01
Recognise a returning customer
Where collection and matching are lawful, linked records can help distinguish repeat visits from new customers and measure the result of an offer.
On this page
- Value does not require a price tag on a person.
- Three different things can look similar in a diagram.
- How a small improvement can create value at scale.
- Why closed user groups deserve precise questions.
- What does a secure offer actually secure?
- Questions that make the flow reviewable.
- Sources and the limits of this explanation.
- Value comes from the connections.
- The information advantage: what the join reveals.
Follow the value as well as the flow
Why linked data is worth asking about.
An email address provides a contact point. A linked history can reveal what someone buys, how often they return and which offers they can use. The combination can support decisions that none of those records could support alone.
The question is how that extra knowledge was created, who can use it, and whether the person was told. Commercial value can exist without a database being sold.
More useful information can change a decision
Value does not require a price tag on a person.
02
Understand purchase patterns
A history of actual purchases can inform audience selection, promotion timing or product recommendations. Tesco Media publicly describes a business based on customer insight and campaign measurement. [1]
03
Reduce losses and unnecessary friction
Fraud detection has its own economic value. Stripe describes using device and activity risk factors to distinguish legitimate shoppers from fraudulent purchasers and bots. [3]
Explanation is not permission. A useful data combination still needs an appropriate purpose, lawful basis, transparency and safeguards. A security justification does not automatically explain a later marketing use. [4]
Tesco’s privacy notice describes coded customer matching, audience creation and aggregated campaign reporting. Those are examples of disclosed commercial uses; they are not evidence that a payment provider’s raw fraud telemetry is available to advertisers. [2]
Identify the field before debating the label
Three different things can look similar in a diagram.
| Information | What needs establishing | The useful question |
|---|---|---|
| Ordinary purchase and account records Order amount, item category or account identifier. | Which merchant collected them, which people they concern and which purposes were explained. | Which record entered the CRM, from which source and under whose instructions? |
| Security results visible to a merchant A risk outcome or decision, where a product makes it available. | The actual field, account permissions and product configuration at the relevant time. | What could this account see or export, and where is the evidence that it did? |
| Raw security telemetry Device or interaction signals held by the provider. | Whether the merchant could access those specific fields at all. A reporting tool’s existence does not establish access. | Show the schema, field lineage, permissions and sender/recipient logs for the alleged transfer. |
These are analytical categories, not assertions that all three are exported by any particular service.
Stripe’s current documentation states that its advanced fraud-detection risk factors are not used for advertising or provided to advertisers. It also describes activity indicators as scoped to a shopping session on a single site or app. Examine the relevant records to identify which fields are accessible, which routes are configured and how any receiving system uses them. [3]
A merchant might also collect similar information independently through its own website or another script. Similar field names or overlapping collection times do not identify which collector supplied a later record.
Illustrative arithmetic · not a market valuation
How a small improvement can create value at scale.
This example models extra contribution from a campaign. It assumes the use is permitted and compares a baseline with an improved response rate. The numbers are invented and are not evidence of any supplier’s revenue.
The result is a business outcome under the assumptions shown. It is neither a sale price per profile nor an estimate of any actual damages.
People reached × (improved conversion − baseline conversion) × contribution per extra order
50 additional orders × £5.00 contribution.
Incremental contribution before data, campaign and compliance costs. The example does not establish causation in an actual campaign.
An eligibility fact can add context
Why closed user groups deserve precise questions.
In an employee or membership benefit scheme, eligibility may reveal an employer, an occupation or a service affiliation. If that fact is linked to purchase history, the resulting record may be more revealing than an ordinary retail account. What is actually verified, retained or shared depends on the scheme; a discount code alone does not establish a person’s job.
A shared CRM or multi-tenant vendor is not, by itself, evidence that two suppliers can see one another’s customers. Examine tenant boundaries, permissions, export destinations and the controller’s instructions. The concern becomes specific when a recipient can access or use information outside the permitted purpose.
If identifying and behavioural information is compromised, it can help make a deceptive approach more convincing. NCSC explains how information from breaches can be used in phishing messages and calls. This is a general security risk. [5]
Keep the origin attached to the inference. “Eligible member”, “frequent purchaser” and “suspected fraud” are different propositions. If one becomes another downstream, ask for the transformation rule, its source records and any review or correction.
Read the industry’s own explanation
What does a secure offer actually secure?
