Personalisation
Personalisation: one campaign, as many messages as customers
The PBM Campaign Platform addresses every customer individually – with their own argument, their own channel, their own timing and the adviser they know. Individual for every recipient – but only with data that is approved for this purpose.
Rules decide. AI suggests. People approve.
Standard
Dear Sir or Madam,
-
Marlene HoffmannExample
Dear Ms Hoffmann, your money becomes available on 31 March – the tax side is worth it.
EmailSender: Adviser C. Reuter
-
Thomas BrandtExample
Dear Mr Brandt, your premium is changing – your cover stays.
LetterSender: Agentur Kaya
-
Jonas WeberExample
Hello Mr Weber, your appointment is confirmed – details via link.
SMS notificationSender: Customer service
Term
What is personalisation?
Personalisation is the automated, individual addressing of every single recipient: content, argument, image, channel, timing and sender are determined per person from their data and behaviour. Unlike classic personalisation, it is not just the salutation that changes but the entire message.
Put more simply: addressing every customer individually – without writing every letter individually. You define rules, and the platform applies them to thousands or millions of contacts.
Classic and attribute-level personalisation compared
| Classic personalisation | Attribute-level personalisation | |
|---|---|---|
| What changes | Salutation, name, perhaps a text module | Text, argument, image, offer, tone, channel, timing, sender |
| Data basis | Name and salutation | Master and contract data, behaviour in the journey, deadlines, segment and product logic |
| Channel | one channel, usually email | the right one for each customer: email, letter, landing page or SMS – depending on consent |
| Timing | one send date for everyone | when the occasion arises, for example 90 days before the end of the contract |
| Sender | the company | the responsible adviser or location |
| Variants | one message, name swapped | as many as rules and attributes produce |
| Control | hardly necessary | Rules, approval, consent check, audit log |
More in the glossary Introductory article on personalisation
What becomes individual
What exactly becomes individual with personalisation?
Nine elements of a message can be controlled per recipient. The example shows Marlene Hoffmann, 58, whose fixed-term deposit expires on 31 March.
-
Texts
The introduction names the occasion: “Your fixed-term deposit expires on 31 March.”
-
Images
Image variant per product or segment – in the corporate identity of the respective brand.
-
Offers & arguments
From a pool of approved arguments: security, flexibility or tax advantage.
-
Tone
Factual and explanatory for a premium adjustment, personal for an invitation.
-
Channel
Email if consent has been given, otherwise letter. SMS for short notifications.
-
Timing
When the occasion arises, such as 90 days before expiry – not on a single cut-off date for everyone.
-
Sender
The responsible adviser with name, photo and telephone number.
-
Landing page content
A personal page with the customer’s own options, protected by an access code.
-
Letter content
Each letter its own PDF: amount, argument, adviser, access code.
Data basis
What data is personalisation based on?
On sources that you already have or that arise in the journey – and on anything else you connect. A rule combines them into exactly one variant per customer.
-
Master and contract data
Product, term, amount, responsible adviser, consents. You define the attributes yourself per customer list – via file import or API.
-
Behaviour in the journey
Opened, clicked, landing page visited, which argument read? Measured via signed links and access codes, without third-party cookies.
-
Life events and deadlines
Contract expiry, end of tariff, end of lease – or an event from your core system, such as a claim notification, that starts a journey immediately. Life events only if the customer has told you about them themselves.
-
Segment and product logic
Dynamic segments, product rules and a pool of approved arguments. Whoever meets the condition today starts today.
-
Further internal & external data sources
PBM’s data pool can be fed from many sources: your own CDP and data stores within the company, data from external data providers (person and milieu data, household and building data and much more) as well as UTM and campaign parameters – via API or file import, usable directly as attributes in rules.
