To-dos (action plan)

Turn measured visibility gaps into concrete, evidence-backed actions with the To-dos rule engine.

2 minute read · Last reviewed: August 14, 2026

Knowing you're behind on a prompt is one thing. Knowing exactly what to publish, where to comment, and which link to go earn is another. That's what the To-dos page is for, previously called Gap Analysis (both names point to the same feature, and the old URL still works).

What counts as a gap

A gap is a prompt or topic where AI cites or favors a competitor but not you, backed by real tracked data from your account. This isn't a generic content checklist, it's built from what's actually happening in your specific tracked prompts.

Your gap list reads like a work queue, not a report: each item is a concrete action, tagged by priority, page area, type, the AI engines it moves, and the effort it takes.

A deterministic rule engine, not free-form AI

Recommendations come from a rule engine, not from an AI model improvising suggestions. Every rule only fires when it finds a real, measured signal behind it, and every example URL shown to you is checked against actual cited sources before it's displayed. If a rule can't find a real URL to back up its suggestion, it's dropped rather than shown with a made-up example.

This matters because an earlier version of this feature used AI to generate suggestions and it sometimes fabricated example URLs that didn't actually exist. That version was replaced with the current, grounded engine specifically because of that problem. If you used this feature a while ago and remember generic or vague suggestions, it's worth another look, since the current version is built to only recommend things it can actually point to real evidence for.

Example recommendation types include publishing a "[Brand] vs [Competitor]" comparison page, creating a page for a question where you never appear, publishing a "[Competitor] alternatives" page, getting into Reddit threads AI already quotes, making a video for a question AI answers with video, publishing a "best [category]" list, adding FAQ schema, or earning links from sites AI already cites.

The "Why" panel

Every recommendation includes a "Why" panel showing the reasoning behind it, a visual bar comparing your visibility to the competitor's on that specific gap, the exact prompts where you're invisible, real reference URLs to look at, and an estimated impact if you act on it.

Working through your list

You can generate new recommendations on demand, scoped to a specific AI model or focus area (content, technical, competitive, backlinks, Reddit, YouTube, social, or press). Items move through a simple workflow: Suggested → To-do → Completed → Archive. You can assign items to a teammate, set due dates, and check off sub-steps as you complete them. If you dismiss something, it won't silently reappear the next time you regenerate recommendations.

You can also add your own manual to-dos alongside the generated ones, and export your whole list as CSV or JSON.