What AI says about your firm’s complaint handling

I asked Gemini, ChatGPT, Claude and Perplexity the same question, changing the name each time.

What information does [name of AI chatbot] summarise when I ask you to recommend a company?

Here’s what came back about reviews and complaints.

Gemini

Ratings and Reviews: Star ratings alongside specific snippets or sentiments quoted directly from customer reviews

ChatGPT

How it compares with similar companies. I’d look at complaints per 1,000 accounts where comparable data is available, how often complaints are upheld, and what they’re about. I’d give more weight to recurring problems than to reviews about how quickly a simple task takes to complete

Claude

Reputation: independent customer reviews and any complaints history

Perplexity

Reputation and reviews: Customer ratings, review themes, testimonials, complaints, and recent feedback from platforms such as Google, industry review sites, or marketplaces

Whilst there’s little doubt AI can help firms improve efficiency, it can also become a major headache. It can act as a ‘zero-click’ digital storefront. Someone can ask who to buy from and get an answer without ever visiting your website.

If this answer draws on what customers say about how you handle complaints, the work your complaints team does can become part of someone else’s decision to buy.

What can AI miss when it summarises customer reviews?

AI can pick up what went wrong in a customer review and leave out what the firm did to put it right. When it condenses several paragraphs into a sentence or two, it can also merge points it sees as similar across different reviews.

Say a customer writes that they took two afternoons off work for a boiler repair and the engineer didn’t turn up either time. The person who picked up the complaint arranged a Saturday visit, and the boiler was fixed without the customer having to take any more time off. An AI summary might pick up the two missed appointments and leave out the Saturday visit. Someone looking for a firm to fix their boiler might make up their mind on this alone.

How to check what AI says about your firm’s complaint handling

Try asking AI in an incognito window what customers say about how your firm deals with complaints. If it links to reviews, open them and check what made it into the answer. Check when the reviews were written, too. Then look at your recent complaint data.

• Are you capturing the issues raised in the reviews?

• If people are still complaining about chasing an answer in recent cases, can you identify where the process is falling short?

If the answer is no, your complaint data isn’t doing its job of pointing you to systemic root causes in your products and/or services.

How to follow up after resolving a complaint

Check that you’ve done what you promised in the final response, then ask the customer to share how you handled their complaint.

A review explaining how you put things right can give AI a fuller picture to draw on.

This starts with a structured complaints process, from capturing the complaint to actively asking for feedback on how it was handled.

I know that the “case closed” button is often hit as soon as the final response is released to the customer, with no follow-up. Now, I get it. There are targets to hit in-house or by a regulator, and the customer’s bound to let you know if something’s not as it should be, right?

Hit the “case closed” button so that the dashboard can turn green if this is what’s needed. But if the customer’s still waiting for something you promised in the final response, there’s still work to do.

Diary the case for a couple of days after you expect the issue to be fixed and send a follow-up message to the customer to check it’s been sorted. For teams using complaint management software with a secure case portal, this can be diaried as an action and the message sent via the secure case portal.

Why ask customers for feedback after resolving a complaint?

Asking for feedback helps you find out whether the customer considers the complaint resolved. You might think a problem’s been sorted when the customer is still waiting for you to do something you promised.

Firms often avoid asking for feedback on complaint handling because they’re worried about rocking the boat. They might reignite a customer’s frustration or get a poor satisfaction score.

Artificial metrics may look good on the outside, but they can hide what’s really going on. Scores may drop initially when you ask more customers for feedback. Their answers give you a more solid foundation to build on.

How a complaint charter can give an AI summary more context

If an AI chatbot finds your complaint charter, it can use what your firm promises alongside what customers say about how you handled their complaints. Publish a charter if you can consistently stick to it. If your charter says you’ll check whether a refund has arrived, make sure someone makes the call.

If you publish your feedback scores and they drop after you start asking more customers, you can explain how you’ve changed the way you collect feedback. You can also explain what you’re doing about the issues customers have raised.

A word of caution if you decide to go down this route. Make sure the charter is easy to find and read on your website. If it gets a bit wordy after collaboration with your compliance team, consider adding an FAQ section with relevant keywords for AI agents to find.

How to ask customers for a review without a long survey

Ask two or three questions customers can answer in 30 seconds. You can take the one-click feedback approach, but a bit more detail will help you understand the score. If you’ve got a complaint charter, connect the request to something you’ve promised in it.

Suggested wording could be along the lines of:

We know long surveys are annoying, so we promise to keep it short.

We’ve just handled your complaint, and it’s important to know how we did. We use your feedback to know whether we’re on track or how we need to improve our internal processes. Please, can you spare 30 seconds for a reality check?

How complaint management software helps teams follow through

Complaint management software can keep follow-up actions with the case so whoever picks it up can see what needs to happen next. If your complaint team must rely on spreadsheets, inboxes and reminders in personal calendars to hold the process together, a manager or team leader hasn’t got the oversight they need to keep cases moving. They need to be able to quickly identify bottlenecks, cases with vulnerability flags, and those that fall into a backlog category. All of these have a higher chance of escalation and customers publicly sharing their experience.

Often, spreadsheet reporting provides more details about the symptom and not enough depth about the root cause of the actual complaint or where the handling process is failing. Without this, there’s a risk of issues recurring, customers remaining unhappy, and AI picking up this information.

If a case handler is unexpectedly absent and their diary is stored on a personal device, it’s harder to reassign cases and identify the next steps. Keeping customers waiting is poor practice, but it happens when there’s no time to catch up.

In Complyr, the follow-up actions on a complaint are part of the case’s workflow. Whoever picks up the case can easily see the next action needed. Complyr’s configurable complaint workflows

Reliable reporting starts with the right data in a single source of truth. Poor data means poor reports.

Spreadsheets can miss important details, weakening your reports. Unless you handle only a handful of complaints a month and can read every file, you risk overlooking information that never reaches a spreadsheet column.

For example, if repair delays appear in customer reviews and active cases, identify where the process is breaking down. An unassigned complaint needs a different fix from a repair an approved contractor has agreed to carry out but hasn’t completed.

Leaders and colleagues across the business need this detail. Complyr’s complaint analytics and reporting

Why following through matters for your firm’s reputation

Complaint and customer service teams are responsible for putting things right, not for causing the original product or service failure. To resolve those problems, they need the right tools and working conditions.

Give your complaints team a system such as Complyr to help them follow through on their promises. The next person asking AI who to buy from could be reading about how your team handled a complaint.