What a rule looks like
In the content editor, each field has its variants. A rule decides who gets which one – for example: portfolio value over €50,000 and tax-relevant → “tax advantage” argument. There is always a default variant so that every contact receives a complete message. The preview shows each variant as a specific customer will see it.
contact.depotwert > 50000 and contact.steuer_relevant Fields
- Subject
- Salutation
- Introduction
- Argument
- Image
- Sender
- CTA
Field · argument · 3 variants
Tax advantageActive in preview
contact.depotwert > 50000 and contact.steuer_relevantWith an amount like this, it pays to consider the tax side from the outset.
Security
contact.risikoprofil == "konservativ" FlexibilityDefault
default Preview as
Marlene Hoffmann
Dear Ms Hoffmann, your contract expires on 31 March …
With an amount like this, it pays to consider the tax side from the outset.
Before / after
What does the difference look like in practice?
Drag the slider. On the left, the standard message that everyone gets. On the right, the message the PBM Campaign Platform puts together for exactly this customer.
Regionalbank Musterstadt eG · Marlene Hoffmann, 58 · Example
Muster Leben AG · Thomas Brandt, 46 · Example
No email consent → automatically letter
Fictitious examples. You define texts, arguments and rules; every variant is checked and approved before sending.
Customer view
How does the customer experience personalisation?
Ms Hoffmann receives a message with her amount, her deadline, her adviser and a personal access code. On the lock page, she enters her date of birth and the code – without these two factors, the page shows nothing. She then sees three options and books an appointment with Christina Reuter directly.
The link contains no customer or contract number, only a signed token. Every access is logged.
Your personal overview
Data protection: no data is displayed without date of birth and code.
Ms Hoffmann, your options from 31 March
- Reinvest
- Pay out
- Advice
AI with approval
What role does AI play in personalisation?
The AI develops content. You make the decisions. The AI writes suggestions, rules determine who receives which variant, and a person approves.
-
AI step: Suggest variants. The AI drafts text variants, subject lines, image variants and translations – for a segment or for individual recipients. It works within your brand rules and only with arguments from your approved pool.
-
AI step: Check automatically. An AI-assisted compliance and tone check runs before every human approval and is always blocking – for example in the case of return promises or a missing mandatory notice.
-
Approval by people: Approve by a human. The standard is the four-eyes principle, adjustable per installation, brand or campaign. Each language version is approved individually.
Approved · four-eyes
All variants are created before sending and cached. At the time of sending, the rules engine decides, not a language model. The AI layer is optional and can be switched off.
AI suggestion · not yet accepted
Your fixed-term deposit expires on 31 March – secure your terms now before it’s too late. You will find three options in your personal overview.
Compliance & tone check
- no return promises
- Mandatory notice present
Tone: “too urgent”
Suggestion: “secure your terms now … before it’s too late” → “You can decide at your own pace how to proceed.”
Created by: J. Neumann · Approval by: C. Reuter (four-eyes)
logged · 25 September 2026 10:42
Privacy policy
How does personalisation remain compliant with data protection?
Through mechanisms, not promises. For us, data protection is not an add-on module but the architecture.
Only approved attributes. Personalisation uses only attributes that your team has created and approved for use. Rules and placeholders that access non-existent attributes do not pass the check.
Consent per channel, before every send. The compliance gate checks every contact. If email consent is missing, the letter goes out – if permitted. A withdrawal stops running journeys immediately and blocks re-entry.
No gaps in the message. If a mandatory value cannot be resolved – such as the responsible adviser – nothing is sent to the contact and the case is flagged during approval.
Traceable. Rules are deterministic: the same data produces the same variant. Approvals, AI actions and access to personal data are recorded in the audit log.
Data subject rights. Access, rectification, erasure and data portability run as guided processes.
The boundary
Where is the line between relevant and intrusive?
A message is relevant if the customer knows the occasion themselves. It becomes uncomfortable when it shows knowledge they would not have expected us to have. Our standard: relevant, not intrusive.
What we deliberately do not do
No live texts on opening. No content is created unchecked at the moment of delivery.
No sending without a consent check. The compliance gate cannot be switched off.
No customer or contract numbers in links. Personal pages only open with a signed token and access code.
No tracking via third-party cookies. Measurement uses our own signed links.
