Vercel acquired Splitbee on October 25, 2022, folding page-view tracking, visitor metrics, top-pages reporting, referrer analytics, and demographic data into Vercel Analytics. The founders, Tobias and Timo Lins, joined Vercel to build it. The pitch was simple: first-party data collection, GDPR compliance, no cookies, no third-party scripts, edge-collected, and free during beta. Three and a half years later, the acquisition is no longer notable on its own terms. What makes it worth revisiting is that Cloudflare has run the same playbook twice, most recently with the Astro Technology Company acquisition in January 2026, and the structural implications for standalone analytics vendors are now legible.
What Vercel bought and what it became
Splitbee was a lightweight web analytics startup. Its product tracked page views, visitors, top pages, top referrers, and basic demographics without cookies or third-party scripts. After the acquisition, Vercel rebranded the capability as Vercel Analytics and positioned it as a first-party, edge-collected, GDPR-compliant alternative to Google Analytics and standalone privacy-focused tools like Plausible.
The integration was architectural, not cosmetic. Because Vercel controls both the hosting runtime and the edge network, ingestion lands on a first-party endpoint on the same domain (/_vercel/insights) rather than a third-party tracker hostname. That first-party path is what underwrites the no-cookie, GDPR-friendly positioning. It does not, however, make collection invisible to ad blockers. [Updated June 2026] Vercel’s own docs are explicit that Web Analytics and Speed Insights “require scripts to do collection of data points” and that “these scripts are loaded on the client-side,” which means the same @vercel/analytics snippet a Plausible or Fathom user would recognize, just pointed at a same-origin route. The realistic claim is that first-party routing dodges the domain-blocklist rules that nuke google-analytics.com, not that it eliminates script-blocker loss entirely.
Web Analytics shipped to general availability on April 19, 2023, so the “free during beta” framing of the original announcement is stale. [Updated June 2026] As documented in Vercel’s current pricing reference (last revised June 16, 2026), it is metered by collected event, where an event is a page view or a custom event. Hobby teams get 50,000 events per month included and cannot buy more: collection pauses after a three-day grace period until the next cycle. Pro teams pay $0.03 per 1,000 events beyond their plan credit. The Web Analytics Plus add-on runs $10 per month per team and is the real upsell, buying UTM parameters, eight properties per custom event instead of two, and a 24-month reporting window against the 12 months a plain Pro plan retains.
The Cloudflare parallel
Cloudflare has run the same verticalization pattern twice. In December 2021, it acquired Zaraz, a third-party script manager that loads external tools through Cloudflare’s edge workers rather than injecting scripts into the browser. The functional overlap with what Vercel did with Splitbee is direct: both acquisitions pull telemetry and tool management inside the hosting boundary.
In January 2026, Cloudflare acquired the Astro Technology Company, the team behind the Astro web framework. This is a different layer of the stack, closer to Vercel’s ownership of Next.js than to analytics, but the structural pattern is the same: absorb an adjacent tool into the platform so that developers never leave it.
The edge-hosting market has performance dynamics that favor the bundlers. Cloudflare Pages scored a median global TTFB of 28ms in a March 2026 comparison, compared to 180-450ms for traditional shared hosting. Vercel scored 8.7/10 for Next.js and React hosting in the same comparison, against Cloudflare Pages at 9.1/10 (note: this comparison carries medium confidence). When the hosting layer is this fast, routing analytics through a first-party endpoint rather than a third-party tracker is a performance argument as much as a privacy one. (The same edge-runtime economics drive the Fluid Compute versus Cloudflare Workers comparison on the compute side.)
Bundled vs. best-of-breed: the coexistence that proves the limit
The Splitbee acquisition did not kill standalone analytics on Vercel’s platform. PostHog maintains an active Vercel Marketplace integration with over 1,000 installs, offering feature flags, A/B testing, and analytics. PostHog’s own integration docs cover Vercel Functions, a Flags SDK, AI SDK observability, and an MCP server for Vercel’s v0 product.
This is the co-opetition pattern that vendor blogs do not name: Vercel built Vercel Analytics to capture the long tail of teams that need basic page-view and visitor data, and it keeps PostHog in the marketplace for teams that need experiment analysis, feature flags, or product analytics depth. The hosting platform gets the easy revenue. The standalone vendor keeps the hard problems.
The strategic question for standalone vendors like Plausible is whether the market for “basic web analytics that is not bundled into hosting” is large enough to sustain them. Vercel Analytics and Cloudflare’s bundled telemetry are not feature-equivalent to these products, and the privacy-compliance positioning that standalone vendors built their brands on is now claimed by the hosting platforms as well. Differentiation has to come from analytical depth, cross-platform portability, or compliance features that bundlers cannot replicate without becoming analytics companies themselves.
Where the standalone tools still win
The competitive field is not one undifferentiated blob of “privacy analytics.” It splits along two faults: how the data is priced, and whether you can take it with you.
On price, the standalone tools are sometimes cheaper at small scale and sometimes not. Plausible’s Starter plan is $9 per month for 10,000 monthly events on a single site, where an event is a page view or a tracked custom event. Fathom Analytics sits in the same band, starting around $15 per month for 100,000 page views across unlimited sites. Vercel’s metered model is the inverse shape: nothing extra until you blow past the plan credit, then $0.03 per 1,000 events with no per-site cap. For a portfolio of small sites that collectively stay under the Hobby 50,000-event ceiling, Vercel is effectively free where Plausible bills per site. For one high-traffic site, Plausible’s flat tiers stop the meter that Vercel keeps running. The cost crossover is real and depends entirely on your traffic shape, which is exactly the comparison the “it’s bundled, so it’s free” pitch elides.
