Execution
7 min readFrom Gap to Published Fix: The Content Workflow
Most visibility tools stop at reporting the score. Cited's content workflow runs the remaining distance: it explains each lost recommendation, ranks the gaps by impact, drafts the missing content grounded in evidence from your own audits, and re-measures after you publish. This page walks the pipeline end to end, and the time it replaces.
Diagnosis: why you were skipped
The Understand tab reads every answer where a competitor won and classifies why. Each lost recommendation lands in one of three buckets:
- Content gap: The winning brands have a page the answer leaned on, a comparison, a pricing explanation, a proof page, and you do not.
- Off-site venue: The answer was built from third-party sources, review sites, communities, publications, where competitors are present and you are absent.
- Technical: Engines could not properly read the page you already have. These route to the readiness scan.
Every diagnosis links back to the stored engine answers behind it, so you can read the evidence rather than trust a label.
The Act queue
The Act tab turns diagnoses into a prioritized work queue, ranked by how many buying conversations each gap costs across your prompt set. High-frequency prompts you lose everywhere rank above long-tail prompts you lose occasionally. Work top-down and each item you clear affects more answers than the one below it.
Content suggestions are checked against your live site and your existing drafts before they reach the queue. If you already have a page that covers the ground, the item arrives as a strengthen recommendation pointing at that URL instead of a duplicate new-page idea.
Briefs and drafts
Each queue item expands into a brief and an editable draft. Drafts are grounded in your audit evidence: the prompts the piece is meant to win, and the kind of sources the engines currently cite in those answers. Formats follow your business type: a B2B SaaS project gets comparison and evaluation formats, a D2C project gets formats that fit consumer buying questions.
For execution-heavy formats, ROI calculators, webinars, video scripts, the deliverable is build instructions plus the citable page copy. Drafts never invent numbers, customer names, quotes or outcomes: anything that must come from you is a clearly marked placeholder.
How the grounding and review layers keep drafts from reading like generic AI content is covered in depth in Why Cited Drafts Are Not Generic AI Content.
Drafts are starting points
Publish drafts through your own review. They are written to be edited: your voice, your proof points and your claims are what make the page yours.
The time math: an analyst-week in about 20 minutes
The workflow exists because the manual version does not survive contact with a calendar. What each stage replaces:
The compressed version is roughly 20 minutes of review at each weekly touchpoint: read the movement, open the top queue item, edit and ship. Teams describe the difference as one workflow replacing a week of scattered analyst work, which is what makes a weekly cadence sustainable for a team of one.
| Stage | Done by hand | Inside Cited |
|---|---|---|
| Collect answers | Ask each engine each prompt, paste answers into a sheet: hours per engine, repeated weekly | Every audit, automatically, stored verbatim |
| Diagnose losses | Read answers side by side, guess why rivals won | Classified per answer with the evidence linked |
| Prioritize | Debate in a meeting | Queue ranked by buying conversations at stake |
| Draft | Brief a writer, wait, revise | Grounded draft ready to edit |
| Verify | Rarely done at all | Next audits re-ask the same prompts |
Publishing and verification
Mark items done as you publish. The next audits re-ask the same prompts, and the trend view shows whether the answers picked up your fix: share of voice, endorsement and citations, before and after. Content that did not move the answers is visible too, which is what lets you iterate instead of guess.
AI-drafted assets are metered per plan: 10 per month on Starter, 40 on Pro, 75 on Agency Pro. Diagnosis and the queue itself are not metered.
Related docs