is how you stop paying it, one piece at a time.

There's a quieter benefit, too. A shared standard turns "good" from one person's taste into something a team can agree on and repeat. Without it, quality lives in whoever happens to review a piece that day, and it drifts every time that person changes. With it, the same bar applies whether the founder, a writer, or a new hire is holding the draft. Scoring doesn't just catch weak pieces — it makes "what we consider good" explicit enough that other people can hit it without you in the room, which is the only way quality survives handing the work off.

Scoring is not an audit

It's worth drawing a clean line between two things that get confused, because they solve different problems. Scoring happens before you publish — it's a gate a new piece has to pass. An audit happens after, across everything you've already put out — it's an inventory of a library you built before you had a standard. One keeps the quality of what's coming; the other fixes the quality of what's there.

You need both, and they reinforce each other. If you only audit, you're forever cleaning up messes you keep making, because nothing stops the next weak piece from going out. If you only score, your back catalogue stays full of the thin content you published before the gate existed. Score to stop the bleeding; audit to drain what's already pooled. We cover the second half of that in how to conduct a content audit; this piece is about the gate.

The five dimensions of quality

A useful score measures a few things that actually predict whether a piece will earn anything, not a long checklist nobody completes. Five dimensions cover most of it. Relevance: is this written for a specific reader with a specific question, or for "everyone," which is another word for no one? Depth: does it say something only you could say, from real experience, or does it restate what's already everywhere?

Clarity: can a reader — and a machine — follow it and lift the point, or is the answer buried under a wall of text? Search fit: does it resolve a genuine question people actually ask, structured so it can be found and cited, without keyword stuffing? Business value: does it build trust and open a path toward a conversation, or is it a pageview that leads nowhere? A piece doesn't need a perfect ten across all five. But a piece that's weak on most of them isn't ready, however finished it looks.

To see the five working together, score a hypothetical. Say you've drafted "Five tips for better email marketing." Relevance: weak — it's aimed at everyone with an inbox, no specific reader. Depth: weak — the five tips are the same five on every other blog. Clarity: fine, it's well-formatted. Search fit: crowded — a thousand near-identical pages already answer this. Business value: near zero — nobody finishes it and thinks of you as the obvious partner. Three clear fails, two passes, and the piece still looks perfectly publishable. That's exactly the content a score exists to catch: competent, finished, and pointless.

The scorecard

Here's the whole thing as a table you can hold a draft against. For each dimension, decide honestly which column the piece is closer to.

Two things about using it in practice. The first is that the columns are deliberately written as descriptions rather than as numbers, because a number invites averaging and averaging is how a piece with two catastrophic weaknesses and three strengths comes out looking acceptable. You are not adding anything up. You are reading five sentences and deciding which one each dimension is closer to.

The second is that the honest answer is often "somewhere in between," and that's fine as long as you say which side of the middle. A dimension you genuinely cannot place is usually a signal that the piece hasn't decided something — most often who it's for, which is relevance wearing a disguise. In that case the useful output of the score isn't a verdict at all; it's the note that sends the draft back with one specific question attached.

The two dimensions people fudge

In practice, three of the five are easy to judge honestly. Clarity, search fit, and relevance are fairly visible — you can see whether the point is buried, whether there's a real query behind the piece, whether it's aimed at someone specific. The two that get fudged, every time, are depth and business value, because they're the ones that ask uncomfortable questions.

Depth is uncomfortable because the honest test is: "would anyone who already knows this subject learn anything here?" A lot of content fails that quietly — it's competent, well-formatted, and adds nothing a dozen other pages don't already say. Business value is uncomfortable because the honest test is "does this actually move someone toward working with us, or does it just exist?" Content that scores well on the easy three and poorly on these two is the exact trap most businesses fall into: polished, plentiful, and inert. Score those two strictly, and you'll cut the pieces that were always going to underperform, no matter how clean the formatting.

Setting the publish threshold

A score is only useful if it forces a decision, so the point of scoring is the threshold, not the number. Decide in advance what a passing piece looks like — for most teams, something like "strong on at least four dimensions, and never weak on depth or business value." Anything that clears it, publish. Anything that doesn't lands in one of two buckets: revise, if a clear fix would get it there, or cut, if the piece was never going to earn its place no matter how much you polish it.

