Skip to main content

GTM, the dataLayer and Your First Conversion · Part 1: What web analytics is, and why tracking breaks on its own

Chapter 1.1 · 1 of 25 · 3 min

What conversion tracking really is, and how it connects to ad spend

Picture a shop.

Free to read

Picture a shop. A clothing shop in Dhanmondi, a neighborhood in Dhaka, that runs Facebook ads, takes orders through a WooCommerce site, and gets paid through bKash (a mobile wallet) and cash on delivery (COD).

The owner spends 2000 taka a day on ads. Ads Manager says there were 14 purchases. The shop's own order list shows 9. So the question is: which number is true, and where did the other 5 go?

Conversion tracking is the work of answering that question. No more, no less.

What an event is

Whenever someone does something on a site, that is an event. Viewing a page, opening a product, adding something to the cart, starting checkout, completing an order. Each one is a separate event.

Not all events are equal. Viewing a page and completing an order are both events, but one is tied to money and the other is not. An event tied to money is what we call a conversion.

Where the money comes in

Facebook decides who sees your ad based on past results. If you never tell Facebook who actually placed an order, what will it learn? Nothing. It will go looking for people who click, because a click is the only thing it can see for itself.

The people who click and the people who come to buy are not the same people. So if you do not send a purchase event, your ads will bring cheap clicks, and the money is wasted.

That is the real job of conversion tracking. Not building reports. Teaching the ad platform.

An extra complication in Bangladesh

COD. With cash on delivery, an order being placed and money being received are not the same thing. Out of 100 COD orders, 30 coming back is normal.

So when do you send the purchase event? When the order is placed, or when the money arrives? Both are valid, but they mean different things. Send it at order time and Facebook learns faster, but the numbers are inflated. Send it when the money arrives and the numbers are true, but it has to be sent from a server, and that is the subject of the next book.

The same split applies to any business where an order can fail after checkout: returns, refunds, chargebacks, invoices that are never paid.

In this book we send the event at order confirmation, and we tell the client in writing that this number is orders, not money received. That one sentence in writing saves a lot of arguments later.

Steps

  1. Pick a site, your own or a client's, that is running ads. In Ads Manager, write down the number of purchases for the last 7 days.
  2. Write down the number of orders for the same 7 days from the site's admin panel.
  3. Work out the difference between the two numbers. Write it as a percentage too.
  4. Think about why the difference might exist. You do not need the answer yet; the reasons come in chapters 1.6 to 1.9.

What not to do

  • Do not treat the Ads Manager number as the truth. It is the platform's own count, inside its own attribution window. The site's order list and Ads Manager will never match exactly, and trying to make them match wastes time.
  • Do not promise the client you will make the two numbers equal. You cannot. What you can do is explain why the gap exists and reduce the part of the gap that should not be there.

Proof

Your notebook has two numbers and the gap between them as a percentage. And you can write in one line why an ad platform needs a conversion event.

Want the full chapter list, or to jump to another chapter?

All chapters in the book