Know how your Stellar app is used, without profiling the people using it.

Shared analytics for Stellar projects: API metrics over OpenTelemetry, web analytics, and a wallet funnel. Every number carries its definition, and the operator view only ever sees aggregates.

Signing in uses a one-time wallet signature. No transaction is sent.

7963980
03:30 PM–05:30 PM · worst bucket 12.9% 5xx
Requests5xxIncident: a bucket with at least 3 errors and 5xx at 3% or moreTimes in your local time zone
Synthetic sample data. The shaded window is where 5xx rose; every number in the app carries its definition.

Sample project · Overview

Synthetic sample data. Not a real project.

Requests
36,10913% versus the previous period

31,840 previously · Compared with the equal period before

p50 latency
48ms5.9% versus the previous period

51ms previously · Median

p95 latency
318ms21% versus the previous period

262ms previously · Approximate

p99 latency
702ms1.7% versus the previous period

690ms previously · Approximate

Server errors
1.4%1.0 pt versus the previous period

0.4% previously · 5xx ÷ all

Status classes

  • 2xx Success

    34,210

    94.7% of responses

  • 3xx Redirect

    612

    1.7% of responses

  • 4xx Client error

    780

    2.2% of responses

  • 5xx Server error

    507

    1.4% of responses

MethodRouteRequestsp955xxTrend
GET/v1/pools18,4201420.2%
POST/v1/quote9,3103183.1%
GET/v1/pools/:id6,204960.1%
POST/v1/swap/submit2,1756401.2%

Three signals, each with its limits stated

API metrics
Requests, latency percentiles and error rates per route, from the OpenTelemetry metrics your services already emit. Metrics only; traces and logs are rejected.
Percentiles are approximations from merged histograms.
Web analytics
Page views, referrers and devices from a pinned, self-hosted Umami, issued only after you verify the domain by DNS or a file.
Visitor counts follow Umami's definitions.
Wallet funnel
Where people drop between connecting a wallet and submitting a transaction, counted from event names alone.
Reported by your frontend. It cannot prove a ledger outcome.

From sign-in to first data point

  1. 1Sign in with your wallet. A one-time signature on testnet. No email, no password.
  2. 2Create a project and an ingest key. The key, never the payload, decides which project a metric belongs to.
  3. 3Dual-export from your service. Keep your existing APM and add Pulse as a second OTLP/HTTP JSON destination.
  4. 4Watch the overview fill in. It refreshes every 15 seconds, with routes, latency and error rate side by side.
Environment
export OTEL_EXPORTER_OTLP_METRICS_PROTOCOL=http/json
export OTEL_EXPORTER_OTLP_METRICS_ENDPOINT=https://<your-pulse-host>/v1/metrics
export OTEL_EXPORTER_OTLP_METRICS_HEADERS="Authorization=Bearer <ingest key>"
export OTEL_EXPORTER_OTLP_METRICS_TEMPORALITY_PREFERENCE=delta

Privacy by omission: no IPs, request bodies or wallet addresses in analytics

What Pulse stores

  • Counters and latency histograms per route, per minute.
  • Wallet event names and counts.
  • A one-way hash of each ingest key secret.
  • Project membership, by Stellar public key, for access control.

What Pulse does not store

  • Traces, spans, logs or request bodies.
  • IP addresses in Pulse's own tables. Web analytics run on Umami, which has its own data handling.
  • Wallet addresses, signatures or transaction XDR in analytics.
  • Anything that would let one project read another's data. Isolation is enforced in the database, not only in the app.

Definitions you can check

Server error rate
5xx responses divided by all responses, over the selected range.
Latency percentiles
Interpolated from merged fixed-boundary histograms. They are approximations, and the top bucket reports its lower bound (10 s).
Wallet events
Reported by your frontend through the tracker. They are not confirmed on the Stellar ledger.

Instrument your first service

Connect wallet