Skip to main content

When to use this

Use this playbook when a buyer question you track gets answered without you, and a competitor is named instead. That is the most valuable gap you can have. The engine has already shown it will recommend a vendor for this question, so the work is to become the better source, not to create demand. A prompt where nobody is named is uncontested but slower to win, because there is no proof yet that engines name vendors for it. You’ll need: edit access to the brand, at least one complete tracking run with competitors configured, and agent credits for the draft. Viewers can read every screen here but cannot use Quotable Agent.

Steps

1

Pick the prompt

In the sidebar’s Actions section, open Opportunities. Under Increase content visibility, each item reads Create content to win “prompt”. Start with the ones marked High impact: those are prompts where a competitor was named in an answer that left you out.Open an item. What this is based on lists Engines missing you, Cited instead (the brands named where you were absent) and the prompt itself.You can also start from the data. In the Analytics section, open Prompts, sort by Visibility and start at the bottom. Add the Brands column from Filters to see which tracked competitors appear in each prompt’s latest answers.
2

Read the answers

Open the prompt from Prompts. On the Responses tab, open each engine’s answer and read it in full. Note three things:
  • Which brands are named, and in what order.
  • What the engine says about the winner: specifics, numbers, a comparison, a use case.
  • Which pages it cited. The Citations tab lists them for the whole prompt.
The Searches tab shows the searches engines ran to answer. Those queries tell you which words your page has to answer, not only the prompt text.
If the cited pages are lists or reviews you are not in, the faster win may be getting onto those pages. See Close a source gap.
3

Draft the page with Quotable Agent

Back on the Opportunities item, click Write with the agent. Quotable Agent opens with a request to write an article that wins the prompt. Before you send it, add what only you know: the angle, the proof you can share, the competitor the page should be compared with, and the length you want.If the request is short, Quotable Agent may ask up to four questions first. Pick answers, type your own, or click Skip all to let it decide.The finished draft is saved as an editable document in the Library and opens beside the chat. Ask for changes in the same chat and it revises the document in place.
4

Make the page answer the question

Edit the draft before you publish. Check that:
  • The opening lines answer the prompt directly, with your brand named.
  • Every claim is true and specific. Replace vague praise with numbers, plans, limits and named use cases.
  • It covers what the winning brand is credited for in the answers you read.
  • Your brand facts match. Quotable Agent writes from your approved brand context; fix a wrong record there, not only in the draft.
5

Publish it on your site

Publish the page on your own domain from your CMS.On the Opportunities item, set Status to Done. Done on records the day it went live; change it if you published earlier. Impact marks that day on your charts.
6

Keep tracking the prompt

Do not edit or delete the prompt. Tracking keeps asking it every day on every engine the brand tracks, so later runs show whether engines start to name you. Deleting a prompt deletes its history.

How to tell it worked

Watch the prompt itself first, then the brand. Give it time and read it honestly:
  • One answer is one sample. The same prompt can name different brands from one day to the next. One good run is not a win, and one bad run is not a loss. Read several runs. On the Overview, set Interval to Weekly in Filters.
  • Engines that search the web move first. An engine that answers from memory changes slowly. On Citations, Citations flow shows the web search rate per engine.
  • Brand-level numbers move less. One prompt is a small share of your set, so your overall visibility score may barely change. Judge the prompt, not the dashboard.
See When numbers shift for every reason a number moves without the engines changing.

Automate it

From Routines, start from the Competitor watch template. Every Thursday it ranks tracked competitors by change in share of voice, lists the prompts where the top two now appear and you do not, and the pages cited in those answers. That list is your next batch of prompts for this playbook. The Monthly opportunities review template checks, on the 1st, whether the prompts behind items marked done moved since, and says when a change is too small or too recent to attribute. See Routines.

Next steps

Close a source gap

Get onto the pages engines already cite for the prompt.

Prompts

Read every engine’s answer, citation and search.

Visibility score

How mentions and position combine into one number.
Last modified on November 19, 2026