Monitoring and support
A connector that tells you what it did
Monitoring is part of the connector, not a tier of it. A support plan adds a person: someone watching the alerts, and someone who fixes the connector when the vendor changes the API. It is optional, and a connector does not stop working without one.
Included, always
What every connector reports
Emitted to your own monitoring stack in a standard shape, so it lands in whatever you already use rather than in a dashboard you have to visit.
| Did it run | Start, finish, duration and outcome of every run |
|---|---|
| How much moved | Rows extracted, loaded and rejected, per dataset |
| How fresh it is | How stale the newest row in each data model is |
| What the API did | Request counts by status, and how often it rate-limited us |
| Where it stopped | The checkpoint, so a failed run resumes instead of restarting |
| Schema drift | New fields and changed types, detected rather than discovered later |
| What is running | The version actually deployed, not the version intended |
Every connector emits the same metric names, so a second connector needs no new dashboard. Payload data never appears in a log line — the logs tell you a run moved 4,812 rows, not what was in them.
Optional
Two halves, and only one of them comes out of your hours
A contract is guaranteed hours a month plus what we do anyway. The minimum is two hours a month over twelve months. The figures are on the pricing page.
Your hours
You decide what they are spent on.
Updates and minor changes, absorbing a schema change from the vendor, investigating and fixing breakage, tuning the cadence or the backfill window, extra monitors, re-running a backfill. Guaranteed monthly. Unused hours carry one month and then expire — we would rather tell you that than let you bank a year of them and find we cannot deliver them in a week.
What we do anyway
Never deducted from your hours.
We receive and triage the alerts, report on run health, apply security patches, keep the dependencies and the SDK current, and answer inside two business days. Keeping your connector patched is our job. So is fixing a defect in what we built — that never comes out of your hours either.
A contract signed with the order also discounts data models — data models only, never the DevOps line or the connector — and that discount is clawed back pro-rata if support is cancelled inside the term. The pricing page carries the three lines a contract is made of, the plans, the minimum and the discount, because that is the only page on this site that carries a number.
Before you ask
What a support plan is not
A plan from a small firm is worth exactly what it can deliver, so here is the shape of it.
| Is it required? | No. The connector is yours and keeps running without one. If we ever imply otherwise, hold us to this sentence |
|---|---|
| Is it 24/7? | No. European business hours. We will not sell you an out-of-hours promise two people cannot keep |
| Is there a contractual response time? | Yes. Two business days on every plan, in European business hours — and that means we have answered with a cause or a plan, not that it is fixed. Faster response and wider coverage are adjustments we price on request |
| Who does the work? | The people who built your connector. That is the whole value, and it is also the key-person risk — worth weighing |
| What if the API vendor breaks it? | We fix it, and it comes out of your hours. If the fix is bigger than the hours you have — a vendor changing authentication can be three days against a four-hour month — we quote the difference or agree to draw on next month. We would rather show you that arithmetic now than send you the invoice later |
| Can we cancel and self-support later? | Yes. The code is in your repository and your team can read it |
Ask what it would cost to look after yours
Tell us the source system and how much it changes. We will give you a number and say plainly whether a plan is worth it for you.
Get an estimate