February 24, 2023

We Asked ChatGPT How To Collect Guest Feedback. It Went Very Wrong.

tinku saini tarka indian kitchen quote

If you’ve been on the Internet lately, you’d find it nearly impossible to avoid topics like “AI” and “ChatGPT”. It seems like every  industry – including restaurants – are trying to stay ahead by envisioning AI use cases that could lead to greater efficiency and better results.

Up till now, AI implementation at restaurants mostly entails a variety of robot automations. However, the next generation of AI, pioneered by companies like OpenAI and tools like ChatGPT, is sparking new interest in how it could be integrated into restaurant operations. This form of text-based AI could potentially help restaurants generate menu ideas, predict food costs, perform data analytics, and much more.

So, if generative AI — specifically in the form of ChatGPT — is so powerful, we had to ask: What does AI think is the best way for restaurants to collect private, direct customer feedback that’s actionable for operational improvements?

And to our surprise, ChatGPT couldn’t have gotten it more wrong.

What ChatGPT Got Wrong

In response to our prompt of  “How should restaurants collect private, direct customer feedback?”, the output – outside a single valid recommendation, was incredibly outdated and impractical.

While our not-quite sentient friend may have the best intentions, it was severely lagging behind the latest feedback management technology and approach. Many of the generated suggestions fail to deliver data-backed, statistically-significant feedback insights, and are in fact preventing restaurants from maximizing the value they could get from guest feedback at scale.

Let’s break down the results from ChatGPT and see why there’s a lot left to be desired in its answers.

The screenshot below shows ChatGPT’s answer:

chatgpt results

[Debunked] 1. Provide simple feedback cards with a few questions and a comment section.

There’s a lot to pick apart here. Let’s start with the methodology: feedback cards.

While feedback cards – or comment cards – were all the rage a few decades ago, they have already been phased out with the onset of more convenient and powerful feedback technology.

First of all, these cards only collect feedback from dine-in experiences. As virtually all restaurants transition into omni-channel services post-COVID, operators must closely monitor guest sentiment across dine-in, takeout, delivery, curbside, pickup, and more.

This is aside from the fact that the last thing customers want to do is hand write a recount of how their experience was.

This is why Tattle uses integration with POS, digital ordering and loyalty programs to collect feedback from all your sales channels, so you can easily benchmark channel performances and isolate problematic areas unique to each delivery method.

And then there’s another problematic area: the survey questions themselves.

A simple survey with only a few questions might sound attractive at first glance, but that doesn’t provide enough data points to pinpoint root causes of issues, or show patterns across time.

Similarly, with text comments, it’s nearly impossible to detect patterns and garner insights across various anecdotes. This means very often the angriest texts with the most number of capitalized words tend to get your staff’s attention, and not the improvement opportunity most correlated with guest satisfaction.

The solution is causation-based, structured survey that makes it extremely intuitive for customers to provide up to 55 points of feedback per submission. In fact, every 9 in 10 customers who start the survey will finish it — and that’s how restaurants using Tattle are seeing 2,000% more feedback volume and better operational efficiency.

[Debunked] 2. Ask for feedback during a meal.

Again, this would only apply to dine-in experiences, and most likely only at full service restaurants. As a result, this method is flawed in a few ways, most notably:

  • Guests might feel compelled to disguise their true feelings so as to avoid having an awkward conversation with the wait staff.
  • This form of feedback collection is based solely on anecdotes, and not measurable, scalable, data-backed patterns.
  • When put on the spot and asked head on, only a vocal minority of customers would likely even provide feedback.

However, there’s value in collecting real-time feedback from customers while the experience is still fresh in their mind. That’s why Tattle automatically triggers a survey email 90 minutes after a transaction, enabling guests to let you know what they actually think right away. This also means you can respond and win back the customer quickly should they be dissatisfied, rather than waiting weeks or months until it’s too late.

[Debunked] 3. Use feedback software like tablets and smartphones.

This has some merit on the surface, in that feedback software has proven to be much more efficient, convenient, and analytical.

However, we do not recommend any feedback platform that requires you to purchase hardware with it. This is because any tablets or handheld devices are not only expensive, but also limit you to only dine-in feedback. Additionally, this type of feedback is not tied to individual transactions or customers, which means there’s no way for you to keep track of the satisfaction level of an individual customer.

A better alternative would be an automated, browser-based, white-labeled survey that customers can fill out on their own device. Thanks to integrations with your ordering techstack, Tattle’s email surveys are pre-populated with the customer’s order details – meaning they only need to rate each operational category and the underlying factors. This user-centric, frictionless survey design helps restaurants collect significantly more feedback in an actionable, scalable way.

[Debunked] 4. Engage with customers on social media.

Like the last one, this point is valid on the surface — brands should always engage with customers on social media, regardless of whether the feedback is positive or negative. This is your chance to show you care, display your professionalism, and drive new sales.

However, this is still markedly insufficient for a strategic, brand-wide feedback approach. Ideally, private feedback (Tattle surveys) should be a filter for public feedback (Google, Yelp, Facebook, etc.).

Tattle succeeds in the fact that it can intercept negative feedback and boost positive reviews online. In using a direct feedback platform first, your restaurants are the first to hear about a dissatisfactory experience — not the entire internet. This means your staff will have the chance to win back that customer with automated, templated email follow-ups and rewards if applicable.

Additionally, Tattle enables brands to prompt happy customers to leave positive reviews through automated email follow-ups. This personalized, scalable approach will effectively boost your overall online presence within a short amount of time.

[Agreed] 5. Follow-up survey submissions via email or SMS.

Let’s end on a high note – our first agreeable point from ChatGPT.

Data has shown that follow-up email surveys or texts are most effective at collecting structured, detailed, and real-time feedback at scale.

The only thing we’d add is this: Automate your follow-up surveys and make the most out of that interaction. What do we mean this?

By automating your follow-up survey emails or texts, you can make sure to capture as many customers as you can without having to train your staff on it. You can set it and forget it, and then watch your feedback submissions rolling in. In fact, an average Tattle partner sees a 10% survey penetration rate and a 94.6% survey completion rate once they start using Tattle’s platform.

Now, how do you make the most out of that survey interaction? The most common use cases are demographic data collection, guest recovery, and reputation management. For example, restaurant marketing teams can add some demographic and behavioral questions at the end of the survey to better understand their audience, or even prompt customer to enroll in the loyalty program.

Secondly, you should leverage your feedback surveys for guest recovery. Once you know the satisfaction level of an individual customer through the survey submission, Tattle can automatically send out a templated apology email and/or reward to incentivize the customer to come back a second time. The emails could be different based on the star rating, or your staff can send manual emails to make it extra personalized.

Not only does Tattle achieve a 70% guest recovery rate using this method, but also you have a higher chance of appeasing angry customers and intercepting potential negative reviews online.

Similarly, your happy customers will receive an email prompt based on their satisfaction level to leave you a nice review on Google or Yelp. Restaurants that set up such automation almost always see their overall rating on review platforms improve over time.


The fact that ChatGPT was still returning outdated feedback management methodologies indicate that many, if not most, restaurants still have tons of room for improvement in terms of CX (customer experience) strategy optimization.

This also means that your restaurant operation is probably on the verge of exciting growth and all you need is a powerful tool to help you get there. What Tattle does is close that gap between you and your customers, while leveraging technology to automatically analyze your guest feedback data and identify the top improvement opportunities at each location.

If you’re interested to see a platform tour or assess whether Tattle would be a good fit for your operation, simply schedule a time with our team here.