What if you could accurately predict which restaurants will fail?

Starting a restaurant is definitely not a ‘get rich quick’ strategy – up to half of all new restaurants fail within their first five years, even in normal times when restaurants aren’t obliged to close their doors because of Covid-19. So being able to predict if a restaurant business will fail is of tremendous importance. Researchers Dr Markus Weinmann from Rotterdam School of Management, Erasmus University (RSM), Christof Naumzik from ETH Zurich and Prof. Stefan Feueriegel from LMU Munich School of Management used online customer ratings to predict a restaurant’s likelihood of failure. Their prediction tool can predict 7 out of 10 failures, even months ahead.

The restaurant market is very volatile and dynamic. Even minor changes – such as updating a  menu or a change of personnel ‒ can make customers switch their loyalty. The success, or even the survival, of service businesses depends on their ability to satisfy their customers. Yet, businesses are often too late to recognise slumping customer satisfaction and suffer the ultimate failure. So an early warning system for restaurants would help them to adjust their service offerings in time.


“Ratings can predict 7 out of 10 business failures, even months ahead”


Three states

The researchers used almost 65,000 customer ratings from more than 900 restaurants as the input for their data analysis, and developed an AI model to estimate the most likely state that a restaurant is in. They found three states:

  1. restaurants that get bad ratings but are still in business,
  2. restaurants that get good ratings and are still in business, and
  3. restaurants that have very dispersed ratings (ranging from very positive to very negative) and are still in business.

They identified that restaurants in State 3 – those with wildly fluctuating reviews, some very positive and some very negative – are most at risk, and have a likelihood of failure which is twice as high as restaurants in the other two states. Their model can detect 7 out of 10 business failures, even months ahead.


Useful indicator

Restaurants using this tool to estimate their current service state (good, bad, or risky) could identify their likelihood of failure. And if they find their reviews predict that they’re in the risky state, they can take steps to improve their service. It could also be useful for investors to assess the current state of a restaurant before committing to backing it or investing more.

Taking it a step further, review platform providers such as Yelp or TripAdvisor could offer this tool as an additional premium service to their restaurant members, making the platforms more useful and attractive for restaurants.


This article was published in Marketing Science under the title I Will Survive: Predicting Business Failures from Customer Ratings.


Rotterdam School of Management, Erasmus University (RSM) is one of Europe’s top-ranked business schools. RSM provides ground-breaking research and education furthering excellence in all aspects of management and is based in the international port city of Rotterdam – a vital nexus of business, logistics and trade. RSM’s primary focus is on developing business leaders with international careers who can become a force for positive change by carrying their innovative mindset into a sustainable future. Our first-class range of bachelor, master, MBA, PhD and executive programmes encourage them to become to become critical, creative, caring and collaborative thinkers and doers. Study information and activities for future students, executives and alumni are also organised from the RSM office in Chengdu, China. www.rsm.nl

For more information about RSM or this article, please contact Danielle Baan, Media Officer for RSM, via +31 10 408 2028 or baan@rsm.nl.

Share this article: