A preview of the Pixel Data dashboard showing sample data: spend today £7,842.18, spend this month £198,241.67, margin £68,732.11, and 1.28M events today.
One command on your infrastructure, or seven lines in your service.
The agent finds the usage data where it already lives — switch CDRs, gateway logs, application events — and starts sending it. Nothing to schema, nothing to model, no data engineering project. If you would rather send events yourself, one plain HTTPS endpoint accepts them from any language.
How integration worksThe install screen sits on “waiting for first event…” until it arrives.
Usually inside fifteen minutes. From that moment the dashboard is live: traffic by the minute, spend today, spend this month, top accounts, top event types. This is your data, not a demo.
See the analyticsUpload a tariff as CSV: match key, rate, increment, minimum.
Longest-prefix matching handles 44 against 447 against 4479 the way a telecom person expects. Plans are versioned with effective dates and are never edited in place, so last month's invoice stays reproducible forever. Before you deploy a change, replay last month's real traffic against the draft and see what it would have done.
Rating in detailEvery amount on every screen is clickable.
Click a figure, get the events behind it. Click an event, get the calculation: the recorded quantity, the increment it was rounded to, the minimum that may have replaced it, the rate that matched and why, and the plan version that was live at that instant. This is the difference between a number and a number you can defend.
See a drill-downUpload a supplier invoice. Get a line-by-line verdict.
Matched, rate difference, timing difference, missing charge — four categories, never one vague “mismatch” bucket. Timing differences are calls that straddled the period boundary and will clear themselves; rate differences are money you are owed. Nobody else does this, because everyone else meters what you sell rather than what you buy.
How reconciliation worksMonthly, per account, as a document your accountant will accept.
Every line traces back to its events, and every rated row is immutable, so an audit trail exists whether or not anyone ever asks for it. Export to CSV, or push the rated amounts into your own payment provider.
Statements and exportsAlerts denominated in pounds, not in metrics.
“Route 234 went margin-negative at 14:02.” “Premium SMS volume is 320% above normal.” “Your supplier changed a rate mid-month.” A threshold breach on a graph is a puzzle; a number with a pound sign in front of it is a decision.
Alerts and forecastingWe launched through telecom because it is the hardest case. Everything below runs on the same universal event schema.
CDRs from FreeSWITCH or Asterisk, prefix-based tariffs with increments and minimums, carrier invoice reconciliation. The agent reads your switch; you do not write code.
Rating handles 1s and 60s billing, 30/6 patterns, and minimum charges. Prefix collisions resolve longest-first.
Token metering per model, per customer, per region. Rate input and output separately, in fractions of a penny, without floating-point drift.
Matched on the model name rather than a dialling prefix — same engine, different match property.
Any countable thing: seats, requests, gigabytes, shipments. If you can send it as an event, we can rate it, chart it and put it on a statement.
One account, one API key, every source into the same dashboards.
If you only have three minutes, this is the story: install at minute one, live data at minute two, money found at minute three.
Minute 1
One command
The agent installs and starts reading your switch.
Minute 2
Live data
The dashboard fills with your own traffic and spend.
Minute 3
Money found
A supplier invoice is audited against what really happened.
Prefer to watch someone else click? The sandbox does the same thing, with sample data.
Install takes one command. If you would rather look first, the sandbox is open — sample data, no card, no signup.