/
/
Find out why customers rate Zendesk support badly
For support leads who want the real causes behind bad Zendesk satisfaction ratings, found by asking their AI assistant to read the tickets behind them.
On this page
Connector
Connect Zendesk, pick your AI tool and paste the prompt.
The problem
Bad ratings arrive without a reason attached
Bad satisfaction ratings in Zendesk often have no comment, so the reason is buried in the ticket thread. Reading each one takes hours, and it is tempting to blame whoever was assigned.
Without PopdockAI
- Filter bad ratings in Zendesk
- Open each ticket and read the whole thread
- Guess the cause and tally it in a spreadsheet
- Blame the assignee on thin evidence
With PopdockAI
- Bad and good ratings are read for the ratio
- Up to 15 bad rated tickets are read in full
- Each is classed by cause with a customer quote
- Agent counts come with volume as context
The prompt
Use this prompt
Copy it into any AI tool connected to PopdockAI. Swap the values in braces for your own.
**Objective:** Explain why customers rated support badly in the last `{lookback-days}` days (default 30), optionally for group `{group-name}`, and recommend fixes.
**Step 1: Resolve the group**
If `{group-name}` is given, call `zendesk-get-group` with group-name set to it to get its group-id. If it matches none or several, list what you found and stop. The ratings tool filters by group-id, not by name.
**Step 2: Pull the ratings**
Call `zendesk-get-satisfaction-rating` with score bad, created.from set to today minus `{lookback-days}` days (YYYY-MM-DD), group-id if resolved, and max-records 200. Also call it once with score good and the same filters, only to compute the good-to-bad ratio.
**Step 3: Read behind each bad rating**
For up to 15 bad ratings, most recent first:
* Use the rating comment when there is one.
* Call `zendesk-get-ticket` with the ticket-id for status, priority, created-date and solved-date.
* Call `zendesk-get-ticket-comment` with the ticket-id and public true to read what the customer was told.
Classify each into one cause: slow response, unresolved or reopened, wrong answer, tone, product problem outside support's control, or unclear.
**Step 4: Report**
1. Totals: bad count, good count, and bad as a percentage of all ratings.
2. Themes table: cause, count, example ticket-ids, one representative customer quote.
3. By assignee: assignee-name with bad and good counts; flag anyone with 3 or more bad ratings, but add context (their volume) before drawing conclusions.
4. Three concrete recommendations tied to the biggest themes.
**Rules & Guidelines:**
* This prompt is read-only. Do NOT call `zendesk-add-ticket-comment`, `zendesk-update-ticket`, `zendesk-update-user` or any other add, update or make tool.
* Do not frame an individual agent as the cause from one or two ratings.
* Use unclear when neither the rating comment nor the thread explains the score.
**Validation:**
* If there are no bad ratings in the period, report the good count and stop.
* If there are no ratings at all, say satisfaction surveys may not be enabled.Who it's for
One answer, three teams that use it
Support managers
See what is really driving bad ratings
Get the causes behind bad ratings, a customer quote for each theme and three recommendations, without reading every ticket yourself.
What were the main reasons for bad satisfaction ratings in Zendesk last month?
Team leads
Coach on patterns, not single tickets
Counts by assignee come with their ticket volume, so you can coach fairly and spot real patterns rather than reacting to one rating.
Break down bad Zendesk ratings for the Billing group over the last 60 days by cause and by assignee.
Customer experience leads
Turn bad ratings into fixes
See whether bad ratings come from slow responses, wrong answers or product problems, so you know which fix will move satisfaction.
Are bad Zendesk ratings this quarter mostly about slow responses or product problems?
How it works
What PopdockAI does when you ask
You name a period and optionally a group, and your AI assistant reads the ratings and the tickets behind them. It is read-only and will not single out an agent from one or two ratings.
01
Ask
You name a lookback period, and optionally a group.
02
Read
It pulls bad and good ratings, then reads up to 15 bad-rated tickets and their replies.
03
Classify
It gives each bad rating one cause, such as slow response, wrong answer or tone.
04
Report
You get totals, themes with a customer quote, counts by assignee and three recommendations.
What PopdockAI provides
The tools behind the answer
Ready made tools for Zendesk and one secure connection to your AI tool. No code, no exports.
zendesk-get-group
zendesk-get-satisfaction-rating
zendesk-get-ticket
zendesk-get-ticket-comment
This report is built from Get data tools only. PopdockAI applies the row, field and parameter rules your admin sets, and every query is written to the audit trail.
The outcome
Fix the causes, not the people
Related use cases
More ways to put this data to work
Assign and reply to a Zendesk ticket
Ask your AI to assign a Zendesk ticket and it resolves the agent, fixes the group if needed, and applies the assignment, status and reply in one update after you confirm.
Learn More
Answer a Zendesk order status ticket from Shopify
Ask your AI to answer a Zendesk order status ticket and it finds the Shopify order, checks it belongs to the requester, drafts a reply with tracking, and posts only after you confirm.
Learn More
Assess the Zendesk customer impact of a Jira issue
Ask your AI how many customers a Jira issue affects and it searches recent Zendesk tickets, groups them by organization, and rates the impact low, medium or high.
Learn More
Escalate a Zendesk ticket to a Jira issue
Ask your AI to escalate a Zendesk ticket and it checks for an existing escalation, creates the Jira issue after you confirm, links it back and notes it on the ticket.
Learn More
Find Zendesk tickets for a Bugsnag error
Ask your AI which customers reported a Bugsnag error and it searches Zendesk in customer wording, confirms each ticket in its conversation, and counts affected organizations.
Learn More
One prompt template with infinite possibilities.
Your AI assistant adapts its approach, tone, and capabilities to match each unique context while you maintain a single, manageable workflow.
Join forward-thinking companies that are transforming their operations with intelligent, secure AI automation.