Connecting TheFork to Microsoft Ads

Architecture

How TheFork and Microsoft Ads actually connect

Two systems that know nothing about each other. Everything below exists to give them one shared reference.

1. Step 1

   TheFork holds the truth

   Marketplace covers carry a per-cover fee and frequently a 30% or 50% food discount. The revenue on those covers is not comparable to a full-price direct booking and should never be averaged with it.

2. Step 2

   Microsoft Ads needs a click to match

   Microsoft matches uploaded conversions to clicks using the MSCLKID, and the conversion goal must be created as an offline conversion type. The click has to fall inside Microsoft's import window — verify the current limit in your account rather than assuming it matches Google's.

3. Step 3

   The identifier is the bridge

   The MSCLKID, captured at the visit and carried into the TheFork reservation. Without it there is nothing to join on and no tool fixes that.

4. Step 4

   The event source decides reliability

   Send seated covers at the actual discounted spend, not the menu price.

5. Step 5

   Deduplication at the destination

   Microsoft deduplicates on the combination of click identifier, conversion name and conversion time. Uploading the same click against a second conversion goal counts twice, which is the usual cause of inflated numbers here.

Diagnose it honestly

Working vs. broken, side by side

Both states look identical from the Microsoft Ads dashboard, which is why this runs for months before anyone notices.

Diagnostic sequence

Diagnostic order

Each step rules out a class of cause. Run them in order and the last one usually has a single explanation standing.

1. 1

   Count reservations carrying the MSCLKID

   The ceiling on everything else. Do this before debugging anything downstream.

2. 2

   Follow one real conversion end to end

   From the ad click through TheFork and into Microsoft Ads.

3. 3

   Check whether the discount is reflected in the imported value

   Usually it is not, and the overstatement is proportional to how heavily you promote.

4. 4

   Confirm auto-tagging is enabled and check the upload diagnostics

   One account setting decides whether any of this works. Then read the per-row errors on the upload rather than the success count.

5. 5

   Read the match rate, not the success message

   Microsoft Ads reports offline conversion uploads with rows accepted and rejected. A successful upload is not a matched conversion — check that conversions actually appear against campaigns afterwards.

Source-of-truth validation

How it gets proven

Against TheFork's own records, not against either platform's dashboard.

• The MSCLKID confirmed present on a live reservation.
• A real conversion followed from click through TheFork to Microsoft Ads.
• Auto-tagging confirmed on, and the accepted count reconciled against conversions appearing in campaign reporting.
• Seated covers reconciled at actual spend against imported conversions, with marketplace volume separated.
• Reconciled over a fully closed window, matched per record rather than on totals.
• Re-checked 30 days later.

Straight talk

When you don't need this

Worth reading before paying anyone, including us.

• If Microsoft Ads already reports conversions that reconcile against TheFork's records within a few percent, this join is working and there is nothing to buy.
• Measure identifier coverage yourself first — pull your last hundred reservations and count how many carry the MSCLKID. Ten minutes, free, and it tells you whether any of this is worth doing.
• If TheFork is purely a marketplace channel for you, the question is whether the discount economics work — not conversion tracking.
• If you are not running Microsoft Ads at meaningful spend, skip it — though for B2B the traffic is often cheaper and better qualified than people expect, which is worth testing before dismissing.