Meta Has Two Gateways Now, and One Is Not a Rename of the Other

By Jackson Kolinski · November 16, 2026 · 8 min read
Meta Has Two Gateways Now, and One Is Not a Rename of the Other

You stood up a Conversions API Gateway at some point, it has been quietly forwarding events ever since, and now every other post mentions something called Signals Gateway. The natural assumption is that Meta renamed the thing you already have. It did not. These are two products, they both exist, and Meta documents them on the same page.

That page is the one worth reading before anybody sells you a migration. What follows is what it says, what the difference means in practice, and how to tell which one your situation actually calls for.

What Meta’s documentation actually says

Meta’s developer documentation has a page titled “Gateway Products: Conversions API Gateway and Signals Gateway”. Both are listed, and it describes them as sharing a common set of components and infrastructure, which it collectively calls the Gateway Products.

The one-line descriptions are the useful part, because they are narrower than most of what is written about them:

  • Conversions API Gateway— send events from your new or existing Meta Pixels to your Conversions API Gateway, which can then forward them to the Meta Conversions API.
  • Signals Gateway— create flexible events pipelines that support various data sources and data destinations.

Read those next to each other and the shape of the difference is immediate. One of them has a fixed source and a fixed destination: pixel in, Meta Conversions API out. The other is a pipeline builder, with sources plural and destinations plural.

The difference that decides it

The Conversions API Gateway solves one problem: getting browser-side pixel events to Meta server-side without you building and maintaining a server integration. It is scoped to Meta because that is all it was ever meant to do, and for a lot of advertisers that scope is exactly right.

Signals Gateway is infrastructure for routing events generally. The phrase “various data sources and data destinations” is doing the work: more than one thing can feed it, and more than one place can receive what comes out. If your situation is one website sending purchases to Meta, that flexibility is capability you will pay for and not use. If your situation is several properties, or events that need to reach more than one platform, it is the product built for that.

So the question to ask is not “which is newer” or “which is better”. It is: do you have more than one source or more than one destination? If no, the gateway you already have is doing the job it was designed for and a migration buys you nothing.

They share a platform, which matters more than it sounds

Meta lists the components both products have in common: account management, host user management, host settings, a Control Plane API, and software updates. The architecture is described as a server application that receives incoming events, transforms them, and sends them to a destination depending on which pipeline type has been configured, plus a web UI for administering and monitoring the instance. Both products expose that admin UI at the same place, a /hub/ path on your gateway endpoint.

Two practical consequences follow. First, the operational knowledge transfers: if you already run a Conversions API Gateway, you are not learning a new system from scratch, you are configuring a different pipeline type on a platform whose administration you already know. Second, the thing that distinguishes the two products is largely which pipeline type is configured, which is a much less dramatic framing than “you need to replatform”.

What neither of them fixes

Gateways get sold on the implication that server-side delivery makes tracking reliable. It makes tracking more durable against the browser, which is a different claim, and three problems survive either product untouched.

Deduplication is still your job

If the pixel is still firing in the browser and the gateway is also sending the same event, you have two events and you need them deduplicated. A gateway forwarding pixel events does not absolve you of event IDs. This is the same trap it has always been, covered in fixing duplicate events between pixel and CAPI, and a gateway deployment is one of the commonest moments to introduce it.

Bad input is still bad output

Neither gateway invents customer data. If the events reaching it carry a weak set of match parameters, they will carry the same weak set on the way out, and your Event Match Quality will reflect that. Server-side delivery improves whether an event arrives, not what is in it.

You lose your usual way of checking

This is the one people underestimate. Once events are leaving a server rather than the browser, opening the page and watching the network tab stops telling you the truth, and Pixel Helper stops being a complete picture. The verification you relied on quietly becomes misleading rather than wrong, which is worse. Our existing piece on the Conversions API Gateway goes further into that, and it applies to both products equally.

Deciding, in order

  1. Count your destinations. Meta only? The Conversions API Gateway is the matching product. More than Meta? That is what Signals Gateway is for.
  2. Count your sources. One website is one source. Several properties, an app, or a server-side system that should feed the same pipeline is the case for the pipeline builder.
  3. Ask what breaks if you do nothing. A working gateway that delivers Meta events is not technical debt because a broader product exists. Migrations introduce duplicate events and gaps; do one for a reason you can name.
  4. Check Meta’s own page rather than a vendor’s. Most writing about this is published by companies that host gateways for a living, which is not a reason to distrust it, but is a reason to read the documentation first.

How to tell which one you are running

For an account you inherited, this is not obvious from the outside, and guessing from how long ago it was set up is unreliable. Both products expose the same admin interface at the /hub/ path on the gateway endpoint, so open that and look at what pipelines are configured. A single pixel-to-Meta flow is the Conversions API Gateway doing its job. A configuration with multiple sources or destinations is a Signals Gateway pipeline.

Two things to write down while you are in there, because nobody ever has them when they are needed later: which cloud account the instance actually runs in, and who pays for it. A gateway is a server, and the commonest way one dies is not a technical fault but a billing relationship that ended when somebody left the company.

If you do decide to move

Treat it as a tracking change rather than an infrastructure change, because the risk is not in the deployment, it is in the overlap.

  1. Expect a period where both are live. That is the dangerous window. Two pipelines sending the same events means duplicates unless event IDs are consistent across both, and a duplicate rate that starts during a migration is easy to attribute to the migration and hard to attribute to the specific cause.
  2. Record your baseline first. Event volume per type, Event Match Quality, and the gap between platform-reported conversions and your own order count. Without those numbers from before, you cannot tell whether the new pipeline is better, worse, or identical.
  3. Cut over one event type at a time if the account matters. Moving everything at once means a single number to compare and no way to localise a regression.
  4. Keep the old instance until you are sure, then decommission it deliberately. An abandoned gateway still running is a source of mystery duplicate events months later.

After you deploy either one

The failure mode for gateways is not a loud outage. It is an instance that stops forwarding while the pixel keeps firing, so volume drops by a share rather than to zero, and the dashboards still show numbers. Things worth checking on a schedule:

  • Event volume at the destination, compared against orders in your own system.
  • Event Match Quality, which moves when parameters stop being passed.
  • Duplicate rates after any deployment that touches the pixel or the pipeline.
  • That the instance is still running at all, which is a server somebody has to own.

That last one is the real difference between a gateway and a pixel: a pixel fails when the page changes, and a gateway can fail because an instance was never restarted after an update. Both are silent.

It is also worth sizing the commitment honestly before you take on either product. A gateway is infrastructure you are hosting, so there is a running cost, there are updates to apply, and there is a person who has to care about it. For a single advertiser sending web events to Meta, a well-implemented Conversions API integration through your existing tooling may reach the same outcome with nothing new to operate. The gateway earns its keep when you want server-side delivery without building that integration, not as a default upgrade everyone should be on.

Taglertrechecks every connected Meta and Google Ads account every thirty minutes and emails you when events stop arriving, match quality drops or duplicates start. It reads what the platforms report about the account, so it is indifferent to whether events came from a browser, a Conversions API Gateway or a Signals Gateway pipeline — which is exactly the property you want from the thing watching a server-side setup, because it cannot be fooled by a pipeline that looks healthy from the inside.

Sources

Stop finding out about broken tracking from your client.

Taglert monitors your pixels and conversion tracking 24/7 and alerts you the moment something breaks.

Start your free trial
  • First scan in ~5 minutes
  • 30-day money-back guarantee
  • No credit card

Related reading