The five data questions that decide a lifecycle project before it starts

The five data questions that decide a lifecycle project before it starts

Jiří Šefler - founder of AJ advert
Jiří Šefler
  • November 20, 2026

Most lifecycle projects start with a whiteboard. Journeys down the left, channels across the top, a timeline underneath: welcome, post-purchase, replenishment, win-back. Someone draws arrows. Someone else asks whether the SMS should come before or after the second email. The data gets one line in the plan: integration, week three.

Week three goes well. The shop connects, orders arrive, contacts sync, and the first journey goes live on schedule.

What the plan doesn’t show is that five decisions were made in week three. Which facts each journey reads. Which system wins when two of them disagree. How old a fact can be. When two records count as one customer. Which permission a message relies on. Each of them decides which customers a journey reaches, and when. Each of them got an answer. Mostly by default.

Every lifecycle project answers these questions, on purpose or by default

A default is a decision someone made for the average business. Sync overnight. Match customers on email. Let the latest update win. For a brand with one shop and one market these are sensible choices, and often the right ones.

They stop fitting quietly as the brand grows. The ERP starts handling returns. A second market brings a second shop. Customer service starts taking orders by phone. None of these changes is a mistake, and each one moves the answer to one of the five questions without anyone deciding it again.

The kick-off is the cheapest moment to answer them on purpose. Nothing is built yet, the people who look after each system can still be asked in one meeting, and changing an answer costs a sentence rather than a rebuild.

One journey to follow: replenishment at a pet-food brand

To keep this concrete, follow one journey through all five questions. It is a modelled example, not a client case.

A pet-food brand sells dry food online in 2 kg and 12 kg bags. Orders come in through the shop. The ERP handles invoicing, stock and returns. Manago AI runs email and SMS. The team wants a replenishment journey: a reminder a few days before a bag is likely to run out, by email first, and by SMS for customers who opted in.

On the whiteboard it is one box and one arrow. Underneath, it depends on facts held in three different places.

1. What does this journey read, and where does each piece live?

Start with the journey, not the systems. Write down every fact the reminder needs in order to be right: which bag was bought, how long that bag lasts, when it actually arrived, whether any of it went back, and whether the customer agreed to hear from you by SMS.

Then write next to each fact where it lives today. For the pet-food brand, bag size and how long it lasts sit in the product catalogue. The delivery date and any return sit in the ERP. SMS consent sits in Manago AI.

The default answer is to build on whatever has already reached the marketing platform. Here that is the order date from the shop, so the countdown starts when the order is placed, not when the bag lands on the doorstep. For a 2 kg bag delivered the next day the difference is small. For a 12 kg bag held at a depot for a week, the reminder arrives while the customer still has a week’s worth left.

Write down: the facts this journey reads, and where each one lives.

2. When two systems disagree, which one wins?

Most facts a journey reads exist in more than one system, and sooner or later two versions differ. The shop marks an order complete once it is paid. The ERP knows that one of the two bags arrived damaged and was refunded. Both are right about what they saw.

The default answer is that the most recent update wins, so the outcome depends on which sync happened to run last. Some weeks the journey sees the refund, some weeks it doesn’t.

The deliberate answer is one writer per field. The ERP writes order status and returns; the shop and Manago AI read them. A correction goes back to the system that owns the field instead of being patched further down the line. And each field gets a person, not a department, who decides what happens when two versions differ.

Write down: for each fact, the one system allowed to write it, and the person who looks after it.

3. How fresh does each piece need to be?

This is the question teams are least used to asking, because the answer usually comes from a sync schedule set once, during the integration.

A nightly sync is a perfectly good answer for a weekly newsletter segment. It is a poor one for a journey that fires shortly after something changes. If a customer cancels a repeat order at seven in the morning and the cancellation reaches the marketing platform with the overnight sync, the reminder scheduled for nine goes out on yesterday’s picture.

The default answer is the integration’s schedule. The deliberate answer ties freshness to the decision rather than to the system: how old can this fact be before the journey acts on something that is no longer true? For replenishment, product data can be a week old. Returns and cancellations need to arrive within hours. The two answers rarely match, which is the point of asking per fact.

Write down: for each fact, how old it can be for this journey to still be right.

4. How do you know it’s the same customer?

The brand’s most loyal customers rarely behave like one tidy record. Someone orders as a guest with a work email, then creates an account with a personal one. Customer service takes a phone order and enters it in the ERP under the name on the invoice. Another member of the household orders the next bag.

The default answer is to match on email. It works most of the time, and it quietly splits these customers in two, each half carrying half the order history. For replenishment, that means a reminder timed off the wrong bag, or sent to someone who reordered last week from their other address.

The deliberate answer starts with a stable customer ID that every integration carries, with email and phone as supporting evidence rather than the only key. Two records become one when a second identifier agrees or a person confirms it. A merge never carries over a permission the customer did not give.

Write down: the key that identifies a customer across the shop, the ERP and Manago AI, and the rule for when two records become one.

5. Which permission allows this message, and how quickly does a withdrawal reach every sender?

Permission for the reminder email and permission for the reminder SMS are two different things, often collected in different places. The SMS opt-in might come from the checkout. The withdrawal might come as a reply to a text, or as a call to customer service that ends up as a note in the ERP.

The default answer is that each system keeps its own record. That works until someone withdraws in one place and the reminder goes out from another.

