Visibility
Prompt suggestions
Where suggested prompts come from, including persona-tagged ones, how auto refill behaves, and where the retired Search Console opportunities went.

Where suggestions come from
Prompts → Suggestions is a
queue of prompts nobody typed in by hand. Every suggestion on it was written by a
research run: DiscoveredBy renders your homepage, has a model pick a handful
of other pages worth reading, scrapes those, and builds a short profile of
the business from what it found. A separate model call then writes a batch of
prompts, and it is told to write them the way buyers actually type: short
searches, plain one-sentence questions, and a few casual sentences describing
a situation, in a buyer's words rather than your site's own service names
(api/services/prompt_style.py#BUYER_STYLE). It is also told to name a place
only where a buyer would, such as "near me" for a local service, because the
country a prompt is tracked in is chosen separately when you track it. These
are instructions to a model, so check each suggestion before you track it.
That call is also handed up to 25 of your most-shown Search Console queries
from the last 28 days, wherever your site ranks for them, each with the page
Google showed most for it
(api/services/prompt_demand.py#top_demand_queries_for_project). Searches for
your own brand names, "site:" searches, and queries one of your active tracked
prompts already covers are left out. The model is told to write at least one
suggestion for each buyer topic in that list your business offers, more for
the most-searched ones, and to cover the most-searched topics first when there
are more topics than suggestions in one run. A short suggestion can be one of
those searches exactly as typed. Nothing moves a Google query onto this screen
by itself: the model decides which searches to write prompts for. (One
exception is a query pulled in from an AI engine's own answer, covered on a
separate page.)
The research pass also reads the project's saved description and business
goal from Project settings. The description helps
focus suggestions on your offering and audience; the goal helps prioritize
the buyer questions that matter to your business. Suggestions must still be
supported by the website and research profile. Empty fields do not prevent
generation. Save any edits before requesting more suggestions: existing
cards are not rewritten when settings change (api/services/research.py#prompt_generation_prompt,
#run_pipeline). This applies to initial discovery, Suggest more, automatic
refills, and scheduled research refreshes.
Suggestions are listed under the theme a run assigned them, and each row
shows the intent and buying stage it was given, plus its Google demand
figure once Search Console has data. The list is ordered by a research score
built from those same three labels (api/services/research.py#_score_candidates),
so the theme holding the highest-ranked suggestion comes first. The score
itself is not shown, because it only restates the labels.
Open a row (its text, or the arrow at its end) to see why it was suggested:
the run's one-sentence rationale, the facts from your site, business profile
or project details it rests on, the Search Console queries it covers, what to
publish if AI answers leave you out, and how and when it was found
(frontend/src/routes/(app)/prompts/suggestions/SuggestionRow.svelte).
Research also attaches a few lines to every suggestion that restate its
intent and stage, or read the same on every row; those are left out
(frontend/src/lib/prompt-suggestions.js#realEvidence). A row flagged "Close
to a prompt you track" is close enough to something you already track that
tracking it too would likely duplicate a measurement rather than add a new
one; see Prompts for what a tracked prompt is.
The flag comes with the suggestion itself, so it stays when your tracked
prompts cannot be loaded; the row then links to the tracked prompt without
its wording, and a note above the list says why
(frontend/src/routes/(app)/prompts/suggestions/+page.svelte#trackedUnknown).
A suggestion can also carry a persona tag. Those come from a different
source: an active persona's own Suggest
prompts request, not this screen's research run, and tracking one creates
that persona's targets rather than General ones. Once any pending suggestion
carries a persona, an audience filter narrows the queue to one persona,
General suggestions only, or everything. It sits beside a search box and
theme, buying stage and intent filters, each of which lists only the values
present in the queue, with their counts
(frontend/src/routes/(app)/prompts/suggestions/+page.svelte#personaChoices).
