What connects today, and what does not
Integration pages are usually a wall of logos with every status set to available. This one is labelled by what you can actually switch on, and two of the entries below say plainly that you cannot switch them on yet.
- Available
- You can turn it on today and it does what it says.
- Limited
- Real and running, but narrower than the name suggests.
- Beta
- Implemented and reachable. Talk to us before you depend on it.
- In development
- We have built the ingestion, but there is no way for you to connect it yet — so it is not available.
Getting findings out of TrackGeo
Nobody opens a dashboard daily. These are the paths a finding takes to reach a person or a system that will act on it.
Slack
Send visibility alerts into a Slack channel — a brand mention dropping, a competitor surging, a run failing, or a spike in Reddit engagement.
What you should know
This is an incoming webhook: you paste a webhook URL into settings. It is not a Slack app, so there is no install flow and no bot user. Email digests do not currently route to Slack — they arrive by email, in-app alert and outbound webhook.
Webhooks
Push events to any endpoint you control, filtered by event type. Prompt runs completing, onboarding scans finishing and digests all fire real events.
What you should know
Delivery is fire-and-forget: there are no automatic retries, and payloads are not signed, so your receiver cannot cryptographically verify the sender yet. Endpoints are validated at connect time so a webhook cannot be pointed at private infrastructure. Configured through the API rather than a settings screen.
Reddit mention tracking
Track mentions of your brand across subreddits you choose, with sentiment labelling, high-engagement alerting, and correlation against the sources AI engines cite.
What you should know
This reads Reddit's public search endpoint rather than the authenticated Reddit API, so coverage depends on what Reddit serves publicly and can be rate-limited. Syncs run on demand rather than on a fixed schedule. Sentiment here is keyword matching, not the model-based classifier used on engine responses.
Build on top of your own data
Authenticated with API keys, and honest about how much surface is open today — there is no point discovering the boundary after you have started building.
REST API
Read your workspace overview programmatically with an API key, for a dashboard or a warehouse of your own.
What you should know
Know the scope before you plan around it: one endpoint is currently exposed to API-key authentication — the workspace overview. The rest of the application's API requires a signed-in session, and there is no general-purpose public REST surface or published API reference yet.
MCP server
Connect an AI assistant or agent to your TrackGeo workspace over the Model Context Protocol, using JSON-RPC over server-sent events, authenticated with an API key.
What you should know
Implemented and reachable, but treat it as beta: the protocol revision is pinned to an older release and server-sent events are the only transport. Talk to us before you build a production workflow on it.
API keys
Generate keys to authenticate the endpoints above. Keys carry a tg_ prefix, are stored only as a SHA-256 hash, and can be revoked at any time.
What you should know
Keys are created and revoked through the API — there is no key-management screen yet. The plaintext key is shown once, at creation, and cannot be recovered afterwards.
If you are planning an integration and the surface you need is not open yet, tell us what you are building — that is what decides what opens next.
Built, but not yet connectable
These two are the reason this page has four statuses instead of one. The data clients and the ingestion are written. The part where you connect your account is not, and that is the part that matters to you.
Google Analytics 4
Set assistant referral sessions and landing-page behaviour beside your visibility data, so you can see whether visibility movement showed up as demand.
What you should know
The client and the ingestion are built, but there is no Google sign-in flow, no way to save a property and no screen to manage it — so there is nothing you can switch on today. We are not going to call that available.
Google Search Console
Bring search clicks, branded query performance and top queries alongside AI answers — and feed real query demand into how prompts are scored for buyer value.
What you should know
The same position as GA4: the client and ingestion exist, the connection flow does not. Until it does, the search-demand signal in prompt scoring falls back to neutral rather than inventing a value.
If you are evaluating TrackGeo specifically to join AI visibility to GA4 or Search Console data, raise it with us first. We would rather have that conversation now than have you find out in week three.
Following an assistant referral all the way through
TrackGeo measures what assistants say. Turning that into observed demand needs the other half of the journey.
Assistant referrals that reach your site are recorded as observed when the referrer identifies the engine. Everything short of that is labelled estimated, inferred or exposure — and never totalled into the observed figure to make a report read better.
How we separate observed from inferredIntegrations, answered plainly
Can TrackGeo connect to GA4 or Google Search Console today?
Not yet. The data clients and the ingestion are built, but there is no Google sign-in flow and no screen to connect a property or a site — so there is nothing you can switch on. We would rather tell you that now than let you buy on the strength of a checkbox. If this is on your critical path, contact us and we will tell you where it stands.
Is the Slack integration a Slack app?
No. It works by pasting an incoming webhook URL into your settings, which means there is no install flow and no bot user. It does send real alerts — brand mentions dropping, competitors surging, runs failing, and Reddit engagement spikes. Email digests currently arrive by email, in-app alert and outbound webhook rather than in Slack.
Are webhook payloads signed?
Not yet, and you should factor that in. Payloads are delivered as-is, so your receiver cannot cryptographically verify that a request came from us, and there are no automatic retries if your endpoint is down. Endpoints are validated at connect time so a webhook cannot be pointed at private infrastructure.
How much of TrackGeo can I reach over the REST API?
One endpoint today: the workspace overview, authenticated with an API key. The rest of the application's API requires a signed-in session, and there is no published public API reference yet. If you need programmatic access to something specific, tell us what you are building — that is what decides what we open up next.
Do you support the Model Context Protocol?
Yes, in beta. TrackGeo runs an MCP server that speaks JSON-RPC over server-sent events and authenticates with an API key, so an assistant or agent can query your workspace directly. The protocol revision is pinned to an older release and server-sent events are the only transport, so talk to us before you build a production workflow on it.
What if the integration I need is not listed?
Then it does not exist yet, and we have not listed it. Everything on this page was checked against the implementation before it was given a status — nothing appears here because it sounded good on a comparison table. Tell us what you need and what it would unblock.
Need something on this page to move faster
Tell us what you are building and what it would unblock. That is what decides the order these ship in.