Skip to content
All field notes
Measurement
Published
Nov 4, 2026
Reading time
11 min read

Microsoft Offline Conversions: MSCLKID or Enhanced Matching?

Audit Microsoft offline conversion imports: separate customer lists from outcomes, verify MSCLKID or enhanced matching, then inspect timing and acceptance.

AP
AdProof Research
Independent measurement desk
A paper customer list remains separate from a dated conversion receipt passing through identity and timing checks.
A contact list, an eligible business outcome and an attributed conversion require different evidence. Conceptual illustration without customer data.
In this guide (9 sections)

A customer list is not an offline conversion file. Before importing anything, identify the business outcome, its actual time, the Microsoft conversion goal and the supported matching route. The current Campaign Management API permits omitting MSCLKID when a hashed email address or phone number is supplied for enhanced conversions. That exception does not guarantee a match or make an old list of paid-ad contacts eligible Microsoft conversions.

An older advertiser question asks whether to reuse existing leads after importing a Google campaign into Microsoft Advertising, whether one lead is enough and whether a Bing click ID is necessary. When reviewed, the original displayed an age of two years and two human replies. Their broad setup suggestions do not establish today's identifier requirements or the eligibility of that advertiser's records.

This guide turns those questions into an import-readiness and acceptance audit. It covers the documented Campaign Management and Bulk contracts; an account's current web-upload template or a connector needs its own field mapping. It does not assume a Google import or a CRM connection transfers conversion attribution between advertising platforms.

Separate the records you actually have#

Start with a small inventory before choosing automation. Ask the business owner to identify what each row represents and how that definition was established.

RecordEvidence it providesWhat is still needed
Customer or prospect contactA record exists in the business systemA defined outcome and its actual occurrence time
Google-attributed leadA result under a separately defined Google workflowEvidence relevant to the Microsoft matching route
Captured MSCLKIDA Microsoft click identifier was retainedAssociation with the correct outcome, goal and eligible interval
Hashed contact identifierAn identifier prepared for a supported enhanced routeCorrect goal/configuration, permitted use and matching acceptance
Uploaded conversion rowA record was submitted through a named channelProcessing, attribution and reporting evidence

Keep a contact's creation time separate from a later qualification or sale time. Do not substitute the upload time for the outcome time to make an older record fit. Likewise, do not invent click IDs, manufacture an outcome or treat every historical contact as a completed sale.

Our Microsoft UET activity and revenue guide owns browser activity, online goal matching and transaction-value diagnostics. An offline import has a different entry point: a business-system outcome must be mapped to an offline goal and the selected matching contract.

For your campaign

Choose the matching route for the outcome

Offline imports need a defined business event, conversion goal and supported identifier path. Discuss which route fits your records.

Discuss measurement options

A short inquiry to scope relevant options; no campaign access is needed.

1. Choose the identifier route explicitly#

Microsoft's current OfflineConversion object reference documents MSCLKID as optional when you provide a hashed email address or phone number for enhanced conversions. It identifies SHA-256 email hashing and an E.164-based phone requirement. Inspect the exact preparation rules for your selected integration before implementing a transformation; a field that merely looks like a hash is not evidence of correct normalization.

This is a scoped API exception, not a claim that every uploader accepts the same minimal columns. Preserve the actual channel, template version and identifier fields used. Microsoft's Bulk Offline Conversion record has its own column requirements and explicitly makes Microsoft Click Id required when hashed email or phone is not being sent. Follow the chosen record contract instead of pasting an API object into a web-upload spreadsheet.

A June 2021 Microsoft Q&A question about call conversions received advice that MSCLKID was unavoidable. That historical answer should not override the current reference's enhanced-conversion exception. It also does not prove that any particular call-tracking service now supports that route.

Where MSCLKID is available, retain its genuine association with the outcome. Where the enhanced route is intended, verify that the implementation and account actually use it. Keep unknown or unavailable identifiers in an explicit exception queue rather than silently substituting another platform's click ID.

