Stape vs Addingwell vs Self-Hosted Server-Side GTM: An Honest 2026 Comparison

Server-side Google Tag Manager has gone from "interesting privacy upgrade" to table stakes for any brand spending real money on ads. With 40%+ of sessions blocked by ad blockers in key markets, third-party cookies fully gone from Chrome, and Safari ITP capping first-party cookies at 7 days, the question in 2026 is no longer whether to run server-side tagging — it is how to host it.

The three serious options are:

Managed SaaS hosting — Stape, Addingwell, Jentis, and a handful of smaller players that run your sGTM container on their infrastructure for a monthly fee.
Self-hosted on cloud infrastructure — Cloud Run, Cloud Compute, AWS App Runner, or Kubernetes, configured and operated by your team or a specialist.
Conversion API gateways — Meta's own Conversion API Gateway, which is a different architecture but shows up on the same shortlist.

Most comparison content online is either a Stape sales page in disguise or a generic "5 things to consider" listicle. This is neither. At TagSpecialist, we build server-side tagging both ways — we deploy clients on Stape when it fits, on Addingwell when it fits, and increasingly on self-hosted Cloud Run when the lock-in or scaling math doesn't work for managed SaaS. Below is what that decision actually looks like in 2026, with honest pros and cons of each.

Pricing note: Vendor pricing changes. Numbers below reflect publicly listed pricing as of mid-2026 — confirm on the vendor's site before quoting them in a board deck. The architectural and feature differences are more durable than the dollar figures.

What "Server-Side GTM Hosting" Actually Means

Every tool on this list does the same minimum job:

Run a Google Tag Manager Server Container (the official gcr.io/cloud-tagging-... image, or a fork) on infrastructure you don't have to manage.
Map a custom domain (e.g., analytics.yourbrand.com) to that container so events look like first-party requests to the user's browser.
Provide an admin dashboard to monitor request volume, errors, and costs.

What separates them — and where the price differences come from — is everything around that core: which conversion APIs are pre-built as templates, how easy it is to deploy a custom transformation, what regions you can host in, what monitoring you get out of the box, and how much engineering time the tool saves you per month.

Option 1: Stape (Managed SaaS, Market Leader)

Stape is the dominant managed sGTM hosting provider in 2026. Originally launched in 2020-2021 as essentially "Heroku for sGTM," it has grown into a full ecosystem with paid Power-Ups, the Stape Store of templates, and managed services for everything from Meta CAPI to TikTok Events API.

Architecture. Stape runs your sGTM container on shared multi-tenant infrastructure (with isolated containers per customer). You get a *.stape.io subdomain by default, and you can map a custom domain — typically a CNAME to a Stape-provided endpoint. Behind the scenes Stape uses a mix of cloud providers and proprietary routing to optimize latency and cost.

Pricing (mid-2026, verify on stape.io):

Free tier (~10K requests/month, single container, EU or US region) — enough to test
Hobby/starter tiers in the $20-50/month range, suitable for sites with up to ~1M requests/month
Business tiers in the $100-200/month range, multi-container, multi-region, priority support
Custom/Enterprise pricing for high-volume advertisers

Power-Ups (paid add-ons) typically cost $5-15/month each and add capabilities like enhanced Meta CAPI deduplication, TikTok CAPI templates, server-side consent helpers, and so on. A typical mid-market deployment ends up at $50-150/month all-in once you add the Power-Ups you actually need.

Where Stape shines:

Fastest time-to-deployment for a stock setup. You can have sGTM + Meta CAPI + GA4 running in under an hour.
Excellent template ecosystem. The Stape Store has pre-built templates for almost every ad platform and CMP, often with a free tier.
Strong documentation and a real community.
Active product development — new templates and Power-Ups ship monthly.