That second bucket is the one that makes scoring worth doing. The willingness to cut a finished piece — to spend the effort of writing it and still not publish it — is what separates a real standard from a rubber stamp. It feels wasteful in the moment. It isn't. Publishing a weak piece costs more than not publishing it, because it dilutes everything around it and earns nothing itself. The draft that dies at the threshold did its job: it told you where your standard is.

Scoring one real draft

Scoring is quick enough that describing it takes longer than doing it, so here is one draft going through the gate and the record it leaves behind. The piece is 1,900 words on client handovers, written from a real position, and the scoring took about four minutes.

The thing worth noticing isn't the verdict. It's that the score got written down next to the draft rather than said out loud in a review call — because a scoring habit that leaves no trace can't tell you anything six months later about which weakness keeps recurring.

The verdict on this one was publish: strong on relevance, depth and business value, fine on clarity, unremarkable on search fit. Four minutes, one line of reasoning per dimension, and a note explaining the depth call — because depth is the dimension a future reader of that file will most want justified.

You can automate the record-keeping without automating the judgment, and that split is the whole trick. An n8n job can open the score file when a draft is marked finished, prompt for the five lines, and file the result; Claude Code can pre-fill the mechanical observations — whether the piece answers a real query, whether the point is near the top, whether it repeats something already published. What no tool should do is issue the verdict, because a model asked whether a draft is good will tell you yes.

Making scoring a habit, not a bottleneck

The risk with any quality gate is that it turns into bureaucracy — a form people fill in without thinking, or a delay that makes publishing feel like paperwork. Keep it light. Scoring should take a few minutes, not a meeting. Five dimensions, an honest read, a decision. If it becomes a ritual where the number is the point and the judgment is skipped, it's actively harmful, because it launders weak content through a process that looks rigorous.

The way to keep it real is to keep it about the decision, not the record. You're not producing a grade for a spreadsheet; you're answering one question — publish, revise, or cut — as honestly as you can, the same way every time. Done right, scoring doesn't slow good content down at all; it clears quickly. What it slows down is weak content, which is exactly what you want slowed.

One structural safeguard is worth building in: whoever wrote the piece should not be the only person who scores it. Not because writers can't be honest, but because four minutes after finishing something is the worst possible moment to ask whether it says anything new. A second reader with the same five dimensions in front of them takes about the same four minutes and catches the two dimensions that get fudged. On a team of one, the substitute is time — score it the next morning, not the moment you stop typing.

What a score cannot tell you

Being honest about the limits keeps the habit from turning into superstition. A score is a judgment made before anyone has read the piece except you, which means there are things it structurally cannot see — and knowing which things stops you trusting the number further than it deserves.

Two failure modes follow from that. The first is treating a high score as a promise: the piece cleared the bar, so it must perform, and when it doesn't the standard gets blamed. It shouldn't be — clearing the gate means the piece deserved publishing, not that it will work. The second is scoring in a way that optimises for the rubric rather than the reader, which is how you end up with content that is measurably strong on all five dimensions and somehow still inert.

The correction for both is the same: the score decides publish, revise or cut, and nothing else. What actually happened to a piece is a separate question, answered months later by looking at what people did with it — and that answer belongs in the audit, not in the gate.

Why a system raises every score

Here's the thing scoring reveals over time: the same weaknesses keep showing up. Pieces are thin because the knowledge behind them was thin. They're irrelevant because there was no clear reader in mind. They're hard to quote because there was no structure to write into. Scoring catches these one at a time — but a good system prevents most of them before a draft exists.

This is also the answer to the objection that scoring is just gatekeeping your own work. It would be, if the score were the end of the process. It isn't — it's an instrument pointed upstream. Every cut carries a reason, and reasons accumulate into a pattern: three pieces cut on depth in a quarter is not three bad drafts, it's a knowledge layer that hasn't been fed. Four cut on relevance is a topic-selection problem, not a writing problem. The gate is where you find that out cheaply.

That's the real relationship between a score and a Content OS. When your knowledge is organized, your topics are chosen deliberately, and your structure is a given, most drafts arrive at the scoring gate already strong — because the system did upstream what the score only measures at the end. Scoring stays valuable as the honest final check. But if you find yourself cutting piece after piece for the same reason, the score isn't the problem to fix. The system that keeps producing that weakness is. A good score is a symptom of good work, and good work is what a system is for.

