Skip to content

Collect feedback with a form and act on it

Build one Probe database that pulls customer feedback into themes, priority and status, then shows what to act on now, starting from a written request.

6 min read
The By theme board showing feedback cards grouped into columns for pricing, onboarding, bugs and feature request

TL;DR

Walk through writing a request to Probe's Paper Agent that builds a feedback database with source, theme, priority and status fields, three views (All feedback, By theme, To act on), then add six example records to see it work. Includes exact credit costs (30 for the database, 43 for examples, 73 total) and screenshots of each step.

If feedback is scattered across email, calls, surveys, and support chat, you end up re-reading the same inbox every week trying to remember what customers actually asked for. This guide builds one Probe database that holds all of it, tagged by source, theme, priority, and status, with a view that shows only what needs action now. Building it costs 30 credits, adding six example records costs 43, for a total of 73.

Key takeaways

Open a new Paper, choose Doc, click Build with AI, and describe the feedback fields and views you want.
Review the proposed database card, then click Apply. Nothing is created until you do.
Open the database and check the three views: All feedback, By theme, and To act on.
Use By theme to spot patterns and To act on to work the queue.

Before you start

The agent builds the database fields and views from your written description of what you want.

What one piece of feedback should track

Each record needs enough on it that you don't have to open the original email or call note to know what to do with it. The database built here gives every piece of feedback six things: the customer, the source it came from, a theme, how many customers raised the same point, a priority, and a status. That's enough to sort, group, and filter without extra digging.

Where feedback comes from

Jotform notes that a good feedback form should be short and ask for exactly what's needed, and the same logic applies to a feedback record: one source field, not four different intake forms to maintain (jotform.com (opens another site)). Zendesk makes a similar point: a feedback form needs a clear purpose, which is also why this database keeps each record to a fixed set of fields rather than open-ended notes (zendesk.com (opens another site)).

Step: Build it with Build with AI

Start a new Paper, choose Doc, and title it "Customer feedback." Under the title, click Build with AI. A box opens asking "What is it for?" Type this request and press Enter:

"A place to collect customer feedback and act on it. Each piece of feedback has the customer, where it came from (email, call, survey, support chat), a theme (pricing, onboarding, bugs, feature request), how many customers raised it, a priority (high, medium, low) and a status (new, planned, done, won't do). I want a board by theme called By theme, a table of all feedback called All feedback, and a view of new or planned high-priority feedback called To act on."

The request typed into the Build with AI box before sending

This step costs 30 credits, charged as two parts of the agent working on the request (16, then 14).

The proposed database card showing the fields and three views before approval

You'll see a Review required card: "Create database Paper draft: Customer feedback," with a short description of the fields (source, theme, customer count, priority, status) and the three views (All feedback as a table, By theme as a board, To act on as a filtered view). Nothing is created yet. Click Apply.

The database applied, showing the empty table with its columns

After applying, the agent confirms: a Customer field as the title, Source, Theme, Customers raised (a number), Priority, and Status (New, Planned, Done, Won't do), with no records seeded yet, as asked.

Step: Add example feedback

In the same chat, type:

"Add six example pieces of feedback from early November 2026 across the sources and themes. Start each with 'Example:'."

This step costs 43 credits, charged across three parts of the agent working on the request (9, 18, and 16).

The six proposed example feedback records before approval

The proposal adds six records dated November 3–12, 2026, one for each combination of source and theme: email/pricing, call/onboarding, survey/bugs, support chat/feature request, email/onboarding, and call/pricing, with varied priority and status. Click Apply.

The six example records applied inside the database

The six records now appear: Dana Reyes, Marcus Lee, Priya Nair, Tom Okafor, Laila Haddad, and Jonas Weber, each prefixed "Example:" so they're easy to spot and delete later.

Step: See every piece of feedback at once

Open the All feedback view. This is a plain table over the database, listing every record.

The All feedback table listing all six example records with their fields

This is the view to scan when you want the complete picture rather than a filtered slice, say, during a weekly review.

Step: Build a board by theme

Open By theme. This is a board (kanban) view with one column per theme: pricing, onboarding, bugs, feature request. Each example record sits as a card in its theme's column.

The By theme board with cards grouped into columns by topic

Grouping by theme is what turns a pile of individual complaints into a pattern, pricing issues stacked against onboarding issues, without re-reading six separate emails to notice it.

Step: Build a To act on view

Open To act on. This view filters down to feedback marked new or planned and high priority, leaving out anything already done or marked won't do. It's the queue you work from, not the archive.

The To act on view showing only new or planned high-priority feedback

This is the view worth bookmarking or checking first thing: it's deliberately small, because it's meant to show only what's unresolved and matters.

What priority and status mean here

Priority (high, medium, low) is your call on how much a piece of feedback matters right now, informed by how many customers raised it. Status is one of new, planned, done, or won't do. Keeping status current is what keeps To act on showing the right records, since that view is built to filter on status and priority together.

What you end up with

One database, three views, holding every piece of customer feedback regardless of which channel it arrived through. You can open any record to add notes, since every row opens as its own page the first time you open it. From here, you'd delete the six "Example:" records and start adding real feedback as it comes in, through email, calls, surveys, or chat, tagging each with its source, theme, priority, and status as you go.

Questions people ask

How do I track customer feedback without a spreadsheet?

Build a database with fields for source, theme, priority, and status, then use a table view for the full list and a filtered view for what's urgent. Probe's Paper Agent can build this from a written request, covered in the steps above.

What fields should a customer feedback tracker have?

At minimum: the customer, where the feedback came from, a theme or category, how many people raised the same issue, a priority, and a status. These six fields are what the database in this guide uses.

How do I know which feedback to act on first?

Filter for high-priority items that are still new or planned, as the To act on view does here. Combining priority with status keeps the list to only what's unresolved and matters.

Can I group feedback by category automatically?

A board (kanban) view grouped by a theme field, like By theme in this guide, gives feedback a column per category, pricing separate from onboarding separate from bugs.

Do it on Probe

This system builds on a new Paper, a Doc, Build with AI, and a written request describing the fields and views you want. Start at Probe's Work Studio.

Was this helpful?