← All research

Lansy Research · Food service and delivery

Real conversations are worth real money

We analyzed more than 22,000 conversations with guests to find out why most upsell attempts never turn into a purchase, which phrasings quietly cost the business money, and how conversations in a chain’s strongest locations differ from its weakest.

Sometimes the difference between a basic order and a larger one is a single sentence said in the 40 seconds at the register.

22,000+
real conversations with guests
100+
locations across food service chains
+9%
gap in average check between a strong and a weak location of the same chain

The sample · Across more than 100 locations

22,000+
conversations
100+
locations
6
revenue growth zones
3
ways to raise the check

Three things the data showed

4 in 5

upsell attempts end without a purchase. Most often the employee names the item but never says what it is for — the guest hears a price, not a reason.

+9%

difference in average check between a strong and a weak location of the same chain, on a comparable menu at comparable prices. The menu is not what separates them; the conversation is.

Three myths

keep teams silent: that guests find offers pushy, that regulars already know the menu, and that a busy shift leaves no time. The data disagrees with all three.

What is inside

  • Six zones of revenue growth found in conversations with guests
  • What the best locations do differently — the specific moves that raise the check
  • Three ways to raise a check: the addition, the substitution, and the upsell
  • The phrasing that spoils a guest’s impression, and what to say instead
  • Why most upsell attempts fail, and how to fix it
  • Natural recommendations without pressure — with examples
  • Handling complaints: the phrasings that keep a guest loyal

Why the data holds up

  • Conversations were taken as they came, not selected for the study
  • One scale was applied across every location and every chain
  • Findings rest on transcripts and specific fragments of the conversations
  • Guest and employee data is anonymized
  • Chains differ: for another network these are hypotheses to test on its own conversations

See the same analysis on your own conversations

A working pilot on your own criteria takes one to two days. The first reviews come back the same week.