Where Stape's limits show up:

Lock-in risk. Migrating off Stape is non-trivial because your custom domain DNS, your Power-Up configurations, and your internal Stape-specific templates all have to be re-pointed or rebuilt. We've helped clients migrate off; it takes 1-2 weeks of careful work.
Pricing scales unfavorably past 5-10M requests/month. At that scale, self-hosting on Cloud Run frequently costs less.
Power-Ups are à la carte. The advertised hobby price doesn't include the Power-Ups most production setups need; the all-in price is closer to the next tier up.
Multi-tenant noise. We've seen latency spikes during peak hours that wouldn't happen on dedicated infrastructure. Rare, but documented.
Strict EU data residency for enterprise EEA customers sometimes requires an architecture Stape can't easily provide.

When to choose Stape: You want sGTM running by next week, you don't have a DevOps team, your traffic is under ~5M requests/month, and you're willing to live with the lock-in tradeoff.

Option 2: Addingwell (Managed SaaS, EU-First)

Addingwell is a French startup that has carved out a position as the GDPR-first, EU-focused alternative to Stape. They're newer (founded 2023-2024), smaller, and have positioned themselves around privacy compliance, simpler pricing, and EU data residency.

Architecture. Like Stape, Addingwell runs your sGTM container on managed infrastructure with custom-domain support. The notable difference is that Addingwell defaults to EU regions and emphasizes data-residency guarantees that matter for EEA customers under GDPR.

Pricing (mid-2026, verify on addingwell.com): Addingwell publishes a free tier of 100K requests/month and pay-as-you-go pricing above that. Specific tier pricing isn't always rendered on their public pricing page, so the most accurate read is to request a quote. From clients we've worked with, the typical paid tier sits in the €19-99/month range, scaling with request volume similarly to Stape but with simpler tier breakpoints and fewer add-ons.

Where Addingwell shines:

EU-first positioning. If your DPO is asking pointed questions about where sGTM data is processed, Addingwell's answers are cleaner than Stape's by default.
Simpler pricing. Fewer tiers, fewer Power-Up add-ons, fewer surprise bills.
Strong CMP integration. Their built-in patterns for Cookiebot, Didomi, and Axeptio (popular in France) are noticeably more polished than the average Stape Power-Up.
Smaller, more responsive support team — often the founders or core engineers respond directly.

Where Addingwell's limits show up:

Smaller template/Power-Up ecosystem than Stape. If you need a niche conversion API integration, you may end up building it yourself.
Less mature for very high volume. Addingwell is solid for typical mid-market sites; for enterprise scale (50M+ requests/month) they're less battle-tested than Stape.
US-region availability is supported but not their default story. If your traffic is primarily US and you want US data residency, validate region options before committing.
Younger company — startup risk. Lower than vaporware risk, but worth modeling into your decision if your tracking is mission-critical.

When to choose Addingwell: You're an EU-based brand, GDPR is a first-class concern, you want a managed solution but with simpler pricing than Stape, and you don't need exotic conversion API integrations on day one.

Option 3: Jentis (Premium Managed, Enterprise EU)

Jentis sits at a different price point and a different positioning from Stape and Addingwell. They are a premium, enterprise-focused server-side data capture platform — closer to Tealium or Segment in market position than to Stape — with their own data orchestration layer above sGTM.

Pricing. Custom enterprise pricing only. From client conversations, typical engagements start at €1,500-3,000+/month and scale up significantly. Priced for enterprise.

Where Jentis shines: Strong DACH and EU enterprise market presence, compliance-first architecture with strong audit trails, and more than just sGTM — a full data capture and orchestration layer above the container.

Where Jentis's limits show up: Price excludes most mid-market brands, and vendor lock-in is higher than Stape/Addingwell because Jentis adds proprietary orchestration above the standard sGTM image.

When to choose Jentis: Enterprise EU brand with serious budget and compliance posture, where Stape feels too DIY and self-hosting is off the table.

Option 4: Self-Hosted on Cloud Run (or Equivalent)

The Google-recommended path, and the one most underestimated. The official sGTM Docker image (gcr.io/cloud-tagging-...) is available for free, supported by Google, and runs natively on Cloud Run, Cloud Compute, GKE, AWS Fargate, or any container platform.

