Getting the most from your runs

Last updated October 29, 2026

Let the weekly schedule do the work

Your first run starts on its own once your project has a plan, prompts, and competitors. It doesn't use up an on-demand run.
Every project is measured weekly, on the same day each week, without you doing anything. Scheduled runs never use up your on-demand runs.
AI answers change slowly. Engines pick up new pages and reviews over days to weeks, not hours, so a weekly run catches real movement.
A trend needs several weeks of runs. One week to the next is often noise; three or four weeks in the same direction is a signal.

Situation	Run now?
Right after setup	No — your first run starts on its own.
You changed prompts, competitors, or your brand	Only if you can't wait for the next weekly run, which picks them up. Make changes in a batch, then run once.
A run failed or came back incomplete	Yes — and a failed run doesn't use up your on-demand runs.
Before a client meeting or report	Only if the latest run is several days old.
You published a fix and want to see if it worked	No — engines haven't picked it up yet. Watch the next two to four weekly runs.
Checking for movement day to day	No — a same-day re-run mostly shows noise.

Your organization gets on-demand runs worth 4 full runs of your plan per billing period — 4 × your plan's prompts, shared across your projects. A run uses as many as its project has, so a small project can run more often than a large one. During a free trial it's 4 full runs (9 during a design partner's free month). It resets when the next period starts.
At most 3 runs of one project in any 24 hours.
Failed runs don't use up your on-demand runs, and neither do scheduled runs or your first run. A failed run that still got answers back does count toward the 24-hour limit.
The Run now button shows how many more runs of that project you have left.

Day 1: check your prompts and competitors. Your first run starts on its own, and results take from minutes to a few hours.
After your first results: drop prompts that don't reflect how your customers ask, add the competitors that keep showing up, then Run now once more.
Weeks 1 and 2: two scheduled runs arrive on their own, giving you a first trend. Start on the top items in the Fix-It Playbook.

Every score has a margin of error. A change smaller than the margin isn't a real change — see What the ± margin means.
Compare runs on your project's Runs History page puts any two runs side by side and only calls a change real when it clears the margin.