Why It’s Important to Measure (and Why It’s Harder Than You Think)

Why It’s Important to Measure (and Why It’s Harder Than You Think)#

Measurement is what makes the difference between “throwing money into the void” and “actually growing your business,” whether you’re Google with a $200 billion ad business or a small business that spends $200 a month on Facebook ads.

Big companies are risking billions. They need to know:

  • Which campaigns brought in sales and which ones just looked nice

  • If that Super Bowl ad was worth the $7 million it cost

  • If their attribution model is lying to them (hint: it probably is)

Small businesses have the opposite problem: they don’t have enough money, so every dollar counts. If you spend $50 on the wrong audience, you’ll have to buy groceries for the week. They need measurement even more than the big players, but they don’t have as many resources to do it right.

What Measurement Really Means#

At its core, measurement in AdTech is trying to answer three questions that seem simple but aren’t:

  1. What happened? Did people see the ad? Do you want to click it? Do you want to buy something?

  2. What caused it to happen? (Was it the ad that made them buy, or would they have bought anyway?)

  3. What can we do better next time? (What do we need to change?)

The first question is not too hard. The second one is where things get deep and costly. The third one is where people make or break their careers.

The Toolkit for Measurement#

Platforms use a few key methods to really measure what’s going on:

Tags (Tracking Pixels)#

Small bits of code that are put on websites and run when someone does something, like loads a page, clicks a button, or buys something. It’s like leaving breadcrumbs, but the breadcrumbs tell the main office what they find. For example, everyone tells you to install that “Facebook Pixel.” It’s a label.

SDKs (Software Development Kits)#

Web-based tags don’t work for mobile apps. Instead, you put an SDK right into your app code that keeps track of everything, like installs, screen views, in-app purchases, and how long someone looked at your checkout page before leaving.

Downside: SDKs make your app bigger and can even crash if the vendor sent you buggy code. You can ask me how I know.

Direct Connections#

Some platforms use APIs to talk to each other directly. Your e-commerce platform sends sales data to your ad platform, which then compares it to data on how many people saw the ads. It’s amazing when it works. When it doesn’t work (which is often), you’re up at 2 AM trying to figure out why test orders keep showing up in production reports.

The Problem with Different Devices#

This is where things get complicated: people don’t just use one device anymore. They see an ad for the product on their phone during lunch, look it up on their laptop at work, and then buy it from their couch on their tablet that night. How do you put those pieces together? How can you tell it’s the same person?

Identity matching tries to fix this by connecting device IDs, email addresses, login information, and behavioral signals to make a single picture of a user. It works sometimes. It may think you and your spouse are the same person because you both use the same iPad. This is why campaigns give out “reach” numbers that seem suspiciously low: they’re trying to make sure they don’t count you three times just because you have three screens.

In AdTech, everything is based on measurement. If you get it wrong, you’re flying blind, optimizing for metrics that don’t matter, or worse, making decisions based on data that’s just plain wrong. If you do it right, you can see what’s working, get rid of what’s not, and turn advertising from an art into a science. Or at least a guess based on a lot of information.