Spend & place patterns
How much, and how often, a household’s activity clusters around a product category. This is the baseline every profile is built on top of.
Data provider — Retail media
Shoppers don’t announce what they’ll buy next. Behaviour says it first.
CohortDesk builds retail and FMCG audience segments out of ordinary digital behaviour — what people read, compare, search for and where they spend their time — and models it into forward-looking propensity. No panels. No declarations. The output arrives as advertising IDs, in the platforms you already buy on.
Observed signal field
Modelled
cohort Propensity, not declaration
cohort Propensity, not declaration
TaxonomySix category pillars
CoverageRetail · FMCG · Pet
RecencyMeasured in days
OutputActivation-ready IDs
01 — Approach
A survey records what people say. We read what they do.
Stated-preference research asks a panel to describe itself, then freezes the answer for a quarter. Behaviour doesn’t hold still that long. We work from the signals people generate in the ordinary course of being online, and let the model keep up with them.
The survey model
Self-reported declarations from a panel — subjective, and rarely refreshed.
Static category framing, with no time dimension attached to it.
No read on the here-and-now need — only on a remembered one.
The CohortDesk model
Segments assembled from real online activity and the places people actually go.
Living signals — interest, frequency, in-market, intent — carrying a timestamp.
A current read on behaviour inside one specific product category.
02 — Method
How a profile gets built.
Nobody hands us a shopping list. We assemble one from the traces people leave behind, then model what those traces imply about the next few weeks. Four streams, correlated — each one weak on its own, and specific once they agree.
Places & device visits
Device-location signals that identify repeat visits to category-relevant venues — beauty retail, electronics, grocery — and how frequently they recur.
Browsing & search
The subjects people read about, compare and search for — measured by pattern and frequency rather than by any single visit, then matched to a branch of the taxonomy.
Correlation & modelling
The streams are correlated against each other. Where they agree, the model assigns a likelihood to buy within a category — and the profile enters a segment.
Output
The dataset resolves to specific advertising IDs, which can be used across most of the technologies available on the market. Segments are rebuilt continuously as behaviour moves.
Recency
Days, not months.
Refresh cadence varies by topic. Exact windows are listed against each segment in the taxonomy.
03 — Taxonomy
Six pillars. One index.
The taxonomy is organised into six pillars, each split into product sub-categories and then into individual segments. Below is the shape of it. The full index — segment names, volumes and availability by market — comes with the taxonomy pack.
01
FMCG & CPG
The taxonomy core. The most fluid, everyday shopping there is — and the foundation of campaign reach, built on high-frequency visits to real product pages.
Frequency
High intent
Sub-categories
02
Lifestyle & Dietary Values
Consumers defined by what they choose to stand for: organic, plant-based, eco, actively healthy. Clear, repeated signals in the better-for-you space — and premium rates to match.
Live intent
Premium
Sub-categories
03 Health & Personal Care The highest-margin end of the shelf. Prestige signals reach people while they are actively comparing premium offers, rather than after they’ve settled. High intent
Sub-categories
04
Life-Stage Triggers
The moments that rewrite a shopping list overnight — a pregnancy, a move, a wedding, a school year. Time-bound signals that land inside the decision window, not after it.
Real-time
In-market
Sub-categories
05 Brand & Retail Switchers A conquesting layer. People engaging with a competitor’s brand or a retail chain while the comparison is still open — captured by behaviour, not by claimed loyalty. Switching intent
Sub-categories
06
Pet Care
One of the fastest-growing categories in e-commerce. Full coverage of the owner’s journey — from nutrition through veterinary services to insurance — enriched with species and breed signals.
High intent
Breed interest
Sub-categories
Qualifiers
Every branch of the taxonomy is qualified by a signal type, so the same category can be bought at different depths of intent.
Interest
Frequent visitors
High-intent browsers
In-market
Real-time intent
04 — Activation
Ready to buy, wherever you buy.
Segments are built to be picked up by the stack you already run — nothing to integrate on your side beyond selecting them.
Delivery
Into your platform
Segments resolve to advertising IDs, which can be used across most of the technologies available on the market. Tell us what you run and the integration spec for your setup comes with the taxonomy.
Refresh
Kept current
Profiles are rebuilt as behaviour moves rather than on a quarterly cycle, so a segment reflects the window it claims to. Cadence is listed per segment.
Governance
Documented sourcing
Built to operate inside the industry’s consent and privacy frameworks. Sourcing documentation, the consent basis and the DPA are supplied alongside the taxonomy, before anything is signed.
Get in touch
Request the taxonomy.
We’ll send the full segment index, the sub-category breakdown, availability by market and the integration spec for your platform. One email from a person who knows the data. No drip sequence afterwards.
- 01The full segment index, pillar by pillar
- 02Sub-category breakdown and signal qualifiers
- 03Availability and volumes by market
- 04The integration spec for your platform
Prefer to talk
hello@cohortdesk.com