At ZATO, we’re always analyzing YouTube’s role in the marketing funnel, but our latest team discussion took a deep dive into something that many advertisers struggle with: incrementality. Specifically, we dissected a fantastic conversation between Andrew Faris of AJF Growth and Olivia Kory of Haus Analytics on the complexities of measuring YouTube’s true impact.

So let's start there, if you want to watch this, make sure you've seen that first!

‍

The video sparked a much-needed conversation within our team about how YouTube fits into broader media strategies, the shift from Video Action Campaigns to Demand Gen campaigns, and the massive headache that is attribution. Let’s break it down.

BTW, you can READ the Haus team's official release on this study here!

Do YouTube Ads Perform? Lessons From 190 Incrementality Tests

‍

‍

Demand Gen: Is This the Future of YouTube Advertising?

With Video Action Campaigns (VAC) phasing out and Google pushing Demand Gen as its replacement, our team has been putting Demand Gen campaigns through their paces. And if our results so far are any indication, there’s something interesting happening here.

Sarah Vlietstra pointed out that we've also seen what Olivia's data noted, that Demand Gen is outperforming our traditional Video Action campaigns in some cases. Olivia shared that they're seen improvement in DTC (Direct-to-Consumer) ROAS and stronger post-engagement behavior, making us wonder:

👉 Is it time to shift most of our YouTube strategy toward Demand Gen?

Unlike Meta ads, where users often convert quickly, YouTube’s impact is more delayed and indirect—it’s driving awareness and assisting conversions that will come later in the journey, often untracked. This fits with what we’re seeing in Demand Gen. Because these campaigns blend video and discovery placements, they seem to be functioning more like Meta’s top/mid-funnel ads rather than traditional YouTube direct-response campaigns. That changes the way we think about performance expectations, attribution, and creative strategy.

Diana Talhelm raised a great point here: since Demand Gen is working well for us, it could be a great way to introduce YouTube to clients, because it fives clearer performance data and makes it easier to show value upfront.

The Big Problem: Measuring YouTube’s True Impact

Ah, measurement. The topic that makes every PPC marketer's heart hurt.

One of the most valuable parts of Andrew & Olivia’s discussion was their deep dive into the messiness of YouTube attribution. Unlike Meta, where users are almost always logged in across devices and actions can be more directly tied to a single person, YouTube’s fragmented ecosystem makes tracking a nightmare.

Here’s what we’re up against:

1️⃣ Cross-device behavior – A user watches a YouTube ad on their Smart TV, then later searches for the brand on their phone. How do we track that influence?
2️⃣ Logged-in status variability – Unlike Meta, YouTube users are often not logged in across devices. That means less direct tracking.
3️⃣ Google’s own limitations – You’d think Google would have all the answers here, but even their own attribution tools (brand lift studies, conversion lift) require a large spend and aren’t accessible to many advertisers.

Olivia specifically called out that most advertisers are underestimating YouTube’s true impact simply because we don’t have the right measurement tools in place. This aligns with our experience—YouTube is almost certainly underreported in attribution models. The tricky part is figuring out by how much.

Sarah brought up a new Conversion Lift option Google is rolling out that applies to Demand Gen as well. However, the requirements to access it are inconsistent—some reps tell us one thing, others tell us another. (Shocking, I know.) One thing we can know for sure, it's definitely not for SMBs!

Incrementality: The Key to Understanding YouTube’s Real ROI?

Andrew and Olivia made one thing very clear: incrementality testing is the best way to understand YouTube’s true value. The problem? Most businesses can’t afford to do it properly.

Incrementality studies require:
✔️ Big budgets – You need a high enough spend to see statistically significant results.
✔️ Patience – These tests require weeks or months to gather enough data.
✔️ Controlled variables – Which is hard in the real world of shifting budgets and campaign changes.

This is a huge challenge for SMBs. Kirk (that’s me) pointed out that most small businesses simply don’t have the luxury of running large-scale incrementality studies. We need a practical way to measure impact without those high barriers to entry.

