← all casesNatural & vegan foods · Nuvemshop
Empório Recanto Ipiranga
Before 05/16–06/14 · After 06/16–07/15/2026 (30 days each)
+38.5%
store conversion rate
Conversion rose from 1.35% to 1.87% — and, most importantly: with 13% fewer visits. The store began converting the same audience better, with each visit worth more.
Store conversion
1.87%
before 1.35%
▲ +38.5%
Transactions
222
before 185
▲ +20.0%
Revenue
R$ 29.4k
before R$ 23.4k
▲ +25.5%
Buyers
211
before 174
▲ +21.3%
New buyers
176
before 146
▲ +20.5%
Sessions / visits
11,878
before 13,712
▼ -13.4%
Product views
14,569
before 15,204
▼ -4.2%
Add-to-cart
6,601
before 7,735
▼ -14.7%
§ the challenge
A large catalog of natural and vegan foods, but with short, generic descriptions — many inherited straight from the supplier. Similar products got confused with one another, information customers need to decide was missing (dietary restrictions, how to use, differentiators), and the pages didn't show up in searches from the very people looking for those terms.
§ what we did
01
Catalog audit
We mapped the highest-potential PDPs and what each one was missing to rank and convert.
02
Rewrite in citable blocks
341 descriptions rewritten with an answer-first opening, a tabulated spec sheet, and real search language (vegan, lactose-free, zero sugar).
03
FAQ and structured data
Complete per-product FAQs and Product schema — readable by humans and retrievable by AI.
04
Publishing and readout
Batch publishing between 06/08–06/29 and a comparison of 30-day windows, before and after.
§ how we measured
a transparent, replicable, no-black-box method
The result only counts if you understand exactly how it was measured.
01
Comparable windows
Two periods of equal length, one before and one after the work.
Before · 05/16 – 06/1430 days
After · 06/16 – 07/1530 days
02
Efficiency, not volume
Traffic dropped 13% over the period — so the gain didn't come from more visits. It came from converting the visits the store already had better. It's the cleanest signal that the page content is performing.
less traffic
vs
more conversion
03
How we calculate
Conversion is always outcome divided by opportunity — no inflated metrics.
transactions
sessions
= store conversion
items purchased
items viewed
= conversion per product
Two 30-day windows, immediately before and after publishing the new descriptions.
Before vs after, side by side
Across optimized products (265 analyzed)
Conversion per view on the optimized products jumped from 17.0% to 26.4% — a gain of +9.3 p.p. (+55%). Purchases stayed practically flat (−3.6%) even with 37.7% fewer views: far less traffic, nearly the same sales — the description held conversion up.
Product-by-product highlights
Product
Views (before → after)
Conversion
Δ conversion
Purchases
NotMilk Semi 1L | NotCo
39 → 30
2.6% → 60.0%
▲ +57.4 p.p.
1 → 18
Not Creme 200g | NotCo
1,119 → 157
5.1% → 59.9%
▲ +54.8 p.p.
57 → 94
Not Milkinho 200mL | NotCo
41 → 39
31.7% → 71.8%
▲ +40.1 p.p.
13 → 28
Queijo Ralado Vegetal Parmesão 50g
113 → 58
27.4% → 58.6%
▲ +31.2 p.p.
31 → 34
AdeS Original Zero 1L
86 → 81
7.0% → 37.0%
▲ +30.1 p.p.
6 → 30
Hypothesis holding in the short term
The description work improved conversion efficiency over a ~30-day span. The gain didn't come from more traffic (which dropped), but from turning existing visits into sales more effectively — the cleanest indicator that the PDP content is performing.
Methodological caveat
This is a correlation analysis. Seasonality, paid-media mix, and promotions during the period also play a role. Recommendation: track the next 4 weeks to confirm the trend.
Full methodological note · Store conversion = transactions ÷ sessions. Conversion per product = items purchased ÷ items viewed. 341 optimized products published between 06/08–06/29. Comparable 30-day windows, before and after.
