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

Glossary term

Error Log

One list across every channel of the errors that reached customers: when found and contained, what customers saw, the checks that missed it, causes and actions.
StrategyconceptUpdated

Definition

An error log is one list, across every channel, of the marketing errors that reached customers, stored alongside the team's other operating records. The tutorial on reviewing a marketing error describes its job as recording enough about each error to spot a pattern without retelling the story, which is why entries are short and point to the evidence rather than repeating it.

The single list is deliberate. A log per channel holds too few entries to show anything, and it hides the case that matters most: one gap letting errors through in email, paid social and the website alike. The guide to handling an AI mistake closes every AI incident with an entry, then points to one error log for all channels, so that AI incidents sit beside the rest and a missing check shows up whether the draft came from a model or a person.

What each entry records, and why causes name processes

The fields are few, and each one earns its place by what it lets a later reader see:

  • When. The date the error was found and the date it was contained. The gap between the two dates measures how long errors stayed in front of customers, and whether that time is shrinking.
  • Where and what. The channel, the kind of work and what customers saw, with a reference to the evidence.
  • The checks that missed it. Each gate that existed, and why it let the error through.
  • Causes. The process causes the review found.
  • Actions. Each agreed change, with one owner and a date, and whether it has been confirmed done.
  • Repeat of. A pointer to any earlier entry with the same cause, which turns separate incidents into a pattern.

Causes name processes, not people. A name in the cause field gives the next reader nothing to change; a missing field, check or handover does. The five whys is how a review gets there: it keeps asking why, on evidence, until the answer is a process or rule that can be altered, and that answer is what the log keeps. Personal data stays out of the entries themselves, with records referred to by reference.

In practice: the monthly read and the risk register

Entries arrive from the procedures that handle each error. The live ad and page error response SOP adds one once the fix has been checked in every place the error appeared, recording when it was found and contained, where it ran, the remedy decided and whether a review follows. It also counts skipping small errors as a mistake, because a repeated error only becomes visible when the small ones were logged too.

The log is then read, not just filled. The tutorial on building a marketing risk register reviews the register monthly, and the incident management playbook reads the error log in the same monthly slot as the register, so every logged error sends the team back to the risk it fell under: did the warning sign appear, and did the response work? The same read confirms that each action was done on its date, because an agreed action left undone is what most often lets an error come back. For AI uses, the AI governance operations playbook reads its incident list monthly with the review logs, and runs mistakes that span channels inside the one error log the incident management programme keeps.

Illustrative example — not a real result

A ticketing firm finds one gate behind three wrong dates

Jompeard Tickets, an invented events ticketing firm, read its error log at the end of a quarter and found 3 entries from different months, each an email that gave the wrong date for an event: a comedy night moved from Friday to Saturday, a festival listed in the wrong month and a matinee sent with the evening date. Three different campaign owners had logged them, and each had been fixed with a correction to between 400 and 2,600 recipients. Read together, the field for checks that missed them named the same step every time: proofreading, which compared each email with its brief but never with the booking system, where dates were changed after briefs were signed. The team added a date check against the booking system to the proof step, owned by the email lead, and the next quarter's log held no date errors.

A Risk Register lists what could go wrong, each risk with a warning sign and a prepared response; the error log records what did go wrong, and the monthly read joins the two. A Change Log logs, with dates, each change to the team's tracking, accounts, reports, lists or rules, so an error that touched the numbers usually earns a line in both. A Decision Log records decisions and the reasons for them, while the error log records failures and the actions they prompted. Each entry usually describes one Marketing Incident, a live error that reached customers, spend or reports.

Common confusions

  • It is the same as a system's error logs. Software writes those automatically. A marketing error log is written by people, one entry per error that reached customers, with the review that followed.
  • Each channel can keep its own. Patterns across channels, such as one gate missing errors in several of them, only show in a single list.
  • A falling count means fewer errors. More entries can mean a team is checking more carefully, and fewer can mean errors are going unrecorded. Confirmed actions and repeats with a known cause say more.
  • Logging closes the matter. An entry whose actions are never checked changes nothing; the log is read and its actions chased.

More terms to look up

Where this term is used

Related