No advertising audiences without advertising consent. Identifiers are hashed before transfer; a withdrawal removes the contact immediately.
No language model in the sending path. A rule decides who gets which variant.
No sensitive life circumstances as a hook. In consulting and examples, we work with occasions the customer knows about – deadlines, contracts, their own enquiries. Not with illness, financial hardship or family events that they have never disclosed.
The read-aloud test
Before a variant is approved, four questions help: Does the customer know the occasion? Do they understand where we got the information from? Does the message help them make a decision? Could their adviser read it aloud to them without it becoming uncomfortable?
Control
What do you control yourself?
You set everything that makes up personalisation in the interface – without a developer ticket.
Your control panel
-
Attributes which data is available in rules and texts Approved
-
Argument pool which arguments may be used at all Active
-
Approval Four-eyes principle, per installation, brand or campaign Four-eyes
-
AI on or off On · can be switched off
-
Contact rules Upper limits, minimum intervals, quiet periods Active
-
Adviser intervention Advisers remove individual customers from a campaign Permitted
Contact frequency
The optimal mix of contact frequency and relevance
As often as useful, as relevant as possible – managed across all channels. If you know the occasion, you don’t need a scattergun approach.
The PBM Campaign Platform counts the contacts per customer across all campaigns and channels. You set the upper limits, minimum intervals and quiet periods. If two campaigns run at the same time, the more important one takes priority; the other waits.
Example: Ms Hoffmann is in the reinvestment campaign and the autumn newsletter. Reinvestment takes priority. The newsletter waits until it is completed.
- Day 2: Reinvestment Email · Announcement
- Day 9: Reinvestment Letter · Reminder
- Day 12: Autumn newsletter Waiting
- Day 16: Goal achieved Appointment booked
- Day 19: Autumn newsletter sent after completion
For SMEs
Does personalisation work without a data science team?
Yes. Personalisation in the PBM Campaign Platform is based on rules, not on your own data models. If you can maintain a customer list, you can write a rule.
Three steps to get started Example plan
-
Choose an occasion – for example: the maintenance contract expires in three months.
-
Use one attribute – for example the age of the machine.
-
Define variants – or have the AI suggest them and approve them.
“Customers with machines older than eight years receive the modernisation argument.”
contact.maschinen_alter > 8“Contacts whose machine age is greater than 8.”
MatchesFrequently asked questions about personalisation
What is the difference between classic and attribute-level personalisation?
Classic personalisation usually swaps the salutation and name. With attribute-level personalisation, the entire message is determined per recipient: text, argument, image, channel, timing and sender. It is based on contract data, deadlines and behaviour in the journey.
What data does personalisation need?
Less than is often assumed. Master and contract data such as product, term, amount and responsible adviser, plus the consents per channel, are usually sufficient. Added to this is behaviour in the journey, such as a visit to the personal landing page. Special categories such as health data are not required.
Is personalisation compatible with the GDPR?
Yes, if there is a legal basis and the mechanisms are right. The PBM Campaign Platform uses only approved attributes, checks consent per channel before every send, stops immediately on withdrawal and logs approvals and access in the audit log. This does not replace a legal assessment of your individual case.
How much effort does personalisation involve?
The effort lies mainly in preparation: defining attributes, formulating arguments, clarifying approvals. After that, the platform applies the rules automatically to every contact. It makes sense to start with one occasion and two or three variants per field.
Who approves AI-generated content?
A human. Every AI suggestion first passes through an automatic compliance and tone-of-voice check and then a human approval, by default according to the four-eyes principle. Each language version is approved individually, and every approval is recorded in the audit log.
Does personalisation work for SMEs without a data science team?
Yes. Personalisation is based on rules that you can formulate in everyday language; the platform translates them into a rule expression and back again for checking. The AI also suggests the column mapping during import and initial texts.
Doesn’t personalisation quickly come across as intrusive?
Only if it reveals knowledge the customer does not expect. The PBM Campaign Platform works with occasions the customer knows about, such as contract deadlines. Contact rules limit the number of messages, and personal pages are protected by an access code.