Skip to content
Specicon Specicon
Book a demo
Case study — American Fragrances

Guesswork out.
Guided journeys in.

A 10,000-bottle fragrance catalogue, one scoped decision, four weeks to proof — how explainable matching replaced "customers also bought."

+2.8×
conversion lift
27%
return rate
4 wks
to proof
18ms
p99 latency
The problem

Ten thousand bottles. One guess at a time. And no one could say why.

Fragrance is personal — the same bestseller is wrong for most people. "Customers also bought" filled the carousel but not the cart: returns crept up, and when a shopper asked why this one?, there was no answer.

What shipped

One decision, live in weeks.

01 · WEEK 1 — SCOPED

"Which fragrance fits this shopper?"

One decision on the Launch tier: a guided finder on the product page and quiz — not a platform migration.

02 · WEEK 2 — INTEGRATED

One endpoint into the storefront.

The catalogue mapped to Product DNA, shoppers resolved to User Barcodes™ — one REST call from the stack they already ran.

03 · WEEKS 3–6 — PROVED

50/50 holdout, in production.

Guided journeys against a control — conversion, returns and repeat rate monitored live, not modelled in a deck.

The numbers

Measured against a holdout, not a hunch.

+2.8×
conversion lift
27%
return rate
99.2%
match rate
18ms
p99 latency

12 weeks · guided journeys vs 50/50 control · live monitoring

wk 2live since
Specicon turned our "customers also bought" guesswork into matches we can actually explain to a buyer. Conversion on guided journeys is up, returns are down.
PR Priya RamanVP Growth · American Fragrances
Your first use case

Bring your catalogue. We'll map the same path.

One scoped decision, one endpoint, one holdout — and numbers you can defend in a quarterly review.

Book a working session Explore solutions