2. Verify the offline goal before preparing rows#

Read back the goal from the intended Microsoft account. Microsoft's OfflineConversionGoal reference documents the offline type, account/customer scope, conversion window, counting policy and enhanced-conversions-enabled property. A goal with a similar display name is not enough; retain its ID, exact name, scope and current state.

Choose one business definition per investigated goal. For example, “qualified lead accepted by sales” needs its own documented acceptance rule; it is not interchangeable with any contact entered into the CRM. Keep downstream sales evidence separate from a modeled lead value.

For a click-ID route, the same goal reference distinguishes All from Unique counting and documents ignoring duplicate offline conversions with the same MicrosoftClickId and ConversionTime. Do not extrapolate that rule into a complete deduplication recipe for hash-only matching or a different collection route. Preserve stable internal outcome references and the exact original submission data so retries can be reviewed without changing the business event.

The question “Can I start with one lead?” is best answered as a validation question. One genuine, permitted, well-defined outcome can support a bounded technical investigation. It cannot establish account-wide matching coverage, a reliable bidding signal or improved lead quality. Define those later evaluations separately.

3. Check outcome age and attribution eligibility independently#

A successful transport operation and a counted conversion have different requirements. Microsoft's Bulk record timing reference requires a conversion time within the preceding 90 days. For counting against a recorded click, the outcome must follow that click and fall inside the configured goal window. The API requires UTC timestamps; its reference explicitly distinguishes this from the web application's timezone options.

Use separate columns in your reconciliation sheet:

CheckQuestion to answer
Actual occurrenceWhen did the defined business outcome happen?
Upload ageDoes that real timestamp satisfy the selected upload route?
Interaction eligibilityCan the intended Microsoft attribution context qualify?
Goal windowDoes the click-to-outcome interval fit the saved goal?
Time representationWas the original time converted correctly for this API or template?
Reporting intervalWhich dates and time basis does the receiving report use?

Extending a goal's window should not become an attempt to manufacture eligibility for an imported list. Save the original setting and consult the documented effect of any proposed change. Do not assume the enhanced identifier exception removes time or attribution requirements.

Microsoft's official offline-conversion code example also warns that the conversion name must match the goal and that a mismatched name can leave a conversion uncounted without an observed error. Its timestamp comments exclude future outcomes. Use these as acceptance checks, not as a script to run unchanged against production.

4. Retain submission and item-level results#

For the Campaign Management route, ApplyOfflineConversions permits up to 1,000 records per request. It documents waiting two hours after creating the offline goal before sending conversions and allowing up to six hours for conversion data to become visible. These timings do not excuse a missing identifier, wrong goal or ineligible outcome.

Save the submitted batch order and a sanitized copy of its fields with a local batch reference. Retain the response's TrackingId and PartialErrors. The operation warns that its error list is not a positional copy of the request list. An HTTP success or a completed connector job therefore should not be summarized as “all conversions attributed.”

Microsoft's BatchError reference defines Index as the zero-based index of the failed request item. Use it with the original batch order, error code and message. Do not align errors with input rows by simply taking the first error as row one, particularly when only some items failed.

For a Bulk or web upload, preserve that channel's actual processing result instead. Keep its row references and errors with the same underlying outcome ledger. Different transport envelopes should not erase the ability to identify what was sent and which record needs review.

5. Investigate unattributed uploads with the diagnostic report#

A current Microsoft API provides more evidence than a single campaign total. GetOfflineConversionReportByGoalIds retrieves offline-conversion summaries and diagnostics for the account, with daily breakdowns and unattributed results. Its documented request supports up to 20 goal IDs and a UTC range of up to seven days, with an optional upload-channel filter.

Have the account's authorized implementation owner retrieve the relevant interval and retain the request scope, returned summaries, alerts and any goal-level errors. Do not treat the existence of this documented operation as proof that your current connector already calls it or that every interface exposes equivalent diagnostics.

Keep upload-day diagnostics separate from a campaign report arranged under its own attribution dates. A diagnostic count is not automatically the denominator for campaign conversion rate or cost per lead. The platform measurement comparison guide explains how to preserve each report's population and clock before joining them.

