In short: The content score is a number from 0 to 100 for one exact draft version. It compares the draft with what SerpDad already holds for that page: the brief, its sourced claims, the page's keywords, the top 10 results SerpDad read and the internal link plan. It is advice. It never blocks sign-off, and it does not predict rankings, traffic or clicks.
What the score is
- A score from 0 to 100, made of eight sub-scores with fixed weights.
- Each sub-score shows the evidence it was decided from and the fixes that would raise it.
- The score belongs to one draft version. After an edit, the old score shows greyed out until the new one is ready.
- It is stored as a receipt, like every other SerpDad decision.
What the score is not
- It is not a prediction of rank, traffic or clicks.
- It is not a copy of any search engine's system.
- It is not a gate. Only three rules block a signed export: zero unsourced sentences, every section drafted and a named sign-off. See Drafts and sign-off.
- It never edits the draft. A fix is a suggestion. Applying it makes a new version, which removes any signature.
- It never publishes.
The eight sub-scores
| Sub-score | Points | What it looks at |
|---|---|---|
| Search intent and format | 15 | Whether the draft reads as the page's format, has the structure for that format, and answers first |
| Sections answer their headings | 15 | Whether each section answers its own heading |
| Topic coverage | 20 | Brief subtopics, shared entities and searcher questions |
| Keyword use | 10 | Focus keyword placement, secondary keywords and natural repetition |
| Claims and sources | 15 | Clean facts, how widely sources are spread, and facts in every subtopic |
| Experience and authorship | 10 | Experience sentences, quotes, a byline, and an author bio and profile link |
| Readability | 10 | Long sentences, long paragraphs, heading spacing, reading grade and banned phrases |
| Internal links | 5 | Links placed and the anchor rules |
| Total | 100 |
Most points (70 of 100) come from counting in plain code. The other 30 come from four language checks that SerpDad runs on its own servers: format read, answer first, sections answer their headings, and searcher questions. These checks need no key of yours and have no cost.
Bands
| Score | Band |
|---|---|
| 80 to 100 | Strong |
| 60 to 79 | Fair |
| Below 60 | Needs work |
Below 70, the sign-off dialog adds a notice with the score and the number of open fixes. You can still sign off with the normal button. No note is needed.
When parts are missing
| Situation | What you see |
|---|---|
| Language checks could not run | The overall score reads "Not available". The code-based sub-scores and their fixes still show. Sign-off stays allowed. |
| Some parts not measured, worth 15 points or less | The overall is worked out from the measured parts, scaled to 100, and marked "Partial" |
| Parts not measured worth more than 15 points | The overall reads "Not available" |
| Full verify has not run | Claims and sources is not measured, and the same rules apply |
A language check whose accuracy has not yet been proven stays hidden and counts as not measured.
When the score is worked out
- On every save: the code-based parts, at once.
- On "Score content": the full score, including the language checks.
- Before sign-off: the full score runs on its own when someone opens the sign-off dialog on a version with no full score yet.
Where you see it
- On the article's Optimize tab, with the score, the band, one row per sub-score, and its evidence and fixes. A fix that points at a section or sentence scrolls to it.
- As a small chip on the Drafts board card and in the draft's side panel.
- In the sign-off dialog, with the open fixes, highest point gain first.
- As the Score content workflow block, which passes every article on with its score. A workflow can branch on "Score: lowest" or "Score: articles below threshold". The block's default notice threshold is 70.
The score shown at signing is stored on the sign-off record. It is not included in exports.