Skip to content
Senpai.ai

Research: We study how markets decide.

Original research into AI search and buyer decisions: what customers ask, how ChatGPT, Gemini and Perplexity answer, and which brands get recommended.

How the research works

Studies

Recent research

Research into real markets, customer decisions and AI behavior, focused on useful commercial signals.

Crypto exchangesNetherlands

The Dutch exchange market has a hidden revenue problem.

  • 841 verbatim Dutch market observations
  • 15 underlying commercial decision states
  • 120 query AI and search test panel
FragranceUnited Arab EmiratesCollected August 2026

Brand ownership changes with the buying situation.

  • 8,102 real shopper expressions, English and Gulf Arabic
  • 100 representative shopping questions
  • 1,500 AI answers, five runs across three assistants
Agentic commerceLebanonCollected September 2026

Does AI know what you actually stock?

  • 3,666 real buyer comments, posts and searches
  • 105 specific buying situations
  • 342 AI answers: 38 decisions, asked three ways

Method

How the research works.

We collect large volumes of evidence from search, communities, reviews and other public sources, then map the decisions underneath it, their commercial importance, and how those decisions resolve across AI.

Market decision mapCrypto exchanges, Netherlands

Decision states shown

  1. Euro cash-out and bank frictionWhite space
  2. Fee arithmetic at sizeOwned
  3. Outgrowing a beginner exchangeContested
  4. Leverage, shorting, perpsOwned
  5. APIs and automationFragmented

Five of the 15 decision states in the study. White space is a decision still open across AI.

841 verbatim observations
From the Netherlands study. The dots illustrate how observations gather into decision states. Their number is illustrative: 360 dots stand in for 841 observations, and the size of a cluster is a drawing choice, separate from market share.

Visibility is only useful when it connects back to real customer demand.

Visibility tracking asks

“Where does my brand appear?”

The research asks

“What are customers trying to choose, which brands or products get recommended for those decisions, why they win, and where there is room to take that demand?”

Evidence sources

  • Search behaviour
  • Reddit
  • Forums
  • YouTube
  • Social discussion
  • Reviews
  • Autocomplete
  • Natural search language
  • Support tickets
  • Sales calls
  • CRM notes
  • AI prompt datasets

  1. Collect the evidence

    Real customer signals from search, communities, reviews and other public sources.

  2. Map the decisions

    Group that evidence around what people are actually trying to choose.

    • Decision states named and bounded
    • The language customers use for each

  3. Weigh the opportunity

    Work out which decisions matter most commercially and which are realistically worth pursuing.

    • Proximity to conversion
    • Expected customer value
    • Frequency and market demand
    • Recency
    • Competitive intensity
    • How realistically it can be won

  4. Test who wins

    See which brands and products get surfaced across ChatGPT, Gemini and Perplexity, and what appears to shape the result.

    • Who owns which decision state
    • How brands are positioned
    • Which sources and citations shape the answer

  5. Find where to act

    Identify the areas where the business has the strongest opportunity to improve visibility and capture more demand.

    • Ranked decision territories
    • What it would take to win each
    • What to do first

What a study answers

The questions the research is built around.

Every study is scoped to a market and a set of decisions inside it, so the findings can be acted on rather than filed.

  • Which decisions customers are trying to make in the category.
  • Which brands and products get recommended for those decisions.
  • What separates the options that get named from the ones that never come up.
  • Where demand is going somewhere else and could be taken back.

Find what's worth changing next.

Bring the market, workflow or business question. We'll start there.

Chat on WhatsAppmail@senpai.ai