Skip to content
SEOAIExpert

RESEARCH · SEARCH INTELLIGENCE

Research into how search is changing.

Original analysis, experiments and research into ecommerce SEO, AI search, product discovery and generative search visibility.

Book a Strategy CallExplore Our Approach
Approach
Documented
Output
Method + data
Claims
Evidence only
RESEARCH DASHBOARDIllustrative
  • Active projects4 in development
  • Questions definedDocumented
  • Data collectionIn progress
  • Findings publishedNone yet
  • MethodologyPublished

Status shown reflects the research programme. Findings publish only after analysis.

Themes

Questions worth investigating.

The programme is deliberately narrow: how ecommerce brands and products are found, described and recommended across search and AI surfaces.

  • AI SEARCH

    How answers are assembled.

    • AI search visibility
    • Citation patterns
    • Recommendation behaviour
  • BRAND SIGNALS

    How brands are described.

    • Brand mentions
    • Entity visibility
    • Source consistency
  • ECOMMERCE

    How products get found.

    • Product recommendations
    • Ecommerce search behaviour
    • Attribute coverage

Projects

Original research.

Each project publishes its question, method, dataset and limitations. Nothing is presented as a finding before the analysis is complete.

  • Research project in development

    Which sources AI answers cite for product comparison queries

    Question
    For comparison-style ecommerce queries, what proportion of cited sources are retailers, manufacturers, editorial publications or forums?
    Method
    Repeated fixed prompt set across AI search surfaces, citations coded by source type.
    Dataset
    Prompt set defined; collection window in progress.

    Findings are published only after analysis is complete.

  • Research project in development

    Product attribute completeness and discovery

    Question
    Does increasing structured attribute coverage on product pages change how often those products are described accurately in generated answers?
    Method
    Paired sample of product pages, before-and-after observation against a stable prompt set.
    Dataset
    Sample selection under way.

    Findings are published only after analysis is complete.

  • Research project in development

    Entity consistency across third-party sources

    Question
    How does inconsistent brand information across independent sources affect how a brand is summarised in AI answers?
    Method
    Source audit for a set of brands, compared against generated brand descriptions.
    Dataset
    Audit framework defined.

    Findings are published only after analysis is complete.

  • Research project in development

    Category query behaviour in answer surfaces

    Question
    For broad category queries, when do answer surfaces return brands, products or informational guidance?
    Method
    Query classification against a fixed category query set, responses coded by output type.
    Dataset
    Query set drafted.

    Findings are published only after analysis is complete.

Evidence

How we approach evidence.

The same sequence applies to research and to client work, and it is documented in full on our methodology page.

  • A question that can be answered with available data
  • A sample defined before collection begins
  • A method that another team could repeat
  • Limitations recorded alongside the finding

  1. Question

    Specific and falsifiable.

  2. Data

    Sample and collection window defined upfront.

  3. Method

    Repeatable and documented.

  4. Analysis

    Coded consistently, not selectively.

  5. Finding

    Stated only as far as the data supports.

  6. Limitations

    Published with the result.

Measurement

Search should be measured, not guessed.

The measurement frame below is the one used across research projects and client reporting alike.

MEASUREMENT FRAMEIllustrative
Signals tracked across search and AI answer surfaces
SignalWhat it tells usStability
MentionsWhether the brand appears at allModerate
CitationsWhether the site is used as a sourceVolatile
RecommendationsWhether products are suggestedVolatile
Entity descriptionHow the brand is summarisedSlow moving
IndexationWhether pages are eligibleStable
CoverageHow much demand is addressedStable

Applications

Research that helps teams make decisions.

  • Strategy

    Research decides which visibility work is worth sequencing first, instead of following industry consensus.

  • Content

    Observed answer patterns inform how information is structured so it can be quoted accurately.

  • Technical SEO

    Findings about crawl, render and structured data feed directly into audit checklists.

  • Product discovery

    Attribute experiments show which product data changes affect how products are described.

  • AI search

    Citation and mention patterns shape where authority work is invested.

FAQ

Questions about our research.

How ecommerce brands and products become visible in search and AI answer surfaces: which sources answers draw on, how entity information affects the way a brand is described, how product attributes affect discovery, and how citation patterns differ between query types.

With documented prompt sets run repeatedly across AI search surfaces, recording which brands, products and sources appear. Each project defines the question, the prompt or URL sample, the collection period and the analysis method before data is gathered, and records the limitations alongside the findings.

As a trend across a stable prompt set rather than a fixed position. We record whether a brand is mentioned, whether it is cited as a source and whether it is recommended, then track how that distribution moves over time and after specific changes.

Question, data, method, analysis, finding, limitations — in that order. We publish the method with the result so the work can be challenged, and we do not present observations from a small sample as a general rule.

AI search surfaces change frequently, so published projects are revisited on a defined schedule and re-run against the same prompt set where possible. Updates note what changed in the surface, not only what changed in the numbers.

Next step

See what the data can tell you about search.

We can run the same measurement frame against your brand: what is indexed, how AI surfaces describe you today, and where the gaps sit.

Book a Strategy Call

No obligation. A senior strategist reviews your store before the call.