Connect Amplitude to Microsoft (Bing) Ads

What it is

Cohort sync sends a behavioral audience's membership to Microsoft Advertising for use in Customer Match-style targeting or remarketing lists, following the same activation pattern as syncing a cohort to any other ad platform.

Event streaming sends individual conversion events (a purchase, a signup, a specific milestone) to Microsoft Ads in near-real time so its bidding algorithms can optimize toward the outcome that actually matters, rather than a proxy like click-through.

The Microsoft Ads source pulls campaign, spend, and click data into Amplitude, which is what makes blended-channel analysis (comparing Microsoft Ads spend against downstream Amplitude-tracked outcomes) possible without exporting from Microsoft's own reporting UI by hand.

When you need it

• Paid search or social spend on Microsoft Advertising needs to optimize toward a real product outcome (activation, a specific purchase, a subscription) rather than a Microsoft-side proxy metric - that calls for the event-streaming direction.
• You want to exclude existing customers from acquisition campaigns, or target a specific high-intent behavioral segment for remarketing - that's the cohort-sync direction.
• Marketing or finance needs Microsoft Ads spend sitting next to Amplitude-tracked conversion and retention data in one place for attribution or ROAS analysis, without manually reconciling two exports every reporting cycle - that's the source direction.

Implementation, done properly

1. Scope the project first: pick which of the three integrations (cohort sync, event streaming, spend source) actually solves the problem in front of you, since each has its own auth flow and its own ongoing maintenance surface, and building all three when only one is used is wasted setup.
2. For event streaming: define the exact conversion event and its value (if revenue-bearing) in Amplitude first, confirm it fires reliably and only once per real conversion, then map it to the corresponding Microsoft Ads conversion action - a duplicate-firing event streamed into an ad platform's bidding algorithm will actively mislead its optimization, not just misreport a dashboard number.
3. For cohort sync: build and validate the cohort in Amplitude, add Microsoft Advertising as a destination with the required authenticated account access, and confirm the matching identifier (typically a hashed email or Microsoft-specific click ID, depending on how the integration is documented at setup time) is populated for a meaningful share of the cohort before trusting the sync's coverage.
4. For the spend source: connect the Microsoft Ads account as an Amplitude source, and be deliberate about which campaigns and date range backfill you pull in - a full-account, full-history pull on first connect can be far larger and noisier than what any current analysis actually needs.
5. Whichever direction you set up, confirm attribution windows and click-ID handling align between Microsoft's own reporting and what lands in Amplitude - a mismatch here is the most common reason "our numbers don't match between platforms" complaints trace back to an ads integration rather than a tracking bug.

How to verify it worked

1. For event streaming: use Microsoft Advertising's own conversion-tracking diagnostics (or a test conversion) to confirm events are arriving, then cross-check the count and timing against the same event in Amplitude's Events explorer for the same time window.
2. For cohort sync: check Amplitude's sync log for a successful run, then confirm in Microsoft Advertising's audience manager that the audience size is in a plausible range and growing/shrinking as expected as the underlying cohort changes.
3. For the spend source: reconcile a single day's spend and click totals between Microsoft's own reporting UI and what shows up in Amplitude, for a campaign small enough to check by hand - a discrepancy here almost always points to a timezone or attribution-window mismatch, not a data-loss bug.
4. Whichever piece you're validating, let it run for at least one full week of real traffic before concluding it's working correctly - a spend or conversion source that looks fine on day one can still reveal gaps once weekly reporting cycles and any delayed-conversion windows complete.

Related: Sync Amplitude cohorts to AppsFlyer · Use case: one funnel from ad click to activated user · Fixing zeroed-out revenue properties