Research / Jinn Research
Jinn Research

Why Most In-Store Sales Conversations End Without a Purchase

82 in 100 shoppers left without buying. Most never said no. Findings from 1,686 recorded sales conversations across jewelry brands, and what they mean for your sales floor.

Why Most In-Store Sales Conversations End Without a Purchase

Retailers measure who walks in and what gets billed. Almost nobody measures the conversation in between.

That conversation is where the sale is won or lost. We recorded 1,686 of them between store associates and shoppers across jewelry brands, and graded how each one ended. This post covers what we found.

1. Key takeaways

  • Four in five conversations end without a sale. 82% across brands (95% CI: 80–84%).
  • Most lost sales aren’t a “no.” 43% of shoppers left with no appointment, follow-up or recorded reason.
  • One in seven leaves warm. 14% agreed to come back or asked for a quote.
  • Conversation quality tracks conversion. Strong conversations converted at 35%; weak ones at 2%.

2. What we listened to

Sales associates wore Jinn badges on the floor. Every conversation was transcribed and scored across nine stages, from the greeting and discovery through price, objections and the close. We also graded how each visit ended, on six levels from “bought” to “clear no.”

1,686conversations
9stages scored per conversation
6outcome levels graded
18%average conversion (12–22% across brands)

3. Five things the data shows

FINDING 01Four in five conversations end without a sale

Most store leaders sense this number. What they can’t see is how those conversations ended.

Ended in a purchase18%12% to 22% across brands
Ended without a purchase82%95% CI: 80–84%

All 1,686 conversations

FINDING 02Most lost sales aren’t a “no.” They just stop.

We graded every outcome. The largest group didn’t object or decline. They simply left.

Bought
22%
Committed to a next step
8%
Warm, will follow up
7%
Left with no clear next step
43%
Left with an unanswered concern
9%
Clear no
12%
Where 979 conversations ended, every outcome graded · 213 bought, 74 committed, 65 warm, 424 no next step, 84 unanswered concern, 119 clear no

A shopper who leaves without a next step hasn’t said no. They just haven’t been helped to say yes.

FINDING 03One in seven shoppers leaves warm

14%agreed to come back or asked for a quote
1 in 5of the shoppers who didn’t buy is still a recoverable sale

These sales are only recoverable if the store remembers what each shopper asked for.

FINDING 04How the conversation goes decides the sale

We scored every conversation on how it was handled, then compared conversion across score bands.

Strong
35%
Developing
15%
Needs coaching
2%
27%of conversations scored strong
21%of conversations needed coaching
Conversion rate by conversation quality score, 1,686 conversations · 95% CI: strong 30.8–39.5%, developing 12.8–17.5%, needs coaching 1.0–4.1% · average 18%

Strong conversations converted at nearly double the 18% average. Conversations that needed coaching almost never did.

FINDING 05It holds when you compare like with like

Longer conversations and eager shoppers convert more on their own. So we tested the score where those are equal.

How often does the score rank the buyer above a non-buyer?
69 in 100among conversations of similar length
65–75 in 100among shoppers with similar buying intent
72 in 100on later months the model never saw

A coin toss would score 50 in 100. All tested on conversations the model didn’t learn from.

4. What this means for your stores

Three places to start:

Treat “no next step” as a coaching gap. Most lost sales weren’t lost to price or a competitor. They ended without an ask.

Work the warm list. Nearly 1 in 5 shoppers who didn’t buy left open to coming back. Knowing what each asked for turns a follow-up into a sale.

Measure the conversation, not just the outcome. Once every conversation is recorded and scored, you can see which ones go wrong, and where.

5. A note on the numbers

Data. 1,686 in-store sales conversations between associates and shoppers across jewelry brands, 2026. Order pickups and duplicates excluded.

Scoring. Each conversation is recorded on a wearable badge, transcribed, separated by speaker and scored against nine stages of a consultative sale.

Validation. Stage weights are learned from what separates purchases from non-purchases of similar length, then tested only on held-out conversations. AUC is 0.73 across brands (95% CI 0.70–0.77 on the largest sample), 0.69 within similar-length conversations, 0.65–0.75 within buying-intent bands and 0.72 on a forward time split. 0.5 = chance.

Intervals. Conversion by score band uses the full dataset, with Wilson 95% intervals. The held-out AUC figures are the stricter test.

6. Coming up in the series

Over the next few weeks, we’ll look at how conversations are scored, and how to make that scoring fair.

Coming soon · 02Most sales scorecards just measure how long you talked
Coming soon · 03Talking more doesn’t sell. Moving forward does.
Coming soon · 04A sales score you can actually bet on
Coming soon · 05Skill or the shopper? Scoring that’s fair to your associates
jinn researchretail intelligenceconversation analytics

About Jinn Research

Jinn Research studies what happens in real sales conversations on the retail floor. Our insights come from showroom, test drive and follow-up conversations captured by Jinn and scored from greeting to close.

Share Your Thoughts

Join the conversation and let us know what you think