Understanding campaign performance: Answers to common conversion tracking questions

Last month, I wrote about the importance of knowing pay-per-click (PPC) fundamentals when you’re hiring new team members.

While you can control the team you hire, it’s harder to find clients, bosses and stakeholders who understand all the details of PPC as well as you do. This can lead to questions about campaign performance when conversions don’t seem to be tracked correctly, which in turn can lead to prickly situations where jobs can be at stake.

A client’s questions about some confusing aspect of AdWords need to be answered correctly if you want to look like the PPC rockstar you are.

Here are a few common conversion questions we are frequently asked and their answers.

Why did you report fewer conversions than what I see?

The first confusing situation we encounter is when a client uses our report engine to share account performance. A few days after the report was automatically delivered, the client calls and complains that the number of conversions in the report doesn’t match what they see in AdWords.

Fumble your response and they’ll think you don’t know what you’re doing, but answer it right and take a huge step in building trust with the client.

Why conversion data may be off

AdWords lets you set a conversion window to select how long to track conversions after a click or interaction, usually set at 30 days.

This means the data associated with a click can change up to 30 days after it happens. A click that happened last month can see its conversion count go up, so long as it’s still within the active conversion window. Metrics like clicks, cost and impressions tend not to change much once reported, but conversions, conversion value, cost per acquisition (CPA) and other conversion-related metrics remain to live well after the date the click happened.

You can set a conversion window so you can track conversions that take a few days to happen in AdWords

Some people mistakenly believe AdWords reports a conversion on the day it happens. That would be true in Google Analytics, but remember that AdWords reporting revolves around clicks, whereas Analytics revolves around actions on a site.

That may seem like a small difference, but it has a huge impact on what happens with conversion data. Analytics will track a conversion on the day it happens, regardless of what previous touch points led up to it.

In AdWords on the other hand, a conversion gets added to the reports on the day of the last interaction, assuming you are using a last-click attribution model.

So, unless you send reports where the last included date is older than your conversion window, clicks may still have an open conversion window, and the associated conversion data can still change.

In practice, if you wanted to avoid this confusion, you’d have to send reports that don’t include clicks from the last 30 days. This is not a practical solution, since clients typically want reports as soon as the month ends.

If you want to get a better sense of how many conversions are still likely to occur after you generated a report, look at the “days to conversion” segment in the AdWords interface.

I don’t do e-commerce, so why are my conversions fractional?

Now you’ve gotten the client to understand why conversion numbers change throughout the month, but they push you on another weird thing in AdWords. The client says they do lead gen so it makes no sense that there are fractional values for conversions because after all, there’s no such thing as half a lead submission.

Or is there?

The industry is moving away from last-click attribution models and, depending on the model you’re using, one conversion can be split between several interactions.

For example, with a linear attribution model, every interaction leading up to a conversion gets an equal share of the credit. In a simple example, if the user clicked on two of your ads for different keywords, each one of those clicks gets half a conversion credited.

Other attribution models split the credit in other ways. I recommend Mona Elesseily’s excellent post about attribution models if you want to understand this better.

Aren’t attribution models just a money grab by Google?

So, now your client understands what fractional conversion numbers are, but they want to know why attribution models matter. Isn’t this just a way for Google or the agency to extract more ad budget without improving performance?

The answer to that is complex. At some basic level, a better attribution model is a way for Google to keep getting lots of money from advertisers, but the complete answer also looks at why this would happen.

Google gets more advertisers by giving better insight into how ads are contributing to business goals. If they can correlate ad spend with real-life sales and conversions, then yes, they will continue to spend money with Google. This should only be done because it makes sense for the business and advertisers are pleased with that performance.

New attribution models are important because they try to answer the complex question of what the real value of online advertising is.

User behavior is dynamic, and you need advanced ways to measure how each touch point contributes to the end goal of getting a conversion. Simply looking at the last click is not good enough anymore.

Explaining attribution to non-digital marketers

This concept makes sense to us as online marketing experts, but we still have to explain it to clients who may be resistant, so let’s use a simple example of a car dealership that leaves digital out of the equation.

A last-click attribution model would be like saying that people visit a dealership because they see the sign by the entrance. Because the sign gets people into the showroom, the dealership decides to invest more in making that sign bigger and flashier while cutting their budget for all other marketing.

Soon after the bigger sign goes up, sales tank. What happened?

They missed the fact that most people drive to a dealership because they’ve been influenced by television commercials and ads in magazines. If the dealership invested nothing in those channels, nobody would visit the dealership, no matter how beautiful their sign was.

Now let’s look at the car dealership example in the digital world and see what might happen if they stuck with last-click attribution while their competitors start to experiment with more comprehensive models.

By relying on last-click attribution to record conversions, and using that data to manage bids, they might start to lower bids for keywords used early in the decision process.

Broad keywords like “safest suvs” would get low bids because those are not what the users click on right before contacting a dealership. By undervaluing these early touch points, competitors are allowed to swoop in with remarketing lists for search ads (RLSA), dynamic remarketing and data-driven bidding strategies.

Competitors catch a user’s attention early and remain engaged with them through a variety of marketing technologies. Keywords that used to immediately precede a lead see a drop in volume and the dealership sells fewer cars.

The ideal attribution model also looks at offline touch points and includes lifetime value, but those can still be harder to track.

Procter Gamble (PG) decreased their digital ad budget by an estimated $200 million in 2017 because they felt their money was better spent elsewhere. That type of measurement is complex, but that doesn’t mean we should remain stuck in the way conversions have been measured in AdWords since they first launched conversion tracking.

Conclusion

Conversions are the lifeblood of any business, so it’s no wonder clients get panicky if they feel their account managers are not tracking things correctly. One of the greatest aspects of online advertising is that it’s highly measurable, which makes it possible to use granular targeting to deliver stellar results.

I hope sharing how we answer questions about seemingly weird conversion issues helps you have better conversations and in turn helps you build trust and profitable relationships.


Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.


About The Author