2026 AMRA Brochure - Product brochure - Page 19
BBC - From Segmentation to Conversation:
AI at the BBC
The BBC transformed its audience segmentation through AI.
AMI, its new persona chatbot, was built with users, powered
by rich internal data, and shaped by real conversations with
the UK public. It’s helping teams make faster, smarter, more
audience-centric decisions.
Judges’ comment
“Audience segmentations can unlock powerful insight for
editorial teams, but it can be cumbersome and challenging
to surface the insight editorial teams need in the timeframe
they’re working to. Enter AMI (Audience Map Interactive)
not a chatbot, but a strategic assistant - adopting a
Retrieval-Augmented Generation (RAG) framework, the
model summarises curated insights which guards against
hallucination, but allows for deeper, instant integration of the
existing segmentation. Used weekly by over 50 colleagues it’s
become a trusted companion for teams working on everything
from brand strategy to editorial planning.”
JCDecaux UK & System1 - Starting Brand
Revolutions through Creative Testing
Optix uses AI trained on eye tracking, neuroscience and
surveys to score OOH creative for Attention, Dwell and Clarity
in real street scenes. Validated against 140 brand lifts and
used by 540+ brands, it drives faster decisions, clearer ads
and stronger brand impact. Judges’ comment
“Creativity is critical to effectiveness, yet testing can be time
consuming and costly – JC Decaux and System 1 set out to build
an AI ‘Optix’ to run iterative testing on OOH creative and ensure
the most effective version is running. Fusing attention science,
neuroscience and survey learning with scene-level (real world
context) simulation, they tested 2,000 creatives to train the AI.
So far 540+ brands have optimised their creative, bene昀椀tting
from the proven correlation of improved full funnel impact
from Optix pre-testing.”
the7stars - ClaireTalk: Turning Data into Dialogue
ITV - ITV Content Similarity
A dual-lens AI tool that transforms how ITV connects
audiences with content. Built in-house, blending behavioural
and thematic data, the Content Similarity model empowers
teams across ITV to make data driven decisions - turning
machine learning into measurable, everyday impact.
Judges’ comment
“In a world where VOD platforms offer viewers more hours of
content than is possible to watch in a lifetime (back-to-back
couch zombie style), surfacing the right content to viewers
becomes critical. ITV built a Content Similarity tool using two
complementary Machine Learning methodologies; viewing
similarity (using in-depth content viewing patterns) and
meta-data similarity using natural language processing.
The resulting output enabled ITV to enhance content discovery
through bespoke ITV content rails, optimise CRM, improve
forecasting and advise on linear scheduling decisions.”
To support brand growth and improve audience
understanding, there needed to be a more effective way to
make the wealth of data studies and audience segmentations
on Iceland consumers accessible to stakeholders — enabling
faster, more informed decision making in a fast-paced, multiagency business. Introducing ClaireTalks, an AI powered,
conversational insight tool designed to bring Iceland’s
segmentations to life. comment
“Sitting on 昀椀ve years of audience intelligence across Iceland’s
core customers, the7stars needed to continue to keep the
consumer at the heart of decision making. Developed entirely
in-house, ClaireTalk was created, built on claimed, behavioural
and synthetic data - it mimics a one-to-one conversation with
Iceland’s core audience. ClaireTalk in昀氀uenced Iceland’s return
to TV after a year-long hiatus and led to changes in campaign
messaging, replacing “mums” with “we” in the iconic slogan
“That’s why we go to Iceland.”
19