If the diagnostic result is unavailable, say precisely which evidence remains missing. “No item-level errors observed” is a narrower conclusion than “Microsoft matched every outcome.” Avoid rewriting unknown results into a successful-import count.

A hypothetical readiness and acceptance audit#

Suppose a business has 200 historical contacts. An analyst identifies 12 genuine sales outcomes that meet the business definition and have documented occurrence times. The remaining 188 are contacts or unresolved records, not conversion rows to upload automatically.

Of the 12 outcomes, the audit finds:

Preparation resultOutcomesNext action
Genuine associated MSCLKID and documented eligible timing8Prepare under the selected click-ID contract
Intended enhanced route with identifier-preparation evidence2Confirm goal/channel requirements and matching acceptance
Missing both a usable click ID and verified enhanced-route evidence2Retain unresolved; do not invent identifiers

Ten outcomes are candidates for a controlled import after the remaining checks. That is 10 / 12 = 83.3% of the defined outcomes with a prepared route, not an 83.3% match rate and not the share of all 200 contacts caused by Microsoft ads.

Assume a subsequent authorized submission of those ten has one identified item error. Nine have no observed item error. The analyst retains the failed record for correction and checks attribution diagnostics for the other nine. Until those diagnostics and the intended performance report are reviewed, nine is a processing observation, not nine attributed sales.

Every number here is hypothetical. A rise in a conversion column after correcting a mapping demonstrates changed reporting; it does not establish additional sales or better campaign performance.

Common mistakes and an acceptance checklist#

Avoid using a list size as an outcome count, copying GCLID into a Microsoft field, replacing event time with upload time, or replaying the whole ledger every time one item fails. Also avoid treating a current API exception as universal support in an unverified connector.

Before calling the workflow ready, retain:

  • The business outcome definition and genuine occurrence time.
  • The intended account, offline goal ID/name and configuration.
  • The selected channel's field contract and matching route.
  • Identifier-preparation evidence kept in the appropriate restricted system.
  • Time, window, value and currency checks for the submitted records.
  • Original batch order, processing response and item errors.
  • Separate attribution diagnostics and performance-report acceptance.
  • A precise retry or investigation action for every unresolved record.

Keep sensitive contact data out of public screenshots, examples and editorial artifacts. Use nonidentifying internal references in review packets. Microsoft's sandbox permits functional offline uploads but does not attribute them into sandbox performance reporting, as its Bulk reference states. A sandbox request test therefore cannot prove production attribution.

For supported campaign delivery already measured by AdProof, retain a scoped delivery export beside the separate CRM and Microsoft evidence. The anonymized fashion-retail delivery-quality case demonstrates an impression baseline; it does not establish CRM imports or offline sales attribution. Contact AdProof to confirm supported inventory and available export fields for the delivery investigation. The account and CRM implementation owners retain responsibility for the offline conversion pipeline.

Frequently asked questions#

Is MSCLKID always required for an offline conversion?

The current Campaign Management object permits omission when a hashed email address or phone number is supplied for enhanced conversions. Verify the goal, chosen channel and identifier requirements. Do not generalize that exception to an unverified spreadsheet template or assume it guarantees attribution.

Can I upload customers acquired through another ad platform?

A contact's existence or another platform's attribution is insufficient. Establish a genuine defined outcome and evidence for the intended Microsoft matching route and eligibility. Keep unsupported or unknown records separate rather than labeling the whole list Microsoft conversions.

Does a successful API response mean every sale was counted?

No. Inspect item errors using the original request order, then review the applicable attribution diagnostics and report. Processing, matching, attribution and report selection are distinct acceptance checks.

Should I change timestamps or resend everything when reports remain empty?

Preserve the real outcome time and the original submissions. Locate the missing boundary using goal, identity, timing and diagnostic evidence. Correct or retry specific records only through the supported process, with their original business identity retained.

Your next step

Choose the matching route for the outcome

Offline imports need a defined business event, conversion goal and supported identifier path. Discuss which route fits your records.

Discuss measurement options

A short inquiry to scope relevant options; no campaign access is needed.

Back to all field notes