Example deliverable · technical diagnosis
Fictional example using synthetic data.
This document illustrates the approach and deliverable; it does not describe an actual client project.
What you receive at the end of a technical diagnosis: the question and its boundaries, what the evidence shows and what is still an assumption, the decisions that need an owner, and how the team validates and takes over the follow-up.
A
"We want to build audiences across our CRM, the webshop and the loyalty programme and use them in our marketing channels. Can our data support that, and what does the target architecture need to account for?"
The organisation is considering a CDP. Adobe Real-Time CDP is one of the options; no platform has been chosen.
Platform selection and vendor comparison, the implementation itself, a legal assessment of the consent texts, and cleansing of the source data.
B
Established means the evidence shows it directly. Hypothesis means it is a plausible explanation that still has to be checked. There are no urgency scores: the order in section C follows which decisions other decisions depend on.
All figures come from the synthetic sample.
| # | Observation | Evidence or assumption | Possible impact | Decision or follow-up needed |
|---|---|---|---|---|
| 1 | "Active customer" has three definitions: an order in the last 12 months (CRM), a login in the last 90 days (webshop), a points transaction in the last 6 months (loyalty). | Established E1, E4. Each definition fits the purpose of the team that uses it. The difference itself is deliberate and not an error. |
None, as long as each definition is used under its own name. | Keep the three definitions, give them distinct names in the documentation. |
| 2 | The marketing audience "active customers" adds up counts from all three systems without stating which definition applies. | Established E3. This is where the difference turns into inconsistent use. |
Audience sizes cannot be compared between campaigns; one person can be counted up to three times. | Marketing decides on one definition for cross-channel audiences, with the source fields named. |
| 3 | 31% of webshop accounts match a CRM record on email address; 18% of loyalty members have no CRM customer number. | Established E2. |
One customer ends up as several profiles; loyalty-only members fall outside CRM-based audiences. | Agree identity rules: which keys link records, in what order of priority, and what happens to guest orders. |
| 4 | The low match rate between webshop and CRM comes from guest checkouts and from customers using a different email address in each system. | Hypothesis E2 shows the gap, not its cause. Suggested by the product owner in E4. |
If the cause lies elsewhere, identity rules built on this assumption will not close the gap. | Check 200 unmatched records together with the e-commerce product owner before the identity rules are fixed. |
| 5 | Consent is recorded in two places, CRM opt-in and the webshop checkbox, with no rule for which one wins. 6% of matched customers have conflicting values. | Established E1, E2. Which system is leading is not documented. |
A customer who withdrew consent in one system can still be selected through the other. | Decide the leading source per purpose and channel. Legal confirms the basis; that assessment is outside this scope. |
| 6 | "Lapsed customers" includes people who now buy in store, where those purchases do not reach the webshop data. | Hypothesis Based on E4; not visible in the sample. |
Loyal store customers receive win-back campaigns. | Establish whether store sales reach the CRM, and with which key. |
A CDP can hold several definitions side by side and link identities, but it does not decide which definition is leading or which consent value wins. Findings 2, 3 and 5 need a decision whichever platform is chosen, Adobe Real-Time CDP included. Those decisions then become requirements for the platform selection: which identifiers, which channels, how fast an audience needs to update.
C
This example is fictional and not one of the case studies. Scope, evidence and deliverable are agreed per engagement.
