AI submissions analysis goes beyond basic metrics by using artificial intelligence to find patterns, surface insights, and give you specific, actionable recommendations for improving your forms. Instead of manually reviewing hundreds of submissions looking for trends, you let AI do the heavy lifting and present you with a clear summary of what is working, what is not, and what to change.
This is especially valuable for high-volume forms where manually reading every submission is impractical. A form that receives 200 submissions a month contains insights you would never spot by skimming — the AI finds them.
How to Run an AI Analysis
- Go to Form Forge > Submissions and choose the form you want to analyze. The AI Analysis controls live here, alongside the submissions table — this is the primary entry point, because the feature analyzes your collected submission and conversion data. If a cached result exists, View cached analysis opens it without spending credits. Run new analysis starts a fresh 5-credit job. The same controls are also available as a per-form action on Form Forge > Analytics and as a secondary shortcut on the form editor screen.
- Make sure the form has a reasonable number of submissions — at least 10 are required, 20 for meaningful results, and 50 or more is ideal.
- Click Run new analysis when you want a fresh result. The 5-credit cost is shown on the badge; you confirm the spend before the analysis starts. Use View cached analysis when you only need to reopen the last cached result.
- The analysis runs as a background job. The panel shows a clear “Analysis in progress…” state while the job runs, then the results panel slides in.
- When complete, the results show a quality score, fill-rate per field, drop-off patterns, sample counts, confidence, source labels, and actionable recommendations. The score label is separated from the metadata badges so cached/sample details stay readable. Field-specific recommendation buttons open the form editor on the target field, scroll it into view, add a neutral AI suggestion marker to the highlighted card, and show the AI note inside Field Settings so you can make the change without hunting through a long form. Cached results are labeled with the last analyzed time and cache expiry so you can tell whether you are looking at fresh data. If the job fails (for example, not enough data, or a temporary AI-service issue), the panel shows a clear error message instead of hanging on the loading state — and no credits are spent on a failed job.
What the AI Analyzes
| Analysis area | What AI looks for | Example finding |
|---|---|---|
| Fill rate by field | Which fields are being left empty most often | “The ‘Company Size’ field has only a 12% fill rate” |
| Drop-off patterns | On multi-step forms, which step loses the most people | “65% of visitors abandon on Step 3 (Payment Details)” |
| Content quality | Whether text field responses are meaningful or low-effort | “The ‘Feedback’ field gets one-word answers 40% of the time” |
| Submission timing | When most submissions come in during the day and week | “Peak submission time is Tuesday 10am-12pm” |
| Field redundancy | Whether certain fields consistently receive the same answer | “‘How did you hear about us?’ gets ‘Google’ 80% of the time” |
| Overall quality score | A summary score reflecting form performance across all dimensions | “Your form scores 72/100 — Good with room for improvement” |
Example Recommendations
The AI provides actionable suggestions, not just raw data. Here are examples of what you might see:
- “The ‘Company Size’ field has only a 12% fill rate. Consider making it a dropdown with ranges (1-10, 11-50, 51-200, 200+) instead of a free-text field, or removing it entirely if it is not essential.”
- “65% of visitors abandon the form on Step 3 (Payment Details). Consider adding trust signals like security badges, moving payment to the final step, or offering a free consultation option that does not require payment.”
- “The ‘How did you hear about us?’ field gets ‘N/A’ or ‘Google’ 80% of the time. The dropdown options may need updating to reflect your current marketing channels.”
- “Submissions peak on weekday mornings. Consider scheduling your team to review and follow up during these hours for the fastest response time.”
> Tip: Run AI analysis periodically — once a month for high-traffic forms, once a quarter for lower-traffic forms. The insights change as your audience, marketing, and submission volume evolve over time.
> Good to know: AI analysis uses the built-in Forge API, so you do not need your own AI API keys. The feature is included with your PRO license. Analysis runs in the background, uses completed submissions plus abandonment/payment-failure signals where available, and does not slow down your site or affect form performance for visitors.
> Applying recommendations: If you run analysis from Submissions or Analytics, use the recommendation CTA to jump straight into the form editor. When the insight targets a specific field, Form Forge opens that field’s settings, marks the card with an AI suggestion marker, and keeps the recommendation visible next to the controls. The CTA labels are intentionally short, such as Make optional or Simplify label. When you apply an editor-side recommendation, the button and Field Settings note confirm the concrete change, such as Changed required to optional. Form-level recommendations, such as splitting a form into steps, still open the editor so you can make the structural change manually.
Common Mistakes to Avoid
- Running analysis on a form with only 5 submissions. You need at least 20 for meaningful patterns. With fewer, the AI does not have enough data to draw reliable conclusions.
- Ignoring the recommendations. The AI gives you specific, actionable steps. If it says a field has a low fill rate, investigate whether you actually need that field.
- Running analysis too frequently. Monthly is ideal for most forms. Running it daily on the same data gives you the same results and wastes API credits.
[Screenshot: The AI analysis results panel showing a quality score, fill rate percentages for each field, a drop-off chart, and three actionable recommendations]
—