Architecture. A single Cloud Run service running the sGTM container, with a custom domain mapped via Cloud DNS or Cloud Load Balancing. Optionally extended with sidecar Cloud Run services for transformations, BigQuery sinks, or custom enrichment — the architecture detailed in our server-side GA4 pipeline post.

Cost. This is where self-hosting becomes interesting. A typical mid-market sGTM workload runs on Cloud Run for $30-100/month — including the container, custom-domain SSL, monitoring, and logs. At higher volumes (10M+ requests/month), Cloud Run still scales linearly while managed SaaS pricing tends to step-function up.

Where self-hosting shines:

Lowest infrastructure cost at any meaningful scale.
Zero lock-in. The container is the official Google image. Migrating between cloud providers is a Docker pull away.
Full control over the underlying environment — region, networking, data residency, IAM, secret management.
Native integration with the rest of your cloud stack (BigQuery, Pub/Sub, Cloud Storage, Cloud Logging) without going through a third party.
The same infrastructure can host the sidecar enrichment services described in our pipeline posts.

Where self-hosting's limits show up:

Setup is real engineering work. Without a specialist, expect 40-80 hours of work to deploy correctly the first time, including DNS, SSL, monitoring, alerting, and Consent Mode v2 wiring.
Ongoing maintenance is yours. When sGTM v3.4 ships and changes the GA4 client, you're the one updating the image, redeploying, and validating that nothing broke.
No pre-built Power-Ups. Every conversion API integration is configured in the GTM UI itself (same as managed SaaS) but with no shortcut templates beyond Google's official ones.
Monitoring is roll-your-own. Cloud Logging gets you 80% of the way; the last 20% (dashboards, alerts, error budgets) is your team's responsibility.

When to choose self-hosted: You spend $50K+/month on ads, your traffic is over 5M requests/month, you have a DevOps function (or an engineering specialist) to maintain it, and the engineering cost is dominated by the cost savings vs. managed SaaS at your scale. Or — you're an EEA brand with strict data residency requirements that no managed vendor can guarantee.

Option 5: Conversion API Gateways (Different Architecture)

Worth mentioning because they often come up in the same shortlist. Meta's Conversion API Gateway is a different architecture — it's not an sGTM replacement, it's a Meta-specific server-to-server pipeline that runs on AWS in your account. It only sends data to Meta. It does not replace GA4, Google Ads, TikTok, or any other destination.

In 2026, CAPI Gateway is most useful as a complement to sGTM (or to a managed sGTM), not a replacement. If you only care about Meta and you're already on AWS, it can be a clean solution. For multi-platform conversion tracking, it's incomplete on its own.

The Hidden Cost: Engineering Time

Most comparisons stop at vendor pricing and miss the larger cost: the engineering time to operate any of these solutions. A rough breakdown:

Option	Setup time (hours)	Ongoing time (hours/month)	Vendor cost ($/mo)
Stape (with templates)	4-12	2-4	$50-200
Addingwell	4-12	2-4	€30-100
Jentis	20-40 (vendor-led)	4-8	€1,500+
Self-hosted Cloud Run	40-80	4-8	$30-100
Self-hosted + custom transformations	80-160	8-16	$50-300

At a US engineering loaded cost of $150-200/hour, the real monthly cost of any of these solutions is dominated by the engineering hours, not the vendor invoice. Stape at $150/month plus 4 hours of internal engineering at $200/hour is $950/month; self-hosted Cloud Run at $50/month plus 8 hours is $1,650/month. The breakeven flips at higher traffic volumes where Stape's per-request pricing dwarfs the engineering delta.

This is the framing that matters: don't optimize for vendor cost — optimize for total cost of ownership. Most mid-market brands save money on Stape or Addingwell despite the higher vendor invoice, because the engineering hours saved are worth more than the savings on Cloud Run. Brands with their own engineering capacity (or a TagSpecialist on retainer) flip the equation.

