Automation & spend reconciliation
Your network bills you for clicks. Your tracker records the ones that arrived. The two numbers are never identical, and the gap is money.
Automation closes that gap with measurement rather than estimates: Roitrack asks the network, zone by zone and hour by hour, what it actually charged you for, compares it with what was recorded, and labels every zone with the difference.
It is off by default, and switching on measurement changes nothing about your spend figures until you separately ask it to.
What you need
A traffic source built from one of the networks Roitrack can talk to, and an API key from that network.
| Network | Zone-level stats | Notes |
|---|---|---|
| PropellerAds | Yes | Bearer token from the panel |
| ExoClick | Yes | API token |
| Galaksion | Not yet | Macros only for now |
| TwinRed | Not yet | Their API has no zone breakdown |
The macros are filled in for you when you pick one of these as the source’s template. Do not change the sub mapping: zone has to arrive in sub2 and the network’s campaign ID in sub1, because that is the pairing the comparison is built on. A campaign whose link puts the zone somewhere else will show every zone as unlinked.
Switching it on
- Open Automation from the sidebar.
- Find your traffic source in Spend & traffic loss adjustment. Only sources that can be driven by API are listed.
- Paste the API key into the source and save it.
- Tick Measure.
That is enough to start collecting. The first round runs on the next tick; Check now runs it immediately, using exactly the code the schedule uses.
Each source has its own interval, 12 hours by default. The default for new sources lives in Settings → Performance.
Reading the result
Once a round completes, two columns are available in any report, at any grouping:
- Zone Loss% — the share of clicks the network billed that never reached your tracker.
- Network Cost — what the network says that zone cost, as opposed to what the macro reported.
Both are off until you enable them in Settings → Columns, like every other column.
The columns are blank, not zero, where nothing has been measured. A zone with no measurement and a zone that delivers everything it bills for are opposite facts, and showing them identically is how a good zone gets blacklisted.
A zone reading not linked is almost never a zone that delivered nothing. It means the network billed clicks that carry no matching zone in your tracker — usually the zone macro is missing from that campaign’s link. Fix the link before acting on it.
Correcting future spend
Adjust spend is the second switch, and it is deliberately separate.
With it on, the measured loss becomes a multiplier applied to the cost of new clicks, per zone. A zone measured at 40% loss has the clicks it does deliver priced at what the network is really charging for them, so the ROI you see is the ROI you are actually getting.
What it never does:
- Rewrite history. Clicks already recorded keep the cost they were recorded with. Each campaign remembers how far it has been reconciled, so no hour is ever counted twice.
- Act on a weak sample. A zone with fewer than 50 clicks on the network’s side is left alone.
- Act on a capped or unlinked zone. Those are flagged, not applied.
Leave this off until the measurements have held steady across several rounds. Turning it on makes your cost figures move, and you want to be sure they are moving to the right place.
The campaign’s own Traffic loss %
There is a second, older way to account for the same gap: the Traffic loss % field on the campaign form. It is an estimate you type, it applies to the whole campaign, and it inflates recorded cost by 100 / (100 - loss) — 30% makes a $0.01 click record as $0.0143.
It works on its own. It needs no API key, no network integration and nothing switched on here, which is how most campaigns use it.
When both exist, the measurement wins and the typed figure is the fallback:
- a zone the reconciler measured, where there is one
- the campaign-wide measurement, where there is one
- the Traffic loss % you typed
- unadjusted
They are never multiplied together. Both describe the same lost traffic, so applying both would charge for it twice — 45% typed against 25% measured would bill a click at 2.27× when the measurement says 1.25×.
So you can leave an estimate in place while measurement is switched on: each zone moves off your estimate and onto a real number as it gets measured, and zones the reconciler cannot measure keep the estimate rather than falling back to nothing.
When it goes wrong
Every call Roitrack makes to a network is recorded — the operation, the status, how long it took, how many rows came back, and the error if there was one. Open API call log at the bottom of the Automation page and filter by Failed.
An expired API key, a network that changed its response, a campaign ID the network does not recognise: all of them show here with the reason. An integration that quietly stopped working is otherwise the kind of thing nobody notices for a month.
The source itself also shows the result of its last check, with the time it ran.
What leaves your server
Only the network’s own campaign IDs, sent to that network’s API, to ask for statistics it already holds. No click IDs, no addresses, no visitor data of any kind. The API key is stored write-only — it is never returned to the browser once saved.