Ask AI (natural-language query)

Spectry offers two ways to ask questions in plain language.

Ask AI

From AI insights, ask context-aware questions about your site and get answers grounded in your data. Examples:

• "What are the most critical issues on my site?"
• "How can I improve my homepage conversion rate?"
• "Why is my checkout page slow?"

It fetches data as it reasons

Ask AI doesn't answer from a fixed snapshot. It decides which data it needs, fetches it, and can go back for more — up to three rounds — before replying. It can pull traffic, top errors, web vitals, page content, detected insights, heatmap summaries, funnels, bounce rates, user flow, engagement, frustration signals, trend comparisons, custom events, feedback responses and survey answers.

Two practical consequences: a broad question ("how is my site doing?") costs more rounds and returns a broader answer, while a specific one ("which pages have the worst INP on mobile this month?") gets a sharper reply. And it can only answer from data you actually collect — ask about heatmaps on a site where the heatmaps consent category is never granted and it will correctly tell you there's nothing there.

Conversations are kept

Threads persist, so you can come back to one and keep asking follow-ups with the earlier context intact rather than restating everything.

Ask your data (natural-language query)

The Ask Your Data page turns a plain-English question into a database query, runs it, and returns the result as a table. Examples:

• "Show me page views per day for the last 7 days."
• "What are the top 10 pages by unique visitors?"
• "How many sessions came from mobile devices?"

Expand Show SQL query to see exactly what was run, and the result shows how many rows matched.

Read the SQL before you quote the number. A plain-English question can be interpreted more than one way — "visitors last week" might mean sessions or unique visitors, and last week might mean the last seven days or the previous calendar week. The generated query tells you which one you got. For figures you'll repeat to other people, cross-check against the analytics hub, which uses the canonical definitions.

Which one to use

• Ask AI for "why" and "what should I do" — it reasons across sources and cites what it looked at.
• Ask Your Data for "how many" — it returns an exact table you can check.