On portability, the standalone tools win on a single feature that bundlers structurally cannot match: export. Plausible, Fathom, and PostHog all let you pull raw events out through a stats API or a full data export, and PostHog will run self-hosted against your own Postgres and ClickHouse if data residency is the constraint. Vercel Web Analytics has no equivalent. There is no documented bulk-export of historical events, and the reporting window is a retention guarantee, not a portability one. When you leave Vercel, the dashboard goes dark and the history stays behind. That is the difference between renting a view of your data and owning the data, and it is the same pattern that runs through Vercel’s broader anti-lock-in messaging: the framework is open, the data path is not.
The depth gap runs the other direction. GA4 remains the free default for teams that will tolerate cookie banners and Google’s data handling in exchange for funnels, audiences, and BigQuery export at no cash cost. PostHog is the maximalist alternative: session replay, feature flags, experiment significance testing, and event autocapture, none of which Vercel Analytics attempts. Vercel’s own docs draw the line in the product itself. A page view or a track() custom event with at most eight properties is the entire data model. There is no funnel builder, no cohort retention curve, no replay. The honest framing is that Vercel Analytics competes with Plausible and Fathom on the “traffic dashboard” job, and does not compete with PostHog or GA4 on the “understand user behavior” job at all.
This is why the absorption pattern has a ceiling. A hosting platform can clone the easy 80% of a traffic counter because the data already flows through its edge. It cannot clone session replay or experiment analysis without hiring a data team and standing up a columnar event store, at which point it is no longer bundling analytics as a convenience feature, it is competing as an analytics company with all the support and roadmap costs that implies. Vercel has shown no appetite for that, and keeping PostHog in the marketplace is the tell.
The decision framework
The choice between bundled analytics and a standalone tool breaks down along four axes.
Depth of analysis. Bundled analytics handle page views, visitors, referrers, and geographic breakdowns. They do not typically include funnel analysis, cohort retention, event-based tracking with custom properties, or statistical experiment analysis. If you need any of those, a standalone product is not optional.
Compliance surface. First-party, cookie-free, edge-collected telemetry reduces GDPR and ePrivacy compliance work. Vercel and Cloudflare both lead with this in their positioning. Standalone tools like Plausible offer the same compliance posture but require a separate integration and a separate data processing agreement. The compliance argument favors the bundler because there is one fewer vendor to audit.
Vendor lock-in. This is the cost that bundlers do not advertise. Vercel Analytics data lives in Vercel. Cloudflare Analytics data lives in Cloudflare. If you move hosting providers, your historical analytics do not migrate with you. Standalone tools that send data via a standard endpoint or a server-side events API are portable across hosting changes. For teams that switch hosting providers every few years, this is the deciding factor.
Procurement overhead. Bundled analytics requires no new vendor evaluation, no contract negotiation, no additional billing relationship. For small teams and solo developers, this is often the only axis that matters. For organizations with procurement processes, the marginal cost of adding one more SaaS tool is frequently higher than the tool’s subscription price.
What this means for 2026
The hosting layer is absorbing the analytics layer. Vercel did it with Splitbee in 2022. Cloudflare did it with Zaraz in 2021 and continued the pattern with the Astro acquisition in 2026. The direction is clear even if the pace varies by platform.
For analytics vendors, competing on convenience against a hosting bundler is a losing position. The win has to come from analytical depth, cross-platform portability, or compliance features that the bundlers cannot replicate without becoming analytics companies themselves.
For teams building on Vercel or Cloudflare, the practical question is not which analytics tool is best. It is whether the bundled option covers enough of what you need that the procurement and integration cost of a standalone tool is not justified. For a growing share of teams, the answer is yes. For teams doing product analytics, running experiments, or building feature-flag-driven delivery, the bundled option is still a supplement, not a replacement.
Frequently Asked Questions
What does standalone analytics add to a Vercel hosting bill?
Vercel’s hosting tiers run from free to $20 per month. A standalone analytics tool on top of that (Plausible starts around €9/month; PostHog’s cloud tier scales with event volume) can double or triple a small team’s monthly infrastructure cost before accounting for integration effort and procurement overhead.
Do Cloudflare’s 2026 layoffs change the outlook for bundled tooling?
Cloudflare cut roughly 1,100 positions, or 20% of its 5,156-person workforce, in May 2026 while posting record Q1 revenue of $639.8 million and explicitly citing AI-driven role obsolescence. The practical effect for analytics bundling is that platform companies are redirecting labor savings into automating their own tooling stacks, which accelerates absorption of adjacent capabilities rather than pausing it.
What specific gaps appear with edge-collected analytics that teams don’t expect?
Edge-collected analytics capture request-level signals (URL, referrer, geolocation, user agent) but cannot track in-page interactions like button clicks, form submissions, or scroll depth without additional client-side instrumentation. Session recording and heatmapping are also outside the scope of edge collection entirely. Teams that need any of these capabilities will outgrow bundled analytics regardless of how well the hosting integration works.
When should a team on Cloudflare Pages skip bundled analytics entirely?
Teams running multi-platform deployments, such as a marketing site on Cloudflare Pages with an application backend on a separate provider, get an incomplete picture from Cloudflare’s bundled analytics because it only instruments traffic on Cloudflare’s network. A standalone tool that aggregates events across hosting environments avoids this blind spot. The same constraint applies to Vercel Analytics for teams splitting workloads across providers.