Decision Framework

graph TD
    A[Need server-side tagging] --> B{Do you have an engineer<br/>who can operate it?}
    B -->|No| C{EU-first / GDPR sensitive?}
    B -->|Yes, internal team or specialist| D{Traffic > 5M req/mo?}
    C -->|Yes| E[Addingwell]
    C -->|No| F[Stape]
    C -->|Enterprise EU| G[Jentis]
    D -->|No, < 5M req/mo| H{Engineering hours expensive?}
    D -->|Yes, > 5M req/mo| I[Self-hosted Cloud Run]
    H -->|Yes| F
    H -->|No| I
    I --> J[Optionally: + Cloud Run sidecar<br/>for transformations]

The framework collapses to three questions:

Do you have engineering capacity to operate sGTM? No → managed SaaS. Yes → self-hosted is on the table.
What's your traffic? Above ~5M requests/month, managed SaaS pricing usually crosses over self-hosted total cost.
What's your compliance posture? EU-first → Addingwell or Jentis. US/global → Stape or self-hosted.

Common Mistakes Teams Make Choosing

Optimizing for the headline price. Stape's $20/month tier is not the price for a production setup; the all-in cost with the Power-Ups you'll actually use is closer to $80-150/month. Same on the self-hosted side: $30/month Cloud Run cost ignores the 40-80 hours of setup engineering. Compare apples to apples.
Not pricing the lock-in. All managed SaaS solutions involve some lock-in. Migrating off Stape or Addingwell is doable but takes 1-2 weeks of work. If the brand is going through a major data architecture rebuild in the next 12-18 months, self-hosted is more migration-friendly than it looks.
Underestimating consent gating complexity. Whichever hosting you pick, you still need to wire Consent Mode v2 and gate every non-Google tag explicitly — that's the topic of our consent compliance post. Managed SaaS does not do consent gating for you; it gives you better defaults at best.
Choosing self-hosted without ops capacity. A self-hosted sGTM that breaks at 2 a.m. when nobody's watching is worse than a Stape sGTM that has the same outage but with Stape's on-call team responding. Pick self-hosted only if you have someone (internal or external) on-call.
Picking based on a single feature. "Stape has a [feature]" is not, by itself, a reason to pick Stape if the rest of the architecture is wrong for your scale. Make the decision on architecture and TCO; the features tend to be table stakes across all the serious options.

How TagSpecialist Helps

We don't sell hosting. We build server-side tagging infrastructure on whichever path makes sense for the brand — and we operate the result, either as a one-time engagement or on retainer.

A typical TagSpecialist engagement includes:

The decision itself. A 60-minute discovery call that runs the framework above against your traffic, team, and compliance posture and produces a documented recommendation.
Implementation on the chosen platform. Stape, Addingwell, or self-hosted Cloud Run — each with full conversion API integration (Meta CAPI, Google Ads Enhanced Conversions, TikTok Events API), Consent Mode v2 wiring, custom domain, and parallel-tracking validation against your existing client-side setup.
Documentation and handoff. Architecture docs, runbooks for the most common failure modes, and a quarterly re-audit cadence.
Optional managed retainer. From $150/month, we operate the infrastructure, handle vendor coordination, and ship updates as Google's sGTM image changes.

The fixed-price engagement runs 2-4 weeks for most clients. Cost ranges from $8,000 (Stape or Addingwell, single platform CAPI) to $18,500 (full self-hosted with Cloud Run sidecars and BigQuery sinks). See the GTM specialist pricing page for the full breakdown.

If you want a no-commitment scoping call that walks through this framework against your specific situation, book a 15-minute audit. For more on what changed in 2026, see Server-Side Tagging Best Practices 2026. For the consent compliance angle, see Consent-Compliant Server-Side Tracking in 2026.

The honest version of the buy-vs-build question in 2026: most mid-market brands should be on Stape or Addingwell and stop second-guessing the decision. A meaningful minority — high-volume, engineering-led, or compliance-sensitive — should be on self-hosted Cloud Run. Almost no one should be on both. Pick once, run it well, and revisit in 18-24 months when the landscape inevitably shifts again.

Stape vs Addingwell vs Self-Hosted Server-Side GTM: An Honest 2026 Comparison