3×Add-to-basket lift
Personalized bundle recommendations, served in real time from Currys' own Google Cloud environment.
Marketing AI · Customer Intelligence
One governed profile per customer. In your own cloud.
Unify every customer signal into governed profiles, segments, predictions and live context, ready for every team and agent. Built in the warehouse you already run, so nothing is copied out.
Web visitsApp eventsEmailLoyaltyStore salesService callsIn your warehouse
EC
Unified profile
Emily Carter
Lifetime valueHigh
Propensity: coats0.82
Churn riskLow
On site now · 1 item in basket
1 of 6 sources linked
Trusted by enterprise brands
Runs in your cloud: BigQuery · Snowflake · Redshift · Azure Synapse · Trino
The problem
Every tool has its own customer. None of them agree.
The CRM, email, loyalty and service systems each hold a different version of the same person. Every new use case means another copy, another pipeline and another privacy review.
Customers aren't recognized
Cookies, devices, emails, loyalty numbers and store purchases are never linked, so most visits stay anonymous.
No single version of the truth
Each system holds a conflicting record, and nobody owns the master.
Store and service data never arrive
Every source needs its own pipeline, so offline customers stay invisible.
Rules don't travel with the data
Consent and PII rules are enforced tool by tool, so new use cases stall in review.
Data arrives too late
Overnight refreshes miss the moment, and real time needs yet another store.
Every tool keeps its own copy
Warehouse and egress costs climb with every platform you add.
What's inside
Seven capabilities. One governed profile.
Prepare the customer once, in your warehouse. Then use it everywhere: in audiences, in models and in the live session.
Prepare once
Data ingestion
Every customer source, batch and streaming, through one managed layer: web, app, store, loyalty, service and orders.
Fixes: Store and service data never arrive
Data transformation
One governed, analysis-ready customer model, with consent and PII rules applied once for every use.
Fixes: Rules don't travel with the data
Identity resolution
Cookies, devices, emails, loyalty numbers and store purchases linked to one person, with household, account and probabilistic matching.
Fixes: Customers aren't recognized
Unified profiles
One golden profile per customer, not one per system, built inside your warehouse.
Fixes: No single version of the truth
Use everywhere
Segmentation
Audiences built in place on live profiles and model scores, in minutes, with no export to refresh.
Fixes: Audiences days out of date
Customer AI models
Propensity, lifetime value and churn models, trained on full behavioral history and scored where the data lives.
Fixes: Raw data that never becomes signal
Real-time customer context
The live basket, session and loyalty state, served on request with no extra operational store.
Fixes: Data arrives too late
Served to any agent or channel through MCP servers, connectors, agents and APIs. Profiles, models and weights stay inside your boundary.
Identity in action
Six records. One customer.
Watch scattered identifiers become one golden profile, scored and live, without a row leaving your warehouse.
- Web · cookie 7f3a…c21WEB
- App · device iOS-22B1APP
- Email · e.carter@…ESP
- Loyalty · #4471-209CRM
- Store · receipt, AustinPOS
- Service · call logCX
EC0 of 6 linked
Golden profile
Emily Carter
- Identity
- One person, one household
- Consent
- Email and push, applied once
- Lifetime value · churn risk
- High · low
- Propensity
- Winter coats, 0.82
- Live now
- On site, 1 item in basket
Ready for audiencesReady for agentsReady in session
Put it to work
One profile. Four ways to win with it.
Customer 360
One view of every customer, online and offline: identity resolution, golden records and profile history.
22M Golden Records
Audiences
Build audiences in minutes, in your warehouse, not in a copy of it. Segments, model scores and the Audience Agent.
$5M spend optimized
In-session marketing
Act on shopper intent while the visit is still happening, with the next best offer decided in the session.
3× add-to-basket
Cross-channel activation
Send the right audience and offer to every channel you use, with suppression kept consistent everywhere.
Ads, email, web and app
Better with product context
Who is ready to buy, and what they can buy.
A profile tells you who is ready. Product Intelligence tells you what is in stock, in their size, at a healthy margin. Syntasa joins both inside your own cloud, so audiences, offers and agents never push what you can't sell.
Explore Product IntelligenceWho buys
Propensity, coats0.82, ready to buy
Size, from past ordersMedium
Live sessionBrowsing jackets
Joined in your cloud
What they can buy
Live price$149, no discount needed
Stock in her sizeMedium, 12 left
MarginHealthy
The offer Emily sees, in sessionWinter puffer, size M · in stock · full priceSent
Illustrative example.
Built for both teams
Marketing gets speed. Data keeps control.
For marketing, CRM and ecommerce
- Audiences in minutes,without a data-team ticket.
- Store and service customers visible,not just web visitors.
- Act in the session,while intent is still high.
- Ask in plain Englishwith the Customer Insights Agent.
For data, platform and security
- Zero copy.No shadow database, no egress.
- One governed model,with consent and PII rules applied once.
- Your cloud, your keys, your governance.Models and weights stay yours.
- Runs on what you have:BigQuery, Snowflake, Redshift, Azure Synapse, Trino.
Proof
Customer intelligence, with receipts.
$5MAnnual display-ad savings
Algorithmic retargeting replaced rules-based audiences, now scaled across Lenovo's global marketing programs.
500×Faster clickstream processing
Clickstream processing moved off a legacy Hadoop environment onto NOW TV's own BigQuery, from ingestion to insight in a fraction of the time.
Trusted
Why Syntasa
A customer view you don't have to copy.
| Dimension | Typical packaged CDP | Syntasa Customer Intelligence |
|---|---|---|
| Where the profile lives | In the vendor's cloud, copied from yours | In your warehouse, zero copy |
| Governance | Rules re-applied in each tool | Consent and PII rules applied once |
| Models | Generic scores, or none | Propensity, value and churn on your full history; weights you own |
| Freshness | Daily or weekly syncs | Live basket, session and loyalty state on request |
| Product context | None | Joined with live price and stock from Product Intelligence |
| Adoption | Replace what you have | Modular: feeds the tools you keep |
FAQ
Questions buyers ask.
It does the job of a composable CDP: identity, unified profiles, segments and activation. The difference is that it runs inside your own warehouse rather than in a vendor’s copy, and it adds customer AI models and live context.
No. Syntasa deploys into the cloud account you already run and builds profiles where the data lives. Only results, such as an audience or a decision, go out to your channels.
Cookies, device and app IDs, emails, phone numbers, loyalty and account numbers, and store transactions, with household, account and probabilistic matching.
Yes. Audiences and decisions sync to the tools you already use, including Google Ads, Meta, Braze, Salesforce and Klaviyo, and any agent can query the profile through MCP servers and APIs.
Streaming sources land continuously, and real-time customer context serves the live basket, session and loyalty state on request.
Customers have gone live in as little as a couple of months, depending on scope and data readiness. A fixed-scope pilot runs eight weeks: connect, serve, then measure on one domain and one channel.
Your customers stay yours. The results move.
See Customer Intelligence on your own warehouse, with your own security review.