Fictura
Hawkings

Hawkings

The growth analyst built into the dashboard. It reads your live Fictura data, answers in plain English, and proposes changes it is never allowed to make itself.

Hawkings is the assistant in the dashboard: the reading-glasses pill in the bottom-right corner of every screen, or Ask Hawkings. It answers questions about your app using your actual event data rather than general advice, and it opens as a panel that pushes the page aside instead of covering it, so the numbers you are asking about stay on screen.

It knows which screen you are on. Asking "can I call a winner yet?" on Experiments and on Paywalls are different questions, and Hawkings reads the room before it reads the data.

What it can see

Hawkings calls the same read APIs the dashboard does, scoped to the app selected in the header:

It can readWhich answers
Headline metrics and the conversion funnelWhere users stop, and how much each step costs you
Events and your tracking planWhether something important is going unmeasured
Cohorts by billing outcomeWhat converters did that cancellers didn't
Paywalls and onboarding flows, with which is liveWhat to test, and against what
Experiment resultsWhether a test has actually resolved
Spend, campaigns and ROAS from paid adsWhich campaigns pay for themselves
App Store Connect subscription productsWhether a paywall references a product that exists
App identity: bundle ID, Team ID, Android package and fingerprints, store addresses, websiteWhether your links can open the app and reach the right store
These docs, including the dashboard mapHow a feature works, and where anything is in the dashboard

What it can and cannot change

Hawkings mostly reads. It can do a few setup tasks when you ask it to in chat, such as saving your app identity fields. It cannot ship a paywall, start an experiment, pause a campaign, or change a budget. When the data argues for a change it drafts a proposal and sends you to the page where you confirm it. The assistant that can move your revenue is deliberately never one sentence away from doing it.

Finding your way

Ask Hawkings where something is, like "where do I put my bundle ID?" or "how do I connect Postiz?". It searches these docs and the dashboard map, then tells you the page and the steps.

When it cannot confirm that something exists, it says so. It points you to the Help page instead of guessing.

Getting a good answer

Hawkings is built to refuse to guess, and that shapes how to ask:

  • It will tell you the sample is too small. Below roughly 100 users in a funnel step it says so rather than reading noise as signal. That is the correct answer, not a failure.
  • It separates measurement from estimate. Industry ranges are marked as rough figures; your numbers are never presented as anything but your numbers.
  • It distinguishes attributed from blended. On revenue questions, campaign-attributed metrics and blended totals are reported separately, and unattributed revenue is called out rather than quietly folded into a campaign.
  • Ask for the decision, not the dashboard. "What's the single highest-impact thing to fix?" gets further than "show me my metrics", because the numbers are already on the screen behind it.

Starting over

⋮ → New chat clears the thread and starts a fresh one. Conversations are kept per app and per person, so your threads are yours. A teammate on the same app does not see them.

Attach a PDF, CSV, or Markdown file with + and Hawkings will read it alongside your live data. See Attaching files for what it accepts and what happens to what you upload.

Answers can be wrong

Hawkings generates its answers from your event data, and generated answers can be mistaken, particularly on small samples or unusual instrumentation. Treat it as a fast analyst with access to your numbers, not as an auditor. Anything it proposes is worth the same sanity check you would give a colleague's suggestion.

On this page