Planit’s June 2025 article describes verified audiences as useful for targeting, retention and customer value. Its security discussion centres on limiting offer leakage and affiliate fraud. That does not establish how member information is protected or reused. [7]
Access to an offer
Verification asks whether someone qualifies. The separate data questions are who retains the verification, what it reveals, who can join it to purchases and which later uses are permitted.
Credit for a purchase
The article uses an approximately 85% solo-contribution metric to claim incrementality. Impact defines Solo by the touchpoints in a conversion path. That alone does not identify what would have happened without the offer. [7] [8]
Analytical distinction: attribution assigns credit within an observed journey; a causal test estimates the difference made by an intervention. Experimental approaches can compare exposed and control groups. Ask which method supports a particular claim. [9]
Ask for the explanation and the supporting record
Questions that make the flow reviewable.
- Which exact field is involved: an order record, an inferred category, a merchant-visible risk result or provider-held telemetry?
- Who collected it, and what links it to the person or transaction in question?
- Which purpose and lawful basis apply at each stage, including any change of purpose?
- Which accounts, suppliers or vendors could access it, and what do the access and export logs show?
- Which decisions used it, and how were its origin, uncertainty and any correction preserved?
- What measurable business benefit is claimed, and what evidence separates that benefit from coincidence or another cause?
Is enriched data automatically unlawful?
No. Enrichment describes combining or adding information. Legality depends on the particular collection, purposes, lawful basis, transparency and other applicable requirements. Potential commercial value does not supply legal permission. [4]
Does fraud prevention authorise commercial profiling?
A fraud-prevention explanation does not, by itself, establish permission for a different marketing purpose. Identify the information actually reused and assess the new purpose and applicable requirements. [4]
Does a shared platform prove that records were pooled?
No. A common vendor, reporting product or timestamp does not prove a transfer. Look for the actual identifiers, tenant permissions, field mappings and records of the transfer or joint access.
Can we put a fixed pound value on every profile?
Not from the sources used here. Value depends on the permitted use, accuracy, coverage, freshness, match quality, costs and demonstrated outcomes. The calculator shows transparent assumptions rather than an unsupported price list.
Does the failure to prevent fraud offence make every questionable CRM use a crime?
No. The Home Office guidance addresses a defined offence involving specified fraud by associated persons intended to benefit a qualifying organisation or its clients, with a reasonable-prevention-procedures defence. It does not establish that every data-protection concern is fraud. [6]
Primary sources checked 29 October 2026
Sources and the limits of this explanation.
- Tesco Media / dunnhumby — primary description of customer insight, audience engagement and measurement. A commercial example.
- Tesco privacy and cookies policy — describes purchase information, coded matching, audience uses and aggregated outputs.
- Stripe advanced fraud detection — provider explanation of device and activity risk factors and its stated advertising and session-scope restrictions. Documentation is an attributed statement, not an independent audit.
- ICO guidance on sharing information to prevent scams and fraud — responsibilities, lawful basis and purpose limitation. The ICO marks this guidance as under review following the Data (Use and Access) Act.
- NCSC data-breach guidance — explains how compromised information can support phishing and deceptive calls.
- Home Office guidance on failure to prevent fraud — scope of the offence and prevention procedures; in effect from 1 September 2025.
- Planit, Importance of Closed User Groups — published 26 June 2025, updated 23 July 2025. A marketing agency’s description of CUG benefits, not an audit of data protection.
- Impact Contribution Report documentation — definition of Solo and other journey metrics.
- AppsFlyer incrementality experiment documentation — an example of test and held-back control groups.
This page explains the value of linked information through educational examples and illustrative calculations.
Value comes from the connections.
A joined profile can be more useful for commercial decisions than its separate source records. Payment outputs, purchases, reviews and redemption events can contribute different attributes. Connectors, a warehouse and a CRM make those attributes available together, reducing the need to build every component from scratch.
The critical change is the use: a field created for a security check can become an input to audience selection or offer eligibility. Compare the two views to follow that change through the same tools.
The information advantage: what the join reveals.
Joined stock-purchase and resale records reveal relationships, timing and concentration. An analyst can build a supplier-and-channel graph and use a longer history to estimate demand or replenishment. Value comes from the business information created by the join.
Where the recipient also has a competing commercial interest, ask which non-public patterns it can see and whether those patterns influence offers, negotiations or approaches to customers. Follow access to the combined tables and the decisions made from them.