Who does the scoring

There's a question underneath all of this that gets skipped surprisingly often, and it decides whether any of the rest works: who is holding the scorecard? A standard is only as reliable as the person applying it, and the obvious answer — the writer scores their own draft — has a known and well-documented problem.

You cannot see the gap between what you meant and what you wrote. The draft reads clearly to you because you have the missing context in your head, supplying it silently every time your eye passes over a thin sentence. That is not carelessness; it is how writing works for everybody, and it is exactly why self-scored specificity and usefulness run consistently high. If one person is the only reader before publication, expect the scores to be roughly a point generous on the two dimensions that matter most, and calibrate the threshold accordingly rather than pretending otherwise.

The cheap fix is a second pair of eyes that is not required to be an expert. A colleague who knows nothing about the subject is better at judging clarity than one who knows it well, because they cannot fill in the gaps for you — if they can't say what the piece argues after one read, the score on clarity is wrong regardless of what you gave it. Five minutes of somebody else's attention catches more than an hour of your own re-reading, and it doesn't need seniority to do it.

For a solo founder with nobody to ask, the substitute is time and a change of medium. Score it the next morning rather than the night you finish, and read it somewhere other than the document — on a phone, or out loud. Both work by breaking the familiarity that hides the gap. A model can help here too, and its useful question is not is this good — it will say yes — but what does this piece claim, and what would someone need to already know for it to make sense? The answer tells you what you left in your head.

Frequently asked questions

What is content scoring?

Content scoring is evaluating a piece of content against a consistent set of quality criteria before you publish it, so you catch weak content before it goes out rather than after. It turns "is this good enough?" from a gut call into a repeatable check across a few dimensions — relevance, depth, clarity, search fit, and business value — that everyone applies the same way.

How is content scoring different from a content audit?

Scoring judges one piece before it's published; an audit reviews your whole existing library after the fact. Scoring is a gate you pass new work through; an audit is an inventory of what's already out there. You use scoring to keep quality high going forward, and an audit to find and fix what was published before you had a standard.

Should I use a numeric score?

A rough number helps if it forces a decision — publish, revise, or cut — but the number matters less than the dimensions behind it. The goal isn't a precise grade; it's a consistent, honest check that stops weak content going out under your name. If a score ever becomes a box to tick rather than a real judgment, it's doing harm, not good.

Doesn't scoring slow down publishing?

A little, and that's the point. The businesses that win with content publish less but better. A short check that catches a thin piece before it's published saves the far larger cost of diluting your authority with content that was never going to earn attention, a ranking, or a citation. Scoring trades a few minutes now for not spreading your credibility thin later.

What are the five dimensions of content quality?

Relevance, depth, clarity, search fit and business value. Relevance asks whether the piece is written for one real reader with one real question. Depth asks whether it says anything only you could say. Clarity asks whether a reader or a machine can lift the point without digging. Search fit asks whether it resolves a genuine query and is structured to be found. Business value asks whether it opens a path toward a conversation. A piece does not need to be strong on all five, but a piece weak on most of them is not ready however finished it looks.

Who should score a draft?

Ideally not only the person who wrote it. Four minutes after finishing something is the worst moment to judge whether it says anything new, so a second reader working from the same five dimensions catches the two that get fudged — depth and business value. On a team of one the substitute is time: score it the next morning rather than the moment you stop typing.

Can AI score content for you?

It can do the mechanical observations and should not issue the verdict. A tool like Claude Code is reliable on whether a piece answers a real query, whether the point sits near the top, and whether it repeats something you already published — and an n8n job can open and file the score record so the habit leaves a trace. What it cannot do is tell you whether a draft is good, because a model asked that question will say yes. Automate the record-keeping, never the judgment.

Where to start

Don't build an elaborate system. Take the next piece you're about to publish and run it through the five dimensions honestly — relevance, depth, clarity, search fit, business value — and be strict on the two people fudge. Decide: publish, revise, or cut. Do that for a few pieces and two things happen. Your published work gets noticeably stronger, and you start to see the patterns in what you almost published — the recurring weaknesses that point back to something upstream worth fixing. Scoring is a small habit with a large effect: it makes "good enough" a decision you make on purpose, instead of one that gets made for you by a deadline.

Keep reading

• How to conduct a content audit — scoring's after-the-fact companion for your existing library.
• How to create content that compounds — why quality, not volume, is what pays off.
• What is a Content OS? — the system that makes most drafts strong before they're scored.