Server-side tracking · Follow-up to the foundation · Fixed price
Part of your conversions never reaches Google and Meta. A server of your own for the measurement brings part of that back.
Ad blockers, Safari and the settings on the iPhone stop the measurement in the browser before it reaches Google or Meta. The order, enquiry or booking is real, but the ads never hear about it and learn from a picture full of holes.
With server-side tracking the report goes first to a server that belongs to you, and from there to Google, Meta and the other platforms. That is more stable, you control what leaves the house and the data stays in Germany. It does not repair broken measurement, which is why it comes after the foundation.
First the tracking check shows how big your gap is. Then you get a fixed price for the server if it pays off. No subscription, no fee per report.
What you end up with
- A server in Germany that belongs to you, fully set up
- Google Analytics, Google Ads and Meta get every conversion once, with value and reference
- The cookie decision is respected on the server too
- Manual, test results, backup and handover to your team
What we do differently
-
Foundation first, then the server
A server does not turn wrong numbers into right ones. We only build it on top of verified measurement.
-
A number before the purchase
We compare your real conversions with the measured ones. That gap decides, not an industry average.
-
No way around the cookie banner
Whoever declines is not measured through the server either. We do not sell a trick.
-
You own the system
Server, setup and manual sit with you. No fee to us per report, no term.
Why browser measurement has gaps
Four things stop the report before it arrives.
These are not exceptions but the normal case for a large share of your visitors.
-
Ad blockers
In Germany roughly every second person uses an ad blocker. It blocks not only ads but also the scripts that report a conversion to Google or Meta.
-
iPhone and app settings
Only about a quarter of users allow apps to track them. With a lot of iPhone traffic the ads miss a noticeable part of the attribution.
-
Safari deletes cookies
Safari limits cookies from the browser to seven days, often to one. Someone who returns after a week and buys looks like a new visitor without a campaign.
-
Third-party cookies are disappearing
Browsers keep restricting cookies from foreign providers. What remains is measurement through your own address and your own server.
How big your gap really is depends on consent rate, browser mix and platforms. That is why we measure it before you buy infrastructure.
Calculate the gapHow it works
The report goes to you first, then to Google and Meta.
Instead of the browser talking directly to Google or Meta, it talks to an address on your own domain. Behind it sits a server in Germany that checks, enriches and forwards the report.
Personal details such as email addresses are made unreadable before forwarding. The IP address is shortened. What must not be sent after a decline at the cookie banner is not sent by the server either.
-
The conversion happens
Order, enquiry, booking or call, on the website, in the shop or in the booking system.
-
Your server accepts the report
Through an address on your domain. To ad blockers that is no known tracker, to Safari it is traffic with the site the visitor is on anyway.
-
The server checks and cleans
Cookie decision, reference against double counting, value and currency. Personal data is made unreadable.
-
Forwarding to the platforms
Google Analytics, Google Ads, Meta, on request TikTok, LinkedIn and your CRM. Every platform gets exactly one report per conversion.
When it pays off
Not every website needs a server of its own.
Server-side tracking complements browser measurement where ad budget, gap and platforms justify the extra effort. With a small budget and a clean foundation it often does not.
Pays off when
- A noticeable ad budget depends on the measurement
- Several ad platforms are served, for example Google and Meta together
- The comparison shows many conversions lost to ad blockers and Safari
- Data should stay in Germany and you want to control what goes out
Does not pay off when
- · Browser measurement is not in order yet
- · The ad budget is small
- · The gap after the foundation is already small and explainable
- · Nobody wants to run the system later and no operation by us is wanted
How we decide: by your own number. Conversions from shop, form or CRM against the measured ones, plus consent rate, browser mix and number of platforms.
From the tracking check
What the shop sends and what arrives at Google.
On the left, a purchase the way the shop reports it technically. On the right, the same purchase the way Google Analytics receives it. The marked lines are the four most common faults we find in shops that grew over time.
dataLayer.push({ event: "purchase", ecommerce: { transaction_id: "SO-118342", value: 189.90, tax: 30.32, shipping: 4.90, currency: "EUR", items: [{ item_id: "SKU-4471", item_name: "Laufjacke Nordwind", price: 92.50, quantity: 2 }] }});- event purchase
- transaction_id (not set)
Without an order number the purchase counts again on every reload of the confirmation page.
- value "189,90"
The amount arrives as text with a comma. Google cannot use it and counts 0.
- tax 30.32
- shipping 4.90
- currency (not set)
Without a currency Google guesses and the revenue in the ad account is off.
- items []
No items: Google does not know what was bought.