The deliberate answer names one place of record for each permission, with a timestamp, a source and a scope, and a window within which a withdrawal reaches every system that can send. Of the five questions, this is the one where an answer is not optional.

Write down: where each permission is recorded, and how long a withdrawal may take to reach every sender.

Five answers on one page

Here is what the pet-food brand’s kick-off produced for the replenishment journey. The middle column is what would have happened anyway.

Question The default answer The deliberate answer Who looks after it
1. What the journey reads Whatever has already reached the marketing platform Bag size and days of use (catalogue), delivery date and returns (ERP), SMS consent (Manago AI) CRM lead
2. Which system wins The latest update ERP writes order status and returns; shop and Manago AI read them Ecommerce manager
3. How fresh The integration’s sync schedule Returns and cancellations within hours; product data within a week Integration owner
4. Same customer A matching email Internal customer ID in every integration; email and phone as evidence CRM lead, with the data owner
5. Which permission Each system keeps its own record One record per permission; withdrawals reach every sender within a stated window CRM lead, with the DPO advising

Deciding these answers takes an afternoon with the right people in the room. Knowing what your own systems answer today takes more than a meeting. Someone has to check which facts the journey really reads, which integration writes each one, and where the shop, the ERP and Manago AI hold different versions of the same fact. That work happens in exports and field-by-field comparisons, and it is the part the whiteboard has no box for.

We have written that part down step by step in Blueprint 01 · Finding the five answers, with a one-page worksheet for writing down what you decide.

Where the answers live in Manago AI

None of the five answers needs new software. Each one has a natural home in the platform the journey runs on.

  • The customer key. Email is the primary identifier of a Manago AI contact, and each contact can also carry an External ID. Putting your internal customer ID there gives every integration the same reference for a person, even when the email changes.

  • What was bought. Orders sent as external events with their products let a journey read which bag was bought, not only that an order happened.

  • Values that change. The expected run-out date moves with every order, so it belongs in a date dictionary detail, which a workflow can use as a date-proximity trigger. Standard details are better kept for information that stays put.

  • Permissions. Manago AI records email marketing consent with its source, alongside mobile marketing consent, and stores custom consents with the date and the version the customer agreed to. Preferences declared in the Preference Center are kept as Key Information, ready for segments and workflow conditions.

When the five answers are written down first, these are configuration choices. When they are not, the same settings still get chosen, only without anyone deciding them.

True on launch day

Answer the five questions at kick-off and the journey launches on facts you chose rather than facts you inherited. That holds on launch day.

Then the brand keeps moving. A second market opens. The ERP changes how it exports orders. A new colleague adds a field that looks just like an old one. Each change can move one of the five answers without anyone deciding it again, which is exactly where this article started.

The next article in this series picks up from there: every field your lifecycle journeys rely on needs four sentences, and the fourth is the one that tells you when the other three stop being true.

Know what your next journey stands on before it launches

Blueprint 01 takes each of the five questions through Manago AI, the shop and the ERP, or wherever fulfilment and finance live in your business: where to look, what to export, which fields to compare and what a mismatch looks like. Five chapters, 22 exports and 23 field pairs, with Worksheet 01 for writing the answers down. Free.

Get Blueprint 01 →

Jiří Šefler - founder of AJ advert
Jiří Šefler
Founder at AJ advert

Founder of AJ advert, a lifecycle architecture consultancy, and a Manago AI partner since 2019, that builds Klints, which keeps the data behind lifecycle journeys consistent across the shop, the ERP and Manago AI.

Latest posts

WhatsApp isn’t just for friends any more: a real guide to WhatsApp marketing
November 18, 2026

WhatsApp isn’t just for friends any more: a real guide to WhatsApp marketing

Think about who’s in your WhatsApp right now. Your mum, probably. A family group still arguing about last week’s politics. The friend who only sends voice notes. And somewhere between them, a shop you once bought something from.

Read more
Martech trends for November 2026: what AI rules out, what Google keeps and what shoppers remember
November 12, 2026

Martech trends for November 2026: what AI rules out, what Google keeps and what shoppers remember

Black Friday has always been where marketing plans meet reality. This year, more of that meeting happens before a shopper reaches the site, in places that report back very little about what was decided there. The signals marketers still collect are getting thinner and harder to interpret, and more of them now come with conditions attached. Shoppers, meanwhile, increa...

Read more
Ready for the rush? Black Friday 2026 ideas for your eCommerce store
November 6, 2026

Ready for the rush? Black Friday 2026 ideas for your eCommerce store

Black Friday falls on 27 November this year, which leaves most eCommerce teams about three weeks. Three weeks is enough to fix the things that usually break at peak: a slow checkout, a segment nobody finalised, a discount that would not survive a legal check, a support inbox with no answers prepared.

Read more
Read more

Make every customer interaction count

Start your free trial and use AI to turn customer data into personalised experiences that drive growth.

Start Free Trial

No credit card required.

All together. Better.

Leadoo AI, Manago AI and Thulium AI are one seamlessly integrated family.

The #1 conversion platform for turning website visitors into active leads with its conversational marketing tools.

Explore Leadoo

Engage, convert and retain your customers with Europe's only AI Customer Data and Engagement Platform built for eCommerce.

Get a demo

Thulium is an omnichannel contact centre system for eCommerce & B2B support that unifies all your customer support.

Explore Thulium