Recommendations are one type of AI insight. Where an anomaly tells you what changed, a recommendation tells you what to consider doing about your analytics, your site, or your traffic.
How recommendations fit in
Zenovay's AI Insights feature generates four kinds of insight:
| Type | What it tells you |
|---|---|
| Anomaly | A metric moved sharply away from its expected value |
| Trend | A metric is drifting in one direction over time |
| Recommendation | A concrete, actionable suggestion based on your data |
| Prediction | A projection of where a metric is heading |
Recommendations are the type covered in this article. They read like a suggestion rather than an alert, for example:
- "Mobile conversion is well below desktop. Worth investigating the mobile checkout flow."
- "Traffic from LinkedIn engages strongly. Consider investing more there."
- "Most key actions aren't tracked as goals yet. Adding goals would sharpen your reporting."
Info
AI Insights, including recommendations, is available on the Pro plan and above.
Where recommendations come from
Recommendations are generated automatically from your own analytics. The AI looks at things like:
- Conversion and engagement gaps (e.g. mobile vs. desktop)
- Traffic sources that convert or engage unusually well
- Metrics that are trending in a notable direction
- Tracking gaps (e.g. high-intent actions with no goal configured)
You don't trigger them manually. New insights appear on a schedule plus event-driven triggers. See Insight frequency for the exact cadence.
Viewing recommendations
There are two places to see insights, including recommendations:
The Insights overview (all websites)
- Open Insights from the sidebar (
app.zenovay.com/insights). - You'll see a team-wide summary: total insights, plus counts by severity, and recent alerts across your sites.
- Click into a website to view its insights in detail.
A single website's Insights tab
- Open the website from Websites (
/domains). - Select the Insights tab.
- Use the Type filter and choose Recommendations to show only recommendation-type insights.
- Click any card to open the full detail view.
You can also filter by severity and search by keyword.
Severity
Each insight carries a severity that signals how much attention it warrants:
| Severity | Meaning |
|---|---|
| Critical | Significant deviation, worth looking at now |
| Warning | Notable, worth reviewing soon |
| Info | Useful context, no urgency |
What's inside a recommendation
Open any insight to see its detail view:
- Title and type — a short summary and the insight type (recommendation, anomaly, etc.)
- Description — what the AI observed and why it matters
- Recommendation — for recommendation-type insights, the suggested action is highlighted
- Metrics — the current value, the expected value, and the deviation
- History chart — how the underlying metric has moved over recent days
- Detected — when the insight was generated
Use the copy button in the detail view to copy the insight's text to your clipboard, for example to paste into a task tracker or a message to a teammate.
Acting on a recommendation
There is no separate "implement / schedule / investigate" workflow. The workflow is intentionally simple: read the recommendation, decide whether it's relevant, and then keep your feed tidy with two actions.
Acknowledge
Mark a recommendation as reviewed. Acknowledged insights are hidden from the default view (you can show them again with the Show acknowledged toggle). This is the way to clear an item you've read and handled.
Dismiss
Remove a recommendation that isn't relevant to you. Dismissing it takes it out of your list.
Both actions are available from the insight card and the detail view. They require editor access to the website; viewers see the buttons disabled.
For more on this workflow, see Acknowledging insights.
Examples of recommendation insights
Recommendations are generated from your real data, so they're specific to your site. Common themes include:
E-commerce
| Pattern in the data | Recommendation |
|---|---|
| Mobile converts far below desktop | Review the mobile checkout flow |
| High bounce on key landing pages | Strengthen those landing pages |
| Strong-converting referral source | Lean into that channel |
SaaS
| Pattern in the data | Recommendation |
|---|---|
| Low conversion on the trial/signup path | Improve onboarding |
| Pricing page is a frequent exit point | Clarify the value on that page |
Content sites
| Pattern in the data | Recommendation |
|---|---|
| Short sessions | Surface related content |
| High bounce on entry pages | Revisit headlines and intros |
Tips
- Filter to recommendations when you want suggestions rather than alerts. Switch the Type filter to Recommendations on a website's Insights tab.
- Acknowledge as you go so your feed reflects only what still needs a look.
- Enable the weekly digest to get a regular email summary of insights instead of checking in manually. See Weekly digest.
Troubleshooting
I don't see any recommendations
- Recommendations only appear once there's enough data to find a meaningful pattern. New sites may show few or none at first.
- Confirm you're on the Pro plan or above, where AI Insights is available.
- Check the Type filter isn't excluding recommendations, and that Show acknowledged is on if you've already acknowledged everything.
The same insight keeps coming back
- Acknowledge or dismiss it. The AI generates fresh insights on its schedule, and handled items stay out of your default view.
A recommendation isn't useful
- Dismiss it to remove it from your list. Improving your tracking setup (adding goals, reducing excluded traffic) gives the AI better signal to work from.