What is the difference between server-side tagging and server-side tracking?
Server-side tagging relays events that still start in the browser, while server-side tracking sends events that start in your backend, such as a paid order or a closed deal in your CRM.
With a server-side GTM container, the browser fires one request to your own subdomain and the container forwards it to GA4, Google Ads and Meta. That gives you speed and control over what each vendor sees. But if the browser tag never fires, there is nothing to forward.
Backend events don't have that weakness. The purchase exists because the order exists, not because a thank-you page loaded. The catch is that backend events carry fewer browser signals, so match quality depends on the hashed email, phone and click IDs you capture.
Most of the builds we ship use both, joined by one shared event ID so platforms merge the two copies instead of counting twice. If you're unsure what you have today, start with a conversion tracking audit.
Sources: Google Tag Manager: Server-side, Meta: Handling duplicate Pixel and Conversions API events
Do I need server-side tracking on Shopify, or are the built-in apps enough?
For many Shopify stores the native Google and Meta sales channel apps are enough to start, because they handle the standard purchase events for you and the Meta app can send them through the Conversions API, not only the browser pixel.
Where they fall short is control. You get Shopify's event definitions, Shopify's values and limited say over what data goes where. That's fine for a store selling one product line on two platforms. It gets messy when you need margin instead of revenue as the conversion value, custom events, several ad platforms that should all agree, or checkout extensions that the apps don't cover well.
A quick check: compare last month's orders in Shopify admin with purchases reported in Meta Events Manager and Google Ads. If the gap is small and the values match, leave it alone. If it isn't, or the same order shows up twice, that's when a server-side GTM layer earns its hosting bill. We cover the store side in our Shopify work.
Sources: Shopify Help Center: Pixels and customer events
Can server-side tracking send offline and CRM conversions back to the ad platforms?
Yes, and for lead generation businesses this is often where server-side work pays off most, because the sale happens days or weeks after the form fill.
The pattern is simple. Capture the click ID and the lead's email or phone when the form is submitted, store them with the lead in your CRM, then send a conversion back when the lead becomes qualified or closes. Google supports this through offline conversion imports and enhanced conversions for leads. Meta accepts the same kind of event through the Conversions API.
Why it matters: if the only conversion your bidding sees is a form fill, the algorithm learns to buy form fills, including spam and tire kickers. When it sees which leads turned into revenue, it learns to buy more of those. I'd rather feed an account fewer, truer conversions than a pile of cheap ones.
The hard part is rarely the API. It's getting sales to mark lead stages consistently.
Sources: Google Ads Help: About offline conversion imports, Google Ads Help: About enhanced conversions for leads
How do I know if my server-side tracking is actually working?
You know it's working when platform-reported conversions line up closely with your real orders or leads, each one counted once, and match quality improves after launch.
There are four checks we run, and you can run them too. First, reconcile: pick a recent week and compare orders in your store or CRM against purchases in GA4, Google Ads and Meta. Second, check deduplication: in Meta Events Manager, browser and server events for the same purchase should share an event ID and merge. Third, look at Event Match Quality in Meta and the enhanced conversions diagnostics in Google Ads. Fourth, confirm consent: decline the banner in a private window and make sure server tags respect it.
If someone hands you a server-side setup and can't show a before-and-after on those numbers, you don't know what you bought. Send the numbers to us and we'll look at them during a tracking audit.
Sources: Meta Business Help Center: About Event Match Quality, Google Ads Help: About enhanced conversions
Will server-side tracking improve my ROAS?
Server-side tracking doesn't create sales, so it won't raise ROAS on its own, but it often improves the ROAS you can act on by giving bidding a more complete and accurate signal.
Two things happen after a good implementation. Reported conversions go up because fewer are lost in the browser, so the ROAS number moves even though nothing about demand changed. Then, over the following weeks, Smart Bidding and Meta's delivery have better data to learn from, which is where real efficiency gains come from, if they come.
Be careful with the first effect. If your team sees a jump the week after launch and calls it a win, you're celebrating a measurement change. We mark the launch date in reporting and compare against orders, not just platform numbers.
If your accounts are already starved for conversions, cleaner data helps more. See how we think about this in bidding strategy.
What can you do with server-side event data beyond recovering lost conversions?
Once events flow through infrastructure you control, you can enrich them, segment them and build audiences around them, which is often worth more than the recovered conversions.
A few examples from accounts I work in. Sending profit or product category with each purchase, so bidding values a high-margin order above a low-margin one. Passing lead stage from the CRM so Meta and Google see quality, not volume. And building custom events from the page view data you already collect.
That last one is underused. In Meta Events Manager, the sample events for a page view show the URLs people came through. With a clean Conversions API setup, you can build custom events around those paths, for example people who arrived from another paid channel and then bought through Meta. That tells you a lot about how channels hand off to each other.
None of this works if the base events are wrong, so we always start with the plan in measurement.
Sources: Meta Business Help Center: About Event Match Quality
