Telefónica announced a study of adaptive advertising on Sept. 7, 2026 , incorporating interests, memories and familiar places to investigate emotional connections with brands. The economic question is which personal reference improves the explanation of an offer and which consumes attention without adding a reason to choose it.
The notice describes moderated, in-person interviews lasting 60 minutes in Madrid and Barcelona during September and October. A sensor headset will record sensory reactions in Barcelona. The announced compensation is a 50-euro Click&Gift gift card per session, rather than a total study budget. Participant numbers, stimulus allocation and purchase outcomes are not published.
Commercial relevance depends on what is personalised
Recognising a place can make a use case easier to understand; recalling an intimate memory can also displace the product from the advertisement. Both experiences might be intense, yet support different advertising decisions. The interview earns its value by recovering that distinction through the participant’s explanation instead of interpreting every reaction as a favourable brand connection.
The EAE Business School survey reported by Statista supplies earlier context unrelated to this experiment. The report published in February 2024 describes 1,000 consumers in Spain, with a profile aged 18 to 70: 56.9% expressed a preference for advice from a real person and 28.9% for notifications about products that interested them. That comparison places relevance alongside the form of interaction rather than solely alongside the amount of available data.
Spain: brand interaction includes personal service and relevance
Four statements about brand interaction. EAE report published in February 2024; survey of 1,000 consumers, profile aged 18 to 70.
Tap or hover for details. All values are also available in the data table.
Collapse notes
EAE Business School / Statista. Original Customer Experience Trends: sample and profile p. 52, values p. 62. Fieldwork dates, interview method and sampling frame are not specified. Statements can overlap and must not be added; their complements do not establish acceptance of personalisation. This does not measure Telefónica customers or advertising effectiveness.
Explore the data table
| Category | Participants expressing the statement |
|---|---|
| I like advice from a real person | 56.9% |
| I like notifications about products that interest me | 28.9% |
| I do not like personalised interactions | 12.9% |
| I do not mind purchase history or being known by name | 10.8% |
The 12.9% declaring that they did not want personalised interactions does not justify describing the remaining 87.1% as willing to accept them. The statements contain different conditions. Nor does knowing purchase history amount to using an intimate memory: the information requested and what someone expects in exchange change.
Acceptance varies with the data, before the advertisement’s response is measured
A separate study, conducted by Kantar for IAB Europe in November 2024 and reported by Statista, asked 10,500 internet users aged 16 or over in 12 European countries how they would feel if different data were used to personalise advertising, with their agreement. Spain contributed 1,000 respondents, but the following percentages describe the combined markets rather than Spain alone.
Twelve European markets: five responses to three types of data
Kantar / IAB Europe online survey, November 2024. Separate percentages for each response, without combining categories.
Tap or hover for details. Select a series above to show or hide it.
Collapse notes
10,500 internet users aged 16 or over in 12 European countries, including 1,000 in Spain. These figures aggregate the countries. The question describes hypothetical advertising use with the user’s agreement; an attitude is not legal consent. Original categories and rounding are preserved, so a row can sum to 99 or 101. Original report: chart 7, p. 19; methodology p. 40.
Explore the data table
| Category | Would easily agree | Would agree | Neutral | Would not want | Would emphatically not want |
|---|---|---|---|---|---|
| Previous online purchases | 8.0% | 33.0% | 30.0% | 21.0% | 8.0% |
| Activity on social media | 7.0% | 22.0% | 28.0% | 28.0% | 15.0% |
| Location within approximately 500 metres | 5.0% | 18.0% | 27.0% | 32.0% | 19.0% |
For purchase history, 33% selected “would agree” and 8% “would emphatically not want”. For approximate location, the same responses were 18% and 19%. These are two particular categories on a five-part scale, rather than total acceptance and refusal. The difference supports a relevant editorial question: how much personal detail an advertisement actually needs to explain the offer.
The percentages do not predict responses in Telefónica’s research room or campaign performance. They do explain why a general preference for relevant advertising cannot justify every type of information. Obtaining, explaining, transforming and reviewing each datum adds work. If a reference improves a use case that can be described with less detail, the more intimate version needs an additional contribution that compensates for its cost.
A recorded reaction needs an explanation
Telefónica’s notice does not identify the sensor type, recorded signal or interpretation scale. Assigning brain reading or automatic emotion recognition to the headset would add undocumented capabilities. Using the device only in Barcelona also makes the two cities’ experiences different. Any subsequent contrast would need to examine that instrumentation alongside stimuli and participants.
The 2019 review by Barrett and colleagues examines facial movements, rather than this headset. Its conceptual distinction remains useful: an observable signal and the emotion assigned to it are different objects. If somebody explains that a memory felt uncomfortable, the intensity of the recording does not turn that discomfort into brand preference.
Better creative work and better personal matching are different investments
An experiment by Matz and colleagues published in 2017 studied advertisements and audiences classified by extraversion for a British beauty retailer. It reached 3,129,993 users, recorded 10,346 clicks and produced 390 purchases over seven days. Congruent combinations presented a purchase odds ratio of 1.54, with a 95% interval from 1.25 to 1.90; the interaction for clicks was not significant. That does not mean a 54-percentage-point improvement or a measurement of memories among Telefónica customers.
The review by Perla and colleagues, published online in 2025 and in the 2026 volume , combined 41 studies. The subset identifying a trait-by-treatment interaction yielded r=0.009, with a 95% Bayesian credible interval from −0.005 to 0.029, based on three independent studies and 11 effects. The interval includes zero and r is not a sales rate. The Matz experiment was excluded from that estimate because of metric-conversion difficulties; the review is not a replication invalidating it.
These precedents investigate personality and digital footprints, rather than the design announced in Madrid and Barcelona. Their economic implication is more specific than a general promise of persuasion: comparing an adaptation with a weak generic advertisement can simply reward a better piece of creative work. Funding a system of versions requires identifying the contribution of personal matching as well as the quality of the explanation.
An interview can reveal which element helps someone understand the offer and which displaces it. A commercial outcome then requires another observation: a completed enquiry, a contract or another decision corresponding to the product. Preserving that sequence avoids purchasing recurring production on the evidence of a reaction. Adaptation deserves to continue when it strengthens a reason to choose that survives the session, even if it is less striking than the memory the participant recognises.
In this article
Marian Vasile
Marian Vasile is a MarTech and growth professional. His work spans product marketing, SEO, paid advertising and deep online tracking, including behavioral analytics and attribution.