Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Glossary term

Tracking Outage

A period when conversion or analytics tracking stops recording, or records wrongly, while sales carry on; its start and end are marked as the outage window.
AnalyticsconceptUpdated

Definition

A tracking outage is a period in which conversion or analytics tracking stops recording, or records wrongly, while real sales or leads carry on as normal. A purchase tag that no longer fires after a checkout update is the familiar case. A tag that starts firing twice is the less obvious one: demand is unchanged but recorded conversions are not, and the guide to bidding through sales and tracking outages treats that window the same way as one where the data went missing.

What separates an outage from a bad week is a source the tags do not touch. If the shop platform or the CRM shows normal activity while the measurement falls or jumps, the measurement is what changed; if every source moves together, the change is probably real.

The outage window runs from the first record that looks wrong to the point where real conversions are recording again in their usual relationship to orders or leads. It is written down with a start, an end and the evidence for both, because every later decision reads from it.

Why it matters more under automated bidding

Under automated bidding a broken tag steers the spend as well as the reports: the strategy sets each bid from the latest recorded results, so a silent purchase tag reads as a collapse in conversion rate. Bids fall on auctions that were in fact converting, and the strategy carries the false lesson for weeks after the tag is mended. A tag that double-counts pushes bids the other way.

Reports built on the same tags inherit the error too: a month containing an outage window reads as a weak month, and next year's comparison repeats the mistake unless the window has been marked.

In practice: confirming, containing, marking the window and comparing periods

The response belongs to the tracking outage response SOP, from confirming the gap to closing the incident. Seen from the term's side, four habits matter most.

  • Confirm against the business's records. An outage is declared only after the measurement has been compared with orders or leads for the same period; made daily, the same comparison spots one in hours rather than days.
  • Contain the effect on bidding. Budgets and targets are held steady, structural edits wait, and where the platform documents data exclusions for the strategy in use, one covers the outage window. Which strategies support it and how the window is entered are in the platform's help centre.
  • Mark the window. Once recovery is verified, the window is annotated in analytics and dashboards and recorded beside every other change. The five whys method for root-cause analysis then follows the gap back past the broken tag to the release step that should have caught it.
  • Keep it out of comparisons. The Google Ads account audit checklist steers reviewers away from comparison periods that include a site move or an outage window, because missing or doubled conversions make a poor yardstick.

Some outages can be expected. A site move changes thank-you page addresses, templates and redirects at once, so the guide to Google Ads during a website migration checks daily conversions against orders or leads in the business's own system for several weeks after launch and treats a widening gap as broken tracking rather than a fall in results.

Illustrative example — not a real result

A paint supplier's cookie banner hides its purchases

Gurnooth Paints, a trade and retail paint supplier, released a new cookie banner at 15:40 on a Tuesday. By Thursday morning Google Ads showed 6 purchases since Tuesday afternoon against 118 orders in the shop system over the same hours, and analytics showed the same fall. The analytics owner declared an outage from 15:40 on Tuesday; triage traced the gap to the banner release. The paid media manager left the target ROAS of 400% and the £250 daily budget untouched and set a data exclusion for the window. The banner was fixed on Friday morning, and once real purchases matched orders again at 13:10 the window was closed, annotated and entered in the change log.

A Conversion Action defines what is counted; an outage is the period when it stops counting what the business actually sells, or counts it twice. Smart Bidding is why an outage costs money as well as data, since its bids rest on the recorded conversions. The Change Log is where the window lives once the incident is closed, beside the other changes that explain steps in the figures. Conversion Lag is easily mistaken for an outage: the latest days look short because some conversions are still to arrive, but they fill in, whereas an outage leaves a gap that does not. A Baseline is the estimate of what a figure would have done without a change, and one drawn from a period that contains an outage window starts from the wrong level.

Common confusions

  • Every drop to zero is an outage. Only a comparison with records outside the tags shows that; a real fall appears in every source at once.
  • An outage means missing conversions. A tag that fires twice is an outage too, inflating results instead of hiding them.
  • A site outage and a tracking outage are the same. When the site is down, nobody can buy, and the technical team's incident process applies. In a tracking outage the sales carry on and only the record of them is wrong.
  • Fixing the tag brings the data back. Events the tags missed are gone from analytics for good; some platforms accept recorded orders as offline conversions afterwards, and the rest of the gap is annotated.
  • Pausing every campaign is the safe response. Pausing loses delivery and learning; the SOP reserves it for campaigns visibly spending too much on the false signal.

More terms to look up

Where this term is used

Related