Chapter 4
Requirements and pre-analysis
 "Zoran Artmagic")
Table of contents
Most failed switchovers do not fail because of the technology but before it: nobody has defined what should be measured, who needs which numbers and what the baseline was before the switch. This chapter is the groundwork that saves weeks later on.
What needs to be in place?
Seven requirements determine whether a switchover goes smoothly:
Hosting for the tracking server, for example TAGGRS or Google Cloud, with access to your domain's DNS settings for the subdomain.
Google Tag Manager and GA4 with administrator rights, plus access to every ad account that is to be connected.
A clean data layer in the store that delivers product views, cart, checkout and purchase with complete product data. In Shopware it comes from an extension, in Shopify from customer events.
A cookie banner with Consent Mode V2 that also passes its signals to the data layer.
Technical know-how for Tag Manager, server containers and the platforms' interfaces.
A data protection review of the planned data flows, internally or by an external data protection expert.
One responsible person who steers the project and owns maintenance after go-live.
If individual points are missing, that is no reason to give up, but it is a reason to clarify them before the switchover rather than halfway through.
Which teams need which numbers?
Tracking is not purely a marketing matter. Before deciding which events to capture, it pays to go round the teams:
Marketing: Conversions and revenue per channel and campaign, cost per order, new versus returning customers.
Sales and B2B: Inquiries, quote requests, lead sources.
Product management: Click paths, checkout drop-offs, use of filters and search.
IT and development: Loading times, tracking errors, browser and device split.
Customer service: Views of FAQ and contact pages, form abandonment.
These requirements become a measurement plan: a list of events, their parameters and the platforms that receive them. It is the most important document in the entire project.
Why is a baseline before the switch so important?
Without a before, there is no after. Record how many purchases GA4, Google Ads and Meta reported over the last few weeks, and put the actual orders from the store next to them. Also note which tags, pixels and apps are currently active. You will need this list later to switch off the old tracking completely, as Chapter 10 explains.
What to watch out for technically
Rate limits: Many platform APIs limit the number of requests per second. At peak times such as Black Week, the server has to absorb this.
Event deduplication: If a browser pixel and the server report the same conversion, both need the same event ID.
Multiple domains and languages: Shopware sales channels and Shopify Markets with their own domains need a well-thought-out plan for containers, subdomains and cookies.
Payment providers with redirects: PayPal, Klarna and others take the customer through external pages. Without a referral exclusion, they break the session and the purchase is credited to the payment provider.
The pre-analysis checklist
Are the tracking goals defined: which events should be captured with which parameters?
Is there administrator access to Google Tag Manager, GA4, all ad accounts and the domain's DNS settings?
Does the store deliver a complete e-commerce data layer, including in checkout?
Does the cookie banner output Consent Mode V2 and pass the status to the data layer?
Is it clear who handles the data protection review and who handles the technical implementation?
Is the baseline documented: measured purchases per platform next to actual orders?
Have the requirements of all teams involved been collected?
Frequently asked questions
For a typical store, a few hours to two days. Completeness matters more than duration: a missing event in the measurement plan costs many times more later on.
A structured data layer in the store that provides events such as product views or purchases with all their details. Tag Manager reads it instead of scraping data from the page. Without a complete data layer, any tracking will have gaps.