Explain why each platform reports a different number of conversions for the same campaign.
Pull up three numbers for last month: the conversions Meta Ads Manager reports, the conversions Google Ads reports, and the total orders or enquiries in your own sales records. Write them next to each other. They won't agree. Before you use any of them to decide where your ad money goes, you need to know why.
Farah sells through an online shop and also has a bazaar stall. At the end of November she did this check. Meta Ads Manager claimed 34 purchases and Google Ads claimed 41 conversions (example figures). Her shop platform recorded 58 orders in total that month from every source, and that 58 includes people who walked up to her bazaar stall and ordered on their phones. If both ad platforms were right, ads alone produced 34 + 41 = 75 sales in a month with only 58 orders in total.
An ad platform sees its job as showing what its own ads achieved. It counts a conversion when it can link a purchase or lead to someone who interacted with one of its ads, within a time window the platform sets. The pixel or tag on your site reports those actions back to the platform. Lesson 3.1 of the course "Paid ads: Meta, Google and TikTok", "What a pixel does and why the auction needs it", explains how that works.
The time window is called the attribution window: how long after an ad interaction a conversion still counts. Meta, Google Ads and TikTok each set their own default windows. The defaults can differ by campaign type, and each platform lets you change them. So a purchase made eleven days after a click may count on one platform and not on another. Defaults change, so check the current ones in each platform's help centre and in your own campaign settings.
Platforms also differ on which date they assign a conversion to. Some use the date of the ad click and others use the date of the purchase. The same sale can therefore fall in October in one tool and in November in another.
GA4 uses its own attribution model, which you met in lesson 5.1. It sees only what happens on your website, and it has no special loyalty to any ad platform.
Some platforms count a view-through conversion: a conversion from someone who saw an ad, did not click it, and later bought. For example, someone scrolls past a shop's ad on Instagram without clicking. Later they type the shop's address into a browser and buy. Meta can count that purchase, because the person was logged in to its app and Meta knows they saw the ad. GA4 can't see the ad view, so to GA4 the same customer arrived directly or through a search. The ad platform claims the sale and GA4 credits direct or organic search. Neither one is lying. They are looking at the same customer with different information.
Platforms also use modelling to estimate conversions they can't observe directly, for example when people decline tracking on their phones. These modelled conversions appear in platform totals with no individual record behind them.
Then there is overlap. Farah's customer from lesson 5.1 clicked a Meta ad, later searched on Google, and bought from the newsletter. Suppose that customer had also clicked a Google search ad along the way. Meta would count the purchase, Google Ads would count it too, and GA4 might credit email, so one sale would carry three claims. Across a month, overlapping claims like these push the totals well past reality. Adding up conversions from every platform almost always gives more sales than were actually made. This is how a business running several channels can feel as if every channel is profitable while its bank account says otherwise.
Decide which system counts the real total, and stop asking the ad platforms for it. Your source of truth is the system that records each actual sale or lead once, with a name or order number behind it. For most businesses this is the shop platform, the booking system or the CRM. For some it is the accounting system. Farah's source of truth is her shop platform, which shows 58 orders. Priya runs a maths centre in Tampines (an invented example), and her source of truth is her enrolment spreadsheet.
GA4 sits between the platforms and your source of truth. It is a good guide to where website visitors came from, but it is also an estimate. Lesson 4.2 explains why.
The other numbers are still useful, as long as each one is used only for its own job:
Meta's purchase count is for comparing Meta campaigns, ad sets and ads with each other, because all of them are counted by the same rules. Google Ads conversions are for comparing Google campaigns and keywords with each other, and they also feed Google's bidding. GA4 is for comparing channels on one consistent basis and for understanding how people behave on your site.
That means three habits change. Stop comparing Meta's numbers directly with Google's, stop adding platform numbers together, and stop reporting platform totals to the owner as sales.
Once a month, set out three things next to each other, the way Farah did: each platform's claimed conversions, GA4's key events, and the real total from your source of truth. Don't try to make them match. Instead, watch how the gap between them moves from month to month.
Suppose Meta has claimed about 60 percent of your real sales for months. Then it suddenly claims 90 percent while real sales stay the same. That means something has changed, either in Meta's counting or in your tracking. Find out what changed before you move any budget.
Building that table for your own business is also the first piece of the attribution note you will write in lesson 5.4.
Take one month. Write down the conversions each platform claims, then compare their total with the real sales in your shop, booking system or CRM.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).