Every finding is documented in the check with its location and the fix. Acceptance is a real test purchase, not the tag preview.
Request the tracking checkWhere the server runs
The first decision: your own server, Stape or Google Cloud.
Where the server runs determines cost, privacy and control. There is no answer that is right for everyone, only the one that fits your traffic and your requirements.
| Point | Stape or Google Cloud Run | Your own server at Hetzner in Germany |
|---|---|---|
| Contract partner | An additional provider in the data path, with Google Cloud a US company | You and a German provider, ISO 27001 certified data centres |
| Cost | Stape from about EUR 20 fixed per month, Cloud Run about 120 to 300 US dollars depending on volume | Fixed server cost by size, no billing per report |
| Privacy review | Transfer to third parties and to the US has to be reviewed and documented | Processing in the EU, shorter review, what goes to Google and Meta still has to be documented |
| Control | Limited, settings and logs sit with the provider | Full, you see and decide every step |
| Operation | The provider looks after the server | Needs operation. We take that on if you want |
Costs from public provider information. We recommend what fits you, not the variant with the highest margin.
What gets connected
The platforms that receive your conversions.
Three form the core of almost every setup. All of them get every conversion once, with value, currency and reference.
-
Google Analytics
The conversion arrives through the server, stable against ad blockers, with the right cookie decision and a clean value.
-
Google Ads
Conversions directly in the ad account, enriched with unreadable customer data so the attribution to the campaign improves.
-
Meta
Browser and server report the same conversion with the same reference. Meta counts it once. How good the matching gets is shown by the account after launch.
-
TikTok and LinkedIn
On request through the same path, without additional scripts in the browser.
-
Your CRM
Enquiries marked as qualified or won in the CRM go back into the ad account with their value.
-
Several websites
Website, shop, booking system and portal report to the same server. The visitor stays recognised across every switch.
Privacy
No way around the cookie banner.
Some sell server-side tracking as a way around consent. It is not. What the visitor declines at the banner is not measured through the server either. What it gives you is control over the data path.
-
Consent stays mandatory
Since May 2024 German law requires consent before cookies are set or read. Fines up to EUR 300,000. A server changes nothing about that.
-
Only send what is allowed
The cookie decision is checked on the server. Without consent nothing goes to Google or Meta, not even anonymised.
-
Data stays in Germany
Processing runs in data centres in Nuremberg or Falkenstein. What goes to Google and Meta leaves them by design, depending on the chosen connection.
-
Personal data is made unreadable
Email addresses and phone numbers are encrypted before forwarding so the platform can only match them, not read them. IP addresses are shortened.
-
Contracts with Google and Meta
Depending on the service you need a data processing agreement or an agreement on joint responsibility. That is not the same at Google and Meta.
-
Privacy policy
The services used, their purpose and possible transfers to the US have to be stated there. The texts stay with you and your advisers.
Practical assessment, not legal advice. Whether your specific setup is permissible is for you to clarify with your privacy advisers.
From the work
What finished projects look like.
Four anonymised cases from different industries, without an attached promise of results. How much becomes visible after a repair depends on consent, browsers, ad blockers and platforms.
-
Online fashion retail
Only half the purchases were visible
- Before
- Main site and shop on two systems with separate measurement paths, the order confirmation counted again on every reload.
- Done
- Measurement plan rewritten, one measurement for both domains, cookie decision and order numbers unified, double counting tested.
- After
- Purchases arrive in Google Analytics and in the ad account with value and a unique number. Accepted with a real test purchase.
No ad cost figures are published for this case.
-
Travel provider
Every booking looked like a direct visit
- Before
- The booking ran across three domains: website, offer flow and a third-party booking system. The connection broke on every switch, no campaign got credit.
- Done
- Handover between the domains set up, redirects cleaned and a return channel from the booking system through the server added.
- After
- A booking stays one visit across all three domains. The confirmation reaches the ad account with value and booking number.
Attribution was measured, not booking volume.
-
Manufacturer, B2B enquiries
Job applications counted as much as purchase enquiries
- Before
- The ads were optimised on form submissions. A large share were applications, complaints and price requests without intent to buy.
- Done
- Enquiry types separated in the form, qualification fed back from the CRM and imported into the ad account with tiered values.
- After
- The ad account sees qualified enquiries separately from the rest. The feedback runs on schedule and is logged.
No effect on close rate or revenue is claimed, the data path is described.
-
Subscription shop on Shopify
Two systems counted the same order differently
- Before
- After the cookie banner was introduced the numbers dropped, a tracking app installed in parallel reported different purchases.