When the queue is empty
The queue only lists suggestions still waiting on a decision, and the Suggestions tab on Prompts shows a count beside its name while any are pending (no count when there are none). If nothing is pending, the reason shown depends on what the most recent research run did: none has run yet for a brand new project, one is reading your site right now, the last one could not reach your site at all, the last one failed outright, or the last one finished without producing anything usable. Only the "could not reach" and "failed" cases offer a retry button; the others just wait for the next run.
A queue that could not be loaded is not an empty one. When the suggestions
themselves fail to load, the page shows an error with Retry in place of
the queue; when the queue is empty and the research status could not be read,
it says it cannot tell whether a set is on its way, again with Retry. Neither claims your
first set is being prepared, and this Retry only reloads the page; it does not
start a research run
(frontend/src/routes/(app)/prompts/suggestions/+page.server.js#load).
Auto refill and its backoff
Beyond your own actions, the pending queue tops itself up on its own once it runs low, without you asking. That top-up is skipped while a research run for the project is already in progress, and after an unproductive run (one that failed, or finished without producing a single new suggestion) it waits before trying again rather than retrying immediately: the wait starts short and roughly doubles after every unproductive run in a row, up to a cap of about a day. After several unproductive runs in a row it stops trying on its own, and only a run you start yourself brings it back, through whichever of Try again or Suggest more the screen currently offers. This automatic top-up is separate from a plan's own on-demand and weekly-refresh entitlements; see Pricing for what your plan includes there.
Tracking or dismissing a suggestion
Tracking a suggestion turns it into a real tracked prompt, in the
suggestion's own country when the project covers it and otherwise in every
country the project covers
(api/services/prompt_discovery.py#_resolve_accept_countries), and
immediately queues a fast first
run outside the daily cycle, the same exception
Prompts describes for
accepting a suggestion. It also counts against your prompt slots the same
way adding one by hand does; see Pricing for what each plan
includes. A row warns you when tracking it would need more room than you
have free, but that warning by itself does not disable the button: click it
anyway and the request still goes to the server, which rejects it with an
error message instead of creating a prompt if there truly is no room. Your
role matters separately: for a viewer the page hides Track, Dismiss and the
selection boxes, with a role notice above the queue explaining why
(frontend/src/routes/(app)/prompts/suggestions/+page.svelte#writable).
To decide several at once, tick their rows, or a theme's own box to take
every row shown under it, and use the bar that appears at the foot of the
screen. It sends them one at a time, so each slot check counts the ones
tracked before it (frontend/src/routes/(app)/prompts/suggestions/+page.svelte#decideCandidates).
The bar also says how many prompt slots the selection uses at most: tracking
skips any country the prompt is already active in, so the real number can
be lower (api/routers/prompt_candidates.py#_count_new_or_reactivated_targets).
Dismissing a suggestion sets it aside; it drops off this queue and will not resurface here on its own. Both tracking and dismissing can trigger the auto-refill top-up described above, so acting on a suggestion can quietly bring a fresh one in behind it.
Search Console opportunities
The Opportunities tab that used to sit beside Suggestions listed pages of yours that ranked well on Google for a search matching a tracked prompt while AI engines rarely cited you for that prompt. It was retired on 13 October 2026, and its rows and dismissals were removed with it. A link to it now opens the Search plan, on the Plan tab, which groups your Google searches and tracked prompts into topics and shows, for each topic, how you do on Google and in AI answers and, where there is one, the next step.
The Search plan does not write suggestions. Its Track a prompt button opens the add dialog on Prompts with the topic's search term filled in, and nothing reaches this queue from it.
Related
- Prompts: what a tracked prompt is, and how a tracked prompt actually gets run
- Search plan: your Google searches and tracked prompts grouped into topics, which replaced Search Console opportunities
- Metrics defined: every project-wide number's exact formula and denominator
- Troubleshooting: what to check when citations are not showing up
- Pricing: what each plan includes for prompt slots and on-demand suggestion generation
Last verified 2026-10-14