picked

Documentation

The dashboard

The sidebar has three groups: Monitor (Prompts, Analytics, Reports), Research (Content Gaps, Competitors) and Create (Create Content), then Settings and Support. This page says what each screen is for, so you know where to look.

LaunchOnIt's dashboard after its first run, October 5, 2026

The screens at a glance

ScreenThe question it answers
PromptsHow often do the AI models name me, prompt by prompt, and which sites do they cite?
AnalyticsWhich of my pages do AI crawlers read, and who sends people from an answer?
ReportsWhat do I send the client this month? Scale only
Content GapsWhat should I write, and where should I get listed?
CompetitorsWho is beating me, and where?
Create ContentCan Picked write the page that should have won? Growth and Scale
SettingsHow do I change models, brand, team or plan?

The filters bar

Prompts, Analytics and Competitors share one bar at the top:

  • Topic. All topics, or one group of prompts. Prompts only: Analytics and Competitors cover every prompt.
  • Date range. Pick a start and an end. Every number on the page covers that range, and changes are against the same number of days before it. Pages open on the last 7 days.

The filters are in the page's address, so a link you send opens with the same view.

Prompts

Your home screen. At the top, two cards:

  • Visibility score: the share of answers in the range that name you, with its change and a line by day.
  • Visibility score rank: where you stand among every brand the answers name, like "#4 of 31 brands named", and the brands around you with their score and sentiment: how the answers present each one, 0 to 100. It fills in after the first run.

Below, one row per prompt:

ColumnWhat it shows
PromptThe question, with its topic
Avg. VisibilityThe share of its answers that name you, and the same per AI model
Avg. PositionYour average place among the brands an answer names. 1 is named first
Brand mentionsThe brands named in its answers, most often first
DemandGoogle searches a month in the US for the closest search

Search the list, or select prompts to move them to a topic or remove them. Add prompts takes one per line and lists ideas from your own data. Run now asks the new prompts and models, and any answer that failed today; the rest waits for the daily run.

A prompt's page

Click a row to open it. At the top, the same two cards for this prompt alone: where you rank among the brands its answers name, and your score by day. Below them, Sources lists every page its answers cited and how often, and Run history has one row per answer: the AI model, whether it named you, how many brands and sources, and when.

Click an answer in Run history to read it. It opens on the side, in three tabs:

  • Overview: the brands in the answer, in order, each with the line the model wrote about it, and the searches the model ran before answering, when it shows them.
  • Answer: the answer as the model wrote it, with its links.
  • Sources: the pages that answer cited, and whether one is yours.

The address of an open prompt, and of an open answer, can be sent to a colleague or a client.

Cited sources

Under the prompts, the sites the AI answers link for your prompts. Citations counts them per AI model, Citations by day splits each day into your site and everyone else (hover a day for the numbers), and Cited domains lists every domain the answers link, with how many answers cited it and whether those answers name you. Only where you're missing keeps the domains whose answers never name you: the places to get listed. Click a domain to see its pages and the prompts they're cited for.

LaunchOnIt's prompts, October 10 to 16, 2026

Analytics

What the AI models do on your site, from your server's own traffic. It needs a connection (Cloudflare, Vercel or a snippet on your server): see AI crawlers for each one and for what the page shows.

  • AI visits: every request an AI crawler made in the range, split by AI model (ChatGPT, Claude, Gemini, Perplexity, Meta AI, Grok, Petal Search and the rest) and by why it came: a live answer, an agent, a search index or training. All AI models and All purposes narrow the whole page.
  • AI visit activity: visits by day. Click a day for its hours and the bots that came.
  • Pages AI read: your pages by visits, with What AI reads (which kinds of page get the visits) and the marks of the AI models that read each one.
  • Recent visits: every visit, newest first, a page at a time. With Cloudflare it refreshes about once a minute while the page is open.
  • Visits from AI answers: people who clicked through from an assistant to your site, by AI model and page.
  • Turned away: the requests your site blocked or answered with an error, by page.

Where an answer's text is cited, not crawled, it's under Prompts: see Cited sources.

launchon.it through Cloudflare, September 17 to October 16, 2026

Content Gaps

Three tabs, each a different way to win a prompt.

