Retail
Ask your shoppers anything.
Get answers in seconds
A synthetic version of your shoppers, built from your customer data.
Test price moves, range changes and propositions before they hit the shelf
What Electric Twin does
Electric Twin turns your customer data into a model of your shoppers.
Ask them anything and get honest answers in about a minute, just like your real customers would give.
No panels, no fieldwork, no six-week wait.
Accurate, independently validated, with up to a 96% match (1-MAE) against real-world surveys.
Electric Twin platform
Your shoppers, always on
This is what it looks like. A question in plain English, get research-grade answers in about a minute.
Understand the ‘why’ behind every number. We give you evidence that stands up when the room asks you ‘how you know’.
merlin.electrictwin.com
We’re moving own-label beef mince from £3.49 to £3.79. Ask @Loyalty_Members whether they absorb it, trade down, or switch store.
M
Merlin
Absorb the rise, stay own-label
62%
Trade down within store
27%
Price-check a rival for the basket
11%
Watch the 11% - they don’t trade down, they take the whole basket to a rival. Ask why, and every answer shows its reasoning.
Ask your shoppers anything…
How the audience is built
Built from data you already have
A synthetic audience is not a audience. It’s a live audience of your customers that you can ask anything, then ask again. We validate against a holdout of real respondents, so it holds up when you take it to the board.
Your data: loyalty, transaction and basket data
Validated against a holdout of real respondents before go-live
Onboarding takes days, not months. Your data stays yours.
@Loyalty_Members
Synthetic audience
Respondents
3,100 synthetic shoppers
Built from
Loyalty and transaction data, UK retail dataset
Validation
up to 96% match (1-MAE) against a real-respondent holdout
Ready to ask
Ask this audience
What retailers test
From price moves to propositions
Built for the teams the business turns to before it commits - insight, trading and brand.
Ask @Loyalty_Members how a 30p own-label price rise lands
Pricing and creative testing
Test price moves and the creative that carries them before they hit the shelf. See who absorbs, who trades down, and why - evidence that stands up when the room asks how you know.
Ask @Weekly_Shoppers what they’d buy if we cut the range by a third
Range and assortment
Test delists, swaps and new lines against the shoppers who actually buy the category, before the reset - not after the sales dip.
Ask @Weekly_Shoppers which of these two propositions they’d believe from us
Proposition and message testing
Validate a new proposition and the positioning behind it with your own shoppers - which claims land, which read as spin, and what they’d need to see to believe it.
Research at the speed of your trading cycle
60s
to responses, instead of 8 weeks
Unlimited
questions and iterations, at no extra cost
50k evals
across 155 countries, continuously benchmarked
Up to 96%
match rate (1-MAE) against real-world data
Why not the usual routes
The honest comparison
A panel survey
Accurate, but slow
Six to eight weeks of fieldwork per question set, and a budget to match. Electric Twin matches panel accuracy and answers in about a minute. Keep the panel for the studies that truly need real respondents.
6-8 weeks, or 60 seconds
A generic LLM
Fast, but it has never met your customers
An LLM alone answers from the internet, not from your shelves - useful for drafting, blind on your shoppers. Modelled from your own data, Electric Twin reaches up to 96% (1-MAE) against real-world answers.
Gut feel
Free, until it isn’t
Experience is a fine compass and a poor pricing model. The price change hits the shelf either way. The difference is whether you tested it on your shoppers first.
£0 today. Expensive after the reset
Proof from the platform
“From rationing research to running it on demand.”
10x
research volume
92%
accuracy (NDAM) vs a real panel, in The Times validation
How it works
Your shoppers. Your audience
Step 1
Build your audience
Synthetic shoppers built from your loyalty and transaction data. No panels, no recruitment.
Step 2
Ask the questions
Type them like you would to a colleague. No survey scripting, no fieldwork window.
Step 3
Get your answers
Research-grade results in seconds, benchmarked against real-world data.
Step 4
Dig deeper
Follow up on any answer and drill into segments until you have the why.
