Recommendations
How Pendium turns a visibility scan into a ranked, prioritized backlog of actions — and how to work them.
A scan tells you where you stand; Recommendations tell you what to do about it. After each visibility scan, Pendium turns the findings into a ranked backlog of concrete actions — content to create, sources to reach out to, and setup to finish — so you always know the highest-impact next move.
In the app it's the Recommendations item in the sidebar.
Where they come from
When a scan completes, Pendium generates strategic insights from the results, and each insight produces zero to a few recommended actions. Those become rows in your Recommendations backlog. Each one carries:
- A title and description (what to do and why it matters).
- A priority — high, medium, or low.
- A category — content, technical, or third-party (outreach).
- A status — pending, in progress, completed, or dismissed.
- Links to the queries it targets, and (for outreach actions) the source URL to reach out to.
The list is ordered by priority, highest first. Re-running scans refreshes the backlog without wiping your progress — an action you've already started stays where it is.
Foundations before tactics
Pendium won't bury you in content tasks before the groundwork is in place. The "Picked for you" card on your dashboard always surfaces the single most important next step, and it follows a strict order: crawlable → measurable → tactical.
In practice, that means it checks your setup from the foundation up and surfaces the first thing that's missing:
- Crawlable — do you have an agent site that's published, with a Profile and FAQ, ideally on your own domain?
- Measurable — is Google Search Console connected so you can prove the impact of your work?
- Ready — are your personas, goals, brand voice, and knowledge base in place?
- Tactical — only once the above is set does the dashboard surface the content- and outreach-level recommendations from your latest scan.
The logic: there's no point recommending a blog post to lift a query if AI can't crawl your site or you can't measure whether it worked. Fix the foundation first.
Working the backlog
- Mark progress — move an action to in progress, completed, or dismissed as you work it. Completing one keeps the queue honest about what's left.
- Turn one into content — many recommendations seed a content workflow directly, so you can go from "write a comparison guide" to a drafted post in a couple of clicks.
- Outreach actions point you at the source to contact — see Outreach.
- Re-scan after you ship to see the impact, and
get_scan_historytracks the trend.
Over MCP
An agent can pull the same backlog with get_recommendations (filter by status or priority) and mark items done with complete_action after it's addressed them.