Periscopy
← All news
MarketeerPRESS · SEP 2, 2026

These are the Portuguese brands AI actually recommends

Marketeer covers the destaque.ai study behind Periscopy: 616 answers from 10 AI assistants, and the gap between being named and being chosen.

616AI answers analysed
10AI assistants
14questions about Portugal
2,993citations in 328 answers

Appearing in an AI answer means a brand is in the conversation, not necessarily in the decision. The study separates how often a brand is named from how often the assistant actually chooses it, and the ratio between the two becomes the recommendation rate.

01

EDP overtakes Salsa when it’s time to choose

Asked for the best Portuguese brands, a median answer lists 17 names: a catalogue, not a pick. Salsa is named most, but EDP is chosen most.

“BEST PORTUGUESE BRANDS” · 38 ANSWERSNAMEDCHOSEN
EDP58%
Salsa23%
Delta Cafés47%
Vista Alegre50%
Bordallo Pinheiro40%
Claus Porto42%
Compal0%
Chosen / named, recommendation rate on the right.

At the far end, 13 Portuguese brands are mentioned and never chosen: Compal is named in 18 answers without a single pick. That is the limit of share of voice, and why the study adds share of recommendation.

02

Who builds a brand’s reputation with AI

328 of the 616 answers showed sources, 2,993 citations in total. The most cited are not the channels brands usually watch.

DECO Proteste105
Reddit88
Wise79
ECO72
Literacia Financeira61
4gnews60
03

Recommended is not the same as approved

A brand counts as chosen even when the answer lists its weak points. TAP is chosen in all 39 answers about the best airline from Portugal, yet 16 of them also raise criticism on price, punctuality or service.

04

The question decides the ranking

Three phrasings of the same banking question, asked to the same assistants, produce very different winners.

Novo Banco87%
Revolut8%
Share of answers that name each bank.
Read the full article on MarketeerSummary by Periscopy. Original reporting © Marketeer.marketeer.sapo.pt

Named or chosen?
See where your brand stands.