Metrisim Logo
Title
Go to content

Validating a new robotic coffee shop concept

Metrisim
Published by Staff author in Implementing theory · 17 August 2025
       
‘Robo Coffee Bar’ will use a visible robot arm and other automations to speed-up coffee making. Get professional barista-level quality in a fraction of the time!

Research objectives:

  • Is 5% market share achievable in the target town?
  • Is there a specific segment that would find the robot more appealing?
  • What pricing will maximize revenue (for a medium sized coffee)?

Attributes and levels to use in constructing alternative products



Attribute importance
Price and wait time were the most important attributes.

Forecast

A share of visit volume prediction of around 17% for the town, an approx. 5% lift due to the switch to a robot and the lower 1 minute wait time. Just over a third of share goes to ‘none’ (consisting of independents and other brands not listed). In total $665 per 1,000 customers (in the broader market), per purchase occasion. Model assumed equal availability of each brand, since 1 branch per brand in town. However, a roll out to the entire US would achieve lower shares as newcomer is unlikely to achieve similar store network sizes.

What segments are there?

Cluster analysis revealed 3 clusters:
Cluster 1: Robo haters - prefer traditional brands, not very time or price sensitive (30%).
Cluster 2: Robo lovers – prefer robot over human; time and price sensitive (12%).
Cluster 3: Indifferents – only care about time and price, don’t care about brand or whether a robot or a human (58%).

Price curves

Price share curves showed Robo Coffee Shop was fairly inelastic to price changes over the tested range, so that revenues increased as price increased. The peak was $4.5 ($773 revenue per 1,000). It is likely this would have dropped off had higher prices been tested.

Conclusions / recommendation

  • The robot arm & shorter wait times lift share +5%.
  • Price: $3.86 to $4.5 lifts revenue +16%.
  • Robo Lover segment share is +26% higher.  

Method

50 US synthetic AI personas were surveyed. Each persona was conditioned on data from a US survey.
It is recommended that human survey respondents are used for business critical decisions.

Want your own custom study?

You can either use Metrisim conjoint analysis software yourself, or if you prefer not to conduct your own study, a design and analysis service is also available.


Back to content