- Done
- Cookie banner wired up properly, double capture switched off, first and follow-up order in the subscription distinguished.
- After
- One order creates one report with one reference. Behaviour on accept and decline tested.
Practical assessment, not legal advice.
How it runs
Switch-over without a measurement gap
The most dangerous phase is the switch itself. Whoever turns off browser measurement before the server measures cleanly tears a gap that the ads optimise on for weeks. That is why both paths run in parallel for a while.
-
tracking check, two working days
Measure the gap
Conversions from your system against the measured ones. Plus consent rate, browser mix and platforms. From that follows whether the server pays off.
-
fixed price
Decision and quote
Where the server runs, which platforms get connected, what your team takes on. You get a fixed price.
-
two to four weeks, complex six to twelve
Build
Server in Germany, address on your domain, platform connections, cookie check, protection against double counting.
-
one to two weeks
Parallel run
Browser and server measure at the same time. We compare the volumes per conversion and document where and why they differ.
-
one day, with a way back
Switch-over
Only when the difference is explainable and stable do we switch. The old state stays saved as a fallback.
-
your choice
Handover or operation
Manual, test results, before-and-after evidence. Then either handover to your team or ongoing operation with updates and monitoring by us.
Prices
The number first, then the fixed price.
The server is a follow-up project after the foundation. Its price depends on platforms, scope of the switch and hosting, so it comes as a fixed-price quote after the check.
Tracking check
EUR 290 net- Conversions from your system against the measured ones
- Consent rate, browser mix and platforms
- A clear statement whether a server pays off
- Fixed price for foundation and server, credited if you commission within 30 days
What the project includes
- from EUR 2,490 net
Server-side tracking
Server in Germany fully set up, address on your domain, Google Analytics, Google Ads and Meta, cookie check on the server, parallel run with before-and-after evidence, manual and handover.
- by quote
Further platforms
TikTok, LinkedIn, feedback from your CRM, several websites or markets.
Afterwards, optional
- by quote
Operation by us
Hosting, updates, security patches, monitoring and intervention on errors. The system still stays yours.
- from EUR 99 per month
Tracking monitoring
So an update or a changed platform interface does not quietly break the measurement.
One-off prices, no term, no fee to us per report. Server costs at Hetzner you pay directly and see on the invoice. All prices net.
Related
What goes together with the server.
-
The foundation first
Measurement plan, Google Analytics, ad account and cookie banner have to be right before the server delivers them more reliably.
See web tracking -
Wiring up the cookie banner properly
Only with consent or also without, without cookies. Both variants and what they mean for your numbers.
See Consent Mode -
Connecting Meta properly
Browser and server report to Meta without counting twice. What improves the matching and what does not.
See Meta Conversions API -
How big is your gap?
The calculator compares your conversions with the measured ones. And the SaaS graveyard shows with 736 entries why a system of your own is safer than a rented tool.
Calculate the gap
Common questions
What is asked most often before a server project.
What is server-side tracking in one sentence?
The report about a conversion does not go from the browser directly to Google or Meta, but first to a server that belongs to you, and from there onwards in a controlled way.
Do we need it if we only use Google Analytics?
Not necessarily. If you spend a noticeable ad budget or need reliable numbers for automated bidding, it pays off earlier than most think. If you only use Google Analytics to look things up, the foundation is usually enough.
Does it work against ad blockers?
Largely. Ad blockers block known addresses of foreign providers. Your own domain and your server are not on those lists. Whoever declines cookies still stays unmeasured.
Do we still need a cookie banner?
Yes. The server creates no legal basis. It respects the visitor’s decision, it does not bypass it.
How long does it take?
Simple setups are live in two to four weeks, complex switches need six to twelve weeks. Add the parallel run so no measurement gap appears.
When do we see results?
The numbers in the ad accounts usually change within two days of the switch. The effect on the campaigns shows after two to four weeks, once the platforms work with the more complete data.
What happens if you are gone?
You own server, setup and manual. Any developer can take it over. It is not maintenance-free: updates, changed interfaces and new banner requirements remain tasks, with you or through us.
Why not a rented tool?
Because then you depend on it: monthly costs, no control over the data path, and if the provider raises prices or shuts the product down you are back at the start. The SaaS graveyard shows how often that happens.
Start the tracking check
Four required answers, the rest is optional. You get a reply by email within one business day, no phone call.
Next step
Measure the gap first, then decide.
The tracking check shows within two working days how many of your conversions Google and Meta do not see and whether a server of your own pays off. If not, that is what it will say.
Credited if you commission within 30 days. No term, no obligation to implement.