Weekly Report Data Analysis

Weekly Incident Report:
September 30 to October 7, 2026

AdStatus recorded nine incidents in seven days. Claude logged four of them. The ad platforms logged five, and four of those were reporting and data delays: your ads kept running, your numbers didn't keep up.

October 7, 2026 | 6 min read | Weekly Report

The Numbers

9
Incidents in 7 days
4
On Claude
4 of 5
Ad platform incidents hit data, not delivery
26 min
Meta's only incident, detection to fix

Source: AdStatus monitoring of official status pages, incidents first detected September 30 to October 7, 2026 (UTC).

Every Incident This Week

First detected (UTC) Platform What broke
Sep 30, 01:01GoogleGoogle Analytics issue
Oct 1, 02:29Microsoft AdvertisingOffline Conversions data delayed up to 11 hours
Oct 1, 16:21ClaudeDelayed credits on the Claude Platform
Oct 1, 18:45Meta (WhatsApp)Meta Business Agent Platform outage
Oct 1, 18:50Microsoft AdvertisingMonetize reporting delayed
Oct 3, 20:45GoogleGoogle Ad Manager reporting delays
Oct 5, 13:09ClaudeElevated errors for Claude Mythos 5.1 and Claude Fable 5.1
Oct 6, 12:28ClaudeElevated errors for Claude Opus 5.5
Oct 7, 01:24ClaudeElevated errors loading usage data on platform.claude.com

Times are when AdStatus first saw each incident on the provider's status page. Where the provider stated an earlier start time, it is given below.

The Ad Platforms Lost Data, Not Delivery

Google, Microsoft Advertising and Meta logged five incidents between them. Four were about data arriving late or not at all. That is the same pattern we flagged in the September reliability report, where 7 of 9 incidents hit measurement or payments rather than delivery.

Microsoft Advertising: two reporting incidents in one day

Microsoft said Offline Conversions data was delayed by up to 11 hours, starting September 30 at 21:01 UTC. Its notice was explicit: reports, the UI and the API "may see stale or inaccurate data", and there was "no impact on the Ad delivery". The same day it posted a second incident: Monetize reporting delays or missing data from September 30 onwards, starting 18:50 UTC. Its October 1 update said the engineering team had identified the issue and data was recovering.

Google: Analytics, then Ad Manager reporting

Google's Analytics incident began on September 29 at 00:41 UTC by Google's own timeline. When we picked it up, Google said service had been restored for some users and expected a resolution for all users soon. On October 3, from 18:00 UTC, Google Ad Manager's Reporting tool ran behind: users could access Ad Manager but might not see the most recent data. Google noted that Data Transfer was unaffected.

Meta: one short outage, and not in Ads Manager

Meta's only incident was the Meta Business Agent Platform, listed under WhatsApp. AdStatus first saw it at 18:45 UTC on October 1, and Meta marked it resolved at 19:11 UTC: 26 minutes from our detection to the fix. Meta's message: "We have recovered from an earlier outage impacting Meta Business Agent Platform, and services have now been restored." Ads Manager logged nothing this week.

Claude Logged More Incidents Than Any Ad Platform

If your team writes copy, briefs or analysis with Claude, this was the week to have a fallback. Claude logged four incidents in the window, and a fifth the day before it opened: elevated errors on claude.ai, Claude Code and Claude Cowork on September 29.

  • Delayed credits, October 1. Credit purchases took time to show up in account balances, and some requests failed for insufficient credit balance. By Anthropic's own timeline it was investigated from 16:20 UTC, fixed and monitored from 18:23 UTC, and resolved at 22:36 UTC.
  • Claude Mythos 5.1 and Claude Fable 5.1, October 5. Elevated errors on requests to both models, investigated from 13:05 UTC.
  • Claude Opus 5.5, October 6. Elevated errors on requests, investigated from 12:24 UTC.
  • Usage data, October 6 (US time). Errors loading the Usage page on platform.claude.com and the Admin API usage report endpoint, from 5 PM PT. Anthropic said Messages API requests were not affected.

Two of the four were model-specific. The other two touched billing and reporting around the API rather than the API itself.

What to Do With This

  • Re-pull Microsoft Advertising conversion reports for September 30 to October 1 before you judge those days. Microsoft said the data could be stale or inaccurate, and an automated rule acting on it in that window acted on wrong numbers.
  • If you publish with Google Ad Manager, treat Reporting numbers for October 3 after 18:00 UTC as provisional. Google said Data Transfer was unaffected, so those files are the better source for that window.
  • If you run WhatsApp agents on Meta Business Agent Platform, check conversations around 18:45 to 19:11 UTC on October 1.
  • If Claude is part of your workflow, set a fallback model for when one model shows elevated errors, and keep a credit buffer before large runs: on October 1, requests failed while purchased credits were delayed.
  • A data delay looks like a performance drop. Before you cut a budget on a bad day, check whether the platform announced a reporting incident for it.

Methodology

All incident data comes from AdStatus monitoring of official status sources: metastatus.com for Meta, ads.google.com/status for Google, status.ads.microsoft.com for Microsoft Advertising and status.claude.com for Claude. AdStatus recorded no incidents for Amazon Ads or Shopify in the window. ChatGPT is not counted: OpenAI posted ChatGPT incidents on October 5 and 6, but our ChatGPT incident log was incomplete this week. Feeds are polled every 5 minutes, and each incident is stored once under the provider's own incident ID, whatever the number of updates.

Counting rules: the nine incidents are distinct incidents first detected between 2026-09-30 and 2026-10-07 UTC. Quoted start times and resolutions are the providers' own; we state a duration only where the provider published one or marked the incident resolved. The Claude incident of September 29 falls outside the window and is not counted.

All times are UTC unless marked otherwise.

Get the Next Incident in Your Slack

AdStatus monitors Meta, Google, Microsoft, Pinterest, Amazon, Shopify, ChatGPT and Claude every 5 minutes and posts to Slack the moment a provider publishes an incident. Subscribers can also ask Claude, ChatGPT or Gemini "is Google Ads down?" through AdStatus for AI assistants. 7-day free trial.

Add to Slack - Free for 7 Days