Asking questions - Help Center

Not everything you ask the agent is a request to change something. A lot of it is just a question — and it answers those from your real data, not from guesswork.

How your sends performed

"How did my last newsletter do?"

"What were my top performing emails this quarter?"

"Which links got the most clicks last month?"

"Where are my readers opening from?"

"Which audience engages the most?"

"How's my deliverability looking — any bounce problems?"

"How much revenue did email drive in the last 90 days?"

Answers come with a report card that deep-links straight into your sending history, so you can go from a one-line answer to the full report in a click.

The numbers are calculated exactly the way the reports page calculates them — same denominators, same bot-open exclusions — so what the agent tells you and what your report shows will always agree. Revenue answers need Stripe or Shopify connected.

What a contact has been doing

"What's the story with sam@example.com?"

"Has jo@example.com opened anything recently?"

You get their profile, their engagement counters and their recent activity, with a link through to them on your People page.

Ask by email address when you can. A name is ambiguous — two Sams is one Sam too many — so if there's any doubt the agent asks for the address rather than guessing at the wrong person.

How something in Audienceful works

The agent can search this help center and answer product and how-to questions directly in chat, with links to the relevant articles.

"How do I set up a custom domain?"

"What's the difference between an audience and a tag?"

"How do I stop bots signing up through my form?"

It answers from the docs rather than from memory, so if something isn't covered it says so instead of inventing an answer — and points you to support@audienceful.com when a question genuinely needs a human.

Research on the open web

Plenty of what you need to know isn't in Audienceful at all. Ask anyway — the agent can search the web, open the top results and read them before it answers.

"What's changed with Gmail's bulk sender requirements this year?"

"What did our competitors announce this quarter?"

"What are people writing about our industry this week?"

"Read https://example.com/their-pricing and tell me how it compares to ours."

Answers come with a "Searched the web for…" row underneath listing every source it used, so you can verify anything before you act on it. It reads the actual pages rather than working from search snippets, which is the difference between a vague summary and something specific enough to use.

The agent reaches for search on its own when a question clearly needs current information — anything about "this week", "the latest", or "what's new" is researched rather than answered from memory. If a first search comes back empty, it broadens the query and tries again instead of giving up.

And research doesn't have to end at an answer. The same capability feeds straight into a draft: "…and write it up as this week's newsletter." See writing from the web and your own site.

What the agent can and can't see

Only your workspace. Every lookup is scoped to the workspace you're working in. It cannot see another customer's data, and it can't see other workspaces of yours.
Only what's really there. Reporting answers are built from your stored send data; contact answers from that contact's record. The agent won't fabricate a number, a contact, or a step in an automation it hasn't loaded.
The open web, when you ask for it. Search results and pages it fetches are reference material for that one request — read, used, and not kept.
Nothing is retained by the AI models. Requests are sent with data collection denied, so your prompts, emails and contact data aren't kept by the providers we route through and aren't used to train them.

As always: the agent can make mistakes. For anything you're about to act on, the report and the contact record are the source of truth — and both are a click away from its answer.