We’ve all seen dashboards that don’t make sense when you look at the numbers, but now that same data could be causing your campaigns to spend your budget chasing the wrong people.
As automation becomes more and more of the ad buying process, from creative generation to bidding, data has become one of the last things advertisers can control, and perhaps the most important. Indeed, automation can only optimize the signals you give it.
Think about it: Which is worse: a brilliant ad delivered to the wrong audience or an average ad delivered to the right audience? The first is spending your budget reaching people you don’t want. The second could be ignored, but if someone commits, at least it’s the right person.
But can you honestly say that the last time you created a campaign, you spent more time checking the data than thinking about the ad copy?
The cost of bad data has changed
Several years ago, poor tracking was a reporting problem.
If a tag fired twice, a conversion was mishandled, a value wasn’t passed correctly, or your offline conversions stopped working for a few weeks, the result was a dashboard that didn’t add up. It was boring, but it had little impact. Eventually, someone would question the numbers during a monthly review, you would trace the problem, correct it, and the data would be useful for the next review.
However, that same data now feeds the algorithm that buys your paid media. Smart bidding strategies don’t wait for you to interpret a report or hit your monthly valuation—they read your conversion data and act on it before you even notice a problem.
The same number, now wrong, has a different result. A wrong number in a report requires an explanation in a meeting.
A wrong number in a conversion used for bidding costs you money because the algorithm doesn’t know it’s wrong. It optimizes for this signal as soon as it sees it, and it does so efficiently.
Google doesn’t understand your funnel or your business
Although conversion actions are labeled in the Google interface as “lead”, “opportunity”, etc., these labels are intended for the organization only. The platform doesn’t actually understand where the conversion event is in your funnel.
All he sees is a conversion event that has a numerical value attached to it (usually representing a currency amount), so he has no idea that a newsletter signup is worth $2 in final value, that a lead is worth $60, and that an opportunity is worth $400. Google sees three conversions. He has no idea that one is worth 200 times the other.
The algorithm does not optimize your business results. This is an optimization for the data you provided to it. If the data is wrong, so will the optimization.
Your customers are searching everywhere. Make sure your brand introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
@media (maximum width: 768 px) { .headline-responsive { font-size: 30px !important; line-height: 1.3 !important; } }
For example, if every form submission triggers the same conversion with the same default value, there is no clean way to distinguish tire removal requests from high-value requests, so the algorithm treats them the same. And since tire kickers are generally cheaper to acquire, you’re inundated with them.
The cost per lead drops from $40 to $25, and the dashboard makes your cost per lead appear more than 35% lower, but the pipeline quickly dries up as truly qualified inquiries quietly drop in half.
3 ways bad data is quietly destroying delivery
Bad data can take many forms, but these are the three problems most likely to derail the campaign.
1. Bad event
Optimizing for a top-of-funnel action, like a page view, when the actual conversion events are happening further down the funnel, causes the algorithm to buy more and more of these cheap events without the bottom-of-funnel activity actually continuing.
2. Bad value
Count all conversions equally (or assign them a flat placeholder value) when their actual value varies by 10x. The algorithm prioritizes the volume of lower-value conversions because they are easier to acquire.
3. No data
This one isn’t discussed enough. Nothing kills a campaign faster than a complete data breakdown.
On the first day, the algorithm wonders where the conversions are. By the second day, we start to assume that they won’t come. On the third day, serious changes are made to the auction. Within a week, most campaigns will have amounted to almost nothing.
How to choose the right signal for Google
So how to fix it?
Let’s take the example of a typical lead generation business. Some leads will never convert, while others are worth 10 times more than the rest.
If your form asks the right qualifying questions, you already know what they are. But if you optimize every submitted lead using a target CPA, you’re telling Google that they’re all equally valuable.
Imagine an account spending $20,000 per month with a target CPA of $40 and generating around 500 leads. Only 150 are eligible, and perhaps only 50 have real added value. The expected value of a lead is $60, the expected value of a qualified lead is $200, and the expected value of a high-value lead is $600. This represents a 10x difference in value.
There are several ways to improve the optimization signal:
- Optimize for a qualified lead:Create a new conversion action, such as “qualified lead,” and only launch it for valuable leads. You can then move your target CPA to this conversion action, knowing that it will ignore worthless leads. The benefit is that you drive the campaign on a more meaningful signal. The downside is that every valuable lead is still treated the same.
- Assign conversion values and useTarget ROAS: Add a monetary value to the qualified lead based on the potential revenue they could generate if they converted into a sale. You can then switch the campaign to target ROAS, allowing Google to optimize the return instead of just counting worthless leads. However, it can still buy a greater number of lower-value leads if it can acquire them at the right price, rather than prioritizing higher-value ones.
- Optimize for a high-value prospect: Create a “high-value lead” conversion event that triggers only for your top-tier leads, with or without conversion value. You can then optimize with a target CPA or target ROAS, depending on whether you want to focus on acquisition cost or return. The benefit is higher quality leads. The downside is that, depending on your spending, the data may be too limited to support this approach until you scale.
These are just a few possible optimization signals without even going deeper into the funnel. You can also apply the same approach to bottom-of-funnel events by creating separate conversion actions for milestones like Contacted Lead, Qualified Contact, or High-Value Contact.
See the full picture visibility of your search.
Track, optimize and win in Google Search and AI from a single platform.
@media (maximum width: 768 px) { .headline-responsive { font-size: 30px !important; line-height: 1.3 !important; } }
Targeting and measurement may be different
It sounds simple, but the conversion event you’re optimizing for and the one you’re reporting on aren’t, and arguably shouldn’t be, the same. We train the algorithm. The other tells you how this training takes place.
In our previous example, a customer or internal stakeholder may want to know the cost per lead, which is a perfectly valid metric. During this time, the campaign is optimized for converting the qualified lead, not the original lead.
You keep the original lead conversion only as a reporting metric, so stakeholders continue to get their cost per lead while the campaign bids on the qualified lead signal that actually drives business value. Same campaign, two conversions, two very different jobs.
Which brings us back to where we started: Did you spend more time verifying the data than writing the ad? In an automated account, data is now your strategy.
The position Bad data teaches AI to waste your advertising budget appeared first on MarTech.