Content Gaps

The prompts where the AI models name other brands and never you. Each row shows the brands named instead, the AI models you're missing on, and the demand. Sort by Best opportunity (many brands named, high demand) and start with the top. "Write the page" starts an article.

Get listed

The pages the AI models cite in answers that name your competitors and not you: roundups, alternatives pages, review sites, directories and threads. Each row has the answers that cite it, the brands those answers name (Names instead: what the answers recommend, not necessarily what the page lists) and your prompts it's cited for. Getting onto one of these pages is often quicker than writing your own.

Click a page and Draft a pitch. In under a minute Picked:

  1. Reads the page. What kind of page it is (a list, an alternatives page, a review, a thread), how it writes each entry, and which products it already lists. If it already names you, there's nothing to ask and the pitch doesn't count.
  2. Finds who runs it, from what the page itself publishes: the author, an email written on the page or its contact page, a "submit a tool" or "write for us" page, the author's X or LinkedIn. Nothing is guessed. When the page publishes no contact, the panel says so and links the site.
  3. Writes the ask. A short email to that person, with your entry written the way the page writes its entries and your name at the bottom, ready to send. The entry is also on its own, for sites that take submissions through a form. The facts about you come only from your site and from Settings > Brand, so fill that in for sharper pitches. In a thread it writes a reply instead, for you to post from your own account with a line that says you run the product.

Edit anything, then Copy email or Open in email: Picked never sends anything itself. Set the status as you go (Drafted, Sent, Replied, Listed, Declined) and filter the table by it.

Every week Picked reads the pages you marked Sent or Replied again. Once one names you, it turns Listed, the company's owners and admins get an email, and the panel shows how many of the answers citing that page name you since, against the 30 days before.

Sometimes the AI cites a whole site, not one page on it: a homepage like saashub.com/ or uneed.best/, usually a directory or a launch platform. There's no author to pitch there, so the panel links the site's submit page instead (or its contact page, or the site), and no pitch is spent.

Pitches are a monthly allowance on every plan; the pricing page has the numbers. A page Picked can't open, one that isn't about your category, or one that already lists you gives the pitch back. Draft again writes a new version and counts as a pitch.

Keyword Gaps

Google keywords your competitors rank for in the top 20 and you don't. The scan needs at least one competitor with a website. It runs on its own within a day of adding one, or straight away with Scan keywords, then refreshes every month: Google's rankings data changes about that often, so a sooner scan would show the same keywords. A competitor added in between shows up at the next scan.

  • Searches a buyer wouldn't make are hidden as off-topic; Show off-topic brings them back.
  • Filter by volume, difficulty, where the competitor ranks, or brand names in the keyword.
  • Track as prompt turns a keyword into the question a buyer would ask an AI, and tracks it.

Click a row for the full picture: every tracked brand that ranks for it, the 12-month trend, the cost per click, and Google's page for it now, with the AI Overview and the sites it cites.

Competitors

Every brand you track, ranked by how often the answers name it. Add competitor takes a name, a website and other names (counted as the same brand). Track on pins a brand to one topic, or tracks it on every prompt.

LaunchOnIt, October 10 to 16, 2026

Under the list, Named by the AI models, not tracked yet shows the brands the answers keep naming that you haven't added. Track brand adds one.

Click a brand for its own page: visibility by day, by AI model, its average position, and Head to head, which splits the answers into you first, both named, neither, and the ones that name it and not you. Those last ones are where to get listed.

Reports

Scale only. A client report as a PDF in your agency's name. See reports for clients.

Create Content

Where Picked writes the page that should have won the answer. Growth and Scale only. See write the article.

Settings

Settings are split in two.

GroupPages
Project settingsGeneral (name, website, other names), AI models, Brand (facts, colour and product screenshots), Publishing (where articles go)
Company settingsCompany, Members, Report branding, Billing & usage

Project settings change one brand. Company settings change the whole account. See team, projects and plans.

Support

Support in the sidebar sends a request to contact@picked.so, and the answer comes to your email.

Next

Reading your results explains what the numbers on these screens mean.

Updated

Find out what ChatGPT and Google AI say about you

Add your domain and the prompts your buyers ask. Unlimited seats for your whole team, and no promises we can't keep.