forumNew topic

Google Ads conversion tracking via GTM, but seeing two different numbers

BBurcu V***Member
Job title
Field sales representative
Sector
Automotive aftermarket
Organization type
120-person company
Joined
May 2023
Message
2
#1

We run an e-commerce store in Lyon selling wholesale and retail home textiles. I mean our monthly ad spend is around 3,500 EUR. We set up both the Google Ads conversion tag and the analytics purchase event via Google Tag Manager, firing off the same order confirmation trigger.

Looking at the ad dashboard last month it reported 142 conversions, while the analytics side only logged 88 orders attributed to Ads sessions. This 54-conversion gap makes it impossible to decide which product lines to allocate our budget to.

What exactly is causing this discrepancy between the two dashboards? Are we double counting, or are we losing data somewhere and how can I verify and debug this in GTM?

DDeniz K***ExpertCommunity member
Joined
Nov 2024
Message
154
Most Helpful#2

Short answer: The discrepancy between Google Ads and your analytics dashboard stems from completely different logging logic, attribution models, and cookie consent handling. Google Ads attributes the conversion to the day the ad click occurred, whereas analytics attributes it to the actual session when the purchase took place; as a result, the two platforms rarely match up 1:1 over the same date range.

To find out whether that 54-conversion difference is caused by technical double-counting or just differing methodologies, go through this checklist in GTM:

1) Check tag firing frequency: In GTM, open the advanced settings of your Google Ads conversion tag and make sure 'Tag firing options' is set to 'Once per event'. If left on 'Unlimited', Ads will log a new conversion every time a customer refreshes the thank-you page or navigates back to it.

2) Deduplicate with Transaction IDs: Always pass your platform's unique order number (Transaction ID) into your purchase tags as a dynamic variable. The Ads engine will automatically drop duplicate fires carrying the same order ID, preventing double counts.

3) Cookie consent settings: Consent choices among users in France directly impact your numbers. While analytics completely drops the session if cookies are rejected, Google Ads can use enhanced conversions and Consent Mode signals to report modeled conversions. Don't make any budget shifts until you place a test order in GTM preview mode and confirm both tags fire strictly once on the thank-you page.

HHasan D***Member
Job title
Customer Relations Manager
Sector
Logistics
Organization type
40-person manufacturing company
Joined
Feb 2025
Message
12
#3

Refresh the thank-you page while running GTM Tag Assistant. If the `purchase` data layer event fires on every page reload, your Ads tag is triggering again. To fix this, either validate unique order IDs server-side or gate the tag with a cookie check so it only fires on the initial post-checkout landing.

MMeryem K***Member
Job title
Graphic Designer
Sector
Leather
Organization type
family business
Joined
Jul 2025
Message
417
#4

Last year when we were spending 5,000 EUR, Ads was reporting 210 conversions while analytics showed 140. Looking into it, we realized mobile shoppers were leaving the thank-you tab open, firing the tag every time they relaunched their browser. Once we mapped the Transaction ID, Ads dropped to around 160 the following month, and the numbers finally aligned.

GGamze K***MemberCommunity member
Joined
Oct 2022
Message
3
#5

What attribution model are you using in Analytics—last click or data-driven? Also, when comparing the reporting windows across both dashboards, are you looking at time of click or time of conversion?

CCerenMember
Job title
QA Tester
Joined
Mar 2024
Message
178
#6

You will never get these two dashboards to match penny for penny. One retroactively attributes to the click date to make its own ads look good while the other balances across all channels. Instead of wasting time trying to zero out the discrepancy, base your numbers on the actual invoices in your own order dashboard.

HHüsniye C***Member
Job title
Information Security Specialist
Sector
Insurance
Organization type
40-person manufacturing company
Joined
Apr 2022
Message
295

Doki · Brand identity · 2024

#7

due to cnil rues in france tons of users reject cookies but analytics just trashes that data completely, but if consent mode v2 is on in ads, it uses ai to model the missing data and adds it to the dashboard part of the gap might be coming from that.

İİbrahim S***Expert
Job title
Store Manager
Sector
Printing
Organization type
20-person company
Joined
Nov 2022
Message
1
#8

Quick checklist to bring the two sources closer: 1) Verify that the Order ID variable is passing properly to both tags. 2) Update the conversion counting method in Ads to 'One' instead of 'Every'. 3) Check if enhanced conversions is enabled. 4) Add a referrer filter to the trigger in GTM to prevent direct page refreshes.

KKader B***Expert
Job title
IT manager
Sector
Agriculture
Organization type
boutique agency
Joined
Mar 2023
Message
233
#9

we went through the exact same thing and almost had a huge falling out with our agency. in the end, we realizd the counter was firing every time customers clicked the tracking link and went back to check their order page. a simple deduplication fix solved all the headache dont sweat it.

ZZaferMember
Job title
Insurance agency
Joined
May 2024
Message
88
#10

Go into the Ads dashboard right away and check your counting method under conversion settings. If you're doing e-commerce, transaction IDs are essential, but as a quick fix switching the count type to 'One' will immediately filter out multiple fires within the same session.

TTülay B***Veteran
Job title
Human Resources Manager
Sector
Seafood
Organization type
300-person organization
Joined
Jan 2023
Message
35
#11

I have a question. Speed is a threshold issue; if you're already fast, making it faster changes nothing.

Good luck with that.

TTülay K***ExpertCommunity member
Joined
Jul 2022
Message
276
#12

Correct. Any comment made without a control group is just a guess.

If you have questions, write them; I'll answer as best I can.

FFurkan A***Veteran
Job title
Supply chain manager
Sector
Education
Organization type
boutique agency
Joined
Oct 2023
Message
1
#13

Saved. The answer varies greatly by industry; there is no one-size-fits-all rule.

I'm also curious if anyone does it differently.

DDoruk S***New memberCommunity member
Joined
Aug 2026
Message
278
#14

I didn't know that. If you get three different answers on a topic, the question was asked wrong.

Don't hesitate to ask; those who don't ask always pay more. That's all, sorry if I went on too long.

MMerve T***VeteranCommunity member
Joined
Mar 2024
Message
282
#15

I feel the same way. A page that answers one question works better than one touching on five topics.

If you have questions, write them; I'll answer as best I can.

EEceMember
Job title
Legal Counsel
Joined
Feb 2024
Message
98

Doki · Backup setup · 2023

#16

It's rare to find an explanation this clear. A page that answers one question works better than one touching on five topics.

Trying to do this alone is the most expensive way. If you post the result here, it will help others too.

GGizem Y***MemberCommunity member
Joined
Feb 2026
Message
40
#17

This approach has a cost, which isn't discussed. When you try to change everything at once, nothing settles.

That's all, sorry if I went on too long.

FFurkan B***Veteran
Job title
Store Manager
Sector
Machinery manufacturing
Organization type
two-branch business
Joined
Jan 2023
Message
126

Doki · Interface design · 2026

#18

I partly agree, partly disagree. When making a decision, first look at what data you have on hand.

Start with a small trial; don't commit to everything at once. That's all, sorry if I went on too long.

TTolga Y***MemberCommunity member
Joined
Dec 2023
Message
2
#19

How did you solve this? If you get three different answers on a topic, the question was asked wrong.

Any comment made without a control group is just a guess. I'm also curious if anyone does it differently.

ÖÖmer N***MemberCommunity member
Joined
Jun 2022
Message
62
#20

Let me write how it's done in practice. When you try to change everything at once, nothing settles.

This is my opinion I'm not claiming it's absolute truth.

This topic has been closed.The moderator marked the topic as resolved. If you have a similar issue, you can open a new topic.
New topic