A few alternatives based on this were presented by Andrew and Olivia and then discussed by our team:
🔹 Branded search lift – If branded search volume increases after launching a YouTube campaign, that’s a good sign YouTube is driving awareness.
🔹 Comparing holdout geos – Running YouTube in one region but not another and comparing results (though this can be tricky).
🔹 Multi-touch attribution models – Imperfect, but better than last-click attribution alone.

The key takeaway? YouTube is almost certainly driving more revenue than what standard attribution models show. The challenge is figuring out how much.

YouTube Shorts: A New Frontier for Demand Gen?

Shorts came up in Andrew & Olivia’s discussion as well, and it’s something our team has been debating internally.

Do Shorts ads belong inside broader YouTube campaigns or should they be segmented into their own campaigns?

Diana argued that integrating all formats into a single campaign allows Google to optimize performance across the funnel. But Chris Reeves countered that Shorts is a completely different experience than long-form YouTube, and engagement behavior is different.

We’re still testing, but early signs suggest that Shorts might play a bigger role in YouTube Demand Gen than most advertisers realize.

The YouTube on TV Dilemma

One major takeaway from Andrew & Olivia’s discussion was the growing dominance of YouTube on TV (Smart TV apps). Diana pointed out that this is where a large chunk of YouTube Reach campaign impressions are happening.

The challenge?
1️⃣ YouTube on TV is harder to track – A user watching on their TV isn’t necessarily clicking or engaging in ways that traditional digital ad models can measure.
2️⃣ Kids’ content is a problem – Even with exclusions, it’s hard to completely avoid wasted impressions on children’s YouTube consumption.
3️⃣ Logged-in status varies – This makes targeting and reporting murkier than other devices.

Andrew & Olivia noted that YouTube on TV isn’t necessarily a bad thing—it just requires a different way of thinking about measurement and user behavior.

Final Takeaways & Next Steps

After discussing Andrew & Olivia’s video, we left the meeting with a few key conclusions:

✅ Demand Gen is showing promise – But we need to continue testing whether its success is due to the campaign type or simply better alignment with YouTube’s role in the funnel.
✅ Attribution is messy – We need to explore new ways to measure YouTube’s real impact beyond just last-click conversions.
✅ Shorts is worth paying attention to – We’ll keep testing whether it should be segmented or integrated into broader campaigns.
✅ YouTube on TV is a wildcard – More research is needed to determine how it fits into attribution models.

At the end of the day, YouTube is a powerful tool—but measuring it properly requires a mindset shift. If we rely solely on traditional attribution models, we’re almost certainly underestimating its impact.

‍

‍

Additional Random Kirk Key Takeaways from the Original Video

‍

1. YT has way fewer advertisers than Meta, and they are spending less than on FB, lots of opp here!
2. Facebook conversion algorithm is just better!
3. Barrier to entry is much higher on YT than Meta for most brands (cost of making videos, learning platform, etc)
   1. production cost (AI advancements help make this lower)
   2. "wasting" money figuring it out
4. 2/3 of YT viewers watch it on a TV screen!!!
   1. How should this change the sort of videos created?
5. The incrementality range is 1-10, which even on the low end is pretty darn amazing! There just seems to be lower risk in trying YouTube than brands think.
6. Weirdly: Google is really bad at reporting its value through YouTube whereas it tries to grab all credit from other channels.
   1. So much opp here for Google to help identify and communicate incrementality better for YT.
7. Starting out? Saturate first on meta, run a holdout test there and watch for efficiency drop,  then you know to move to test YT.
8. Even tho we were annoyed by Google killing Video Action Campaigns for Demand Gen, DG has actually been outperforming VAC in Haus clients!
9. YT TV performs worse than YT... so if you have limited dollars start with YT and don't lean into YT TV yet(this matches other CTV analyses they've seen)
10. YT Shorts: 
    1. Olivia isn't sure why you would break out shorts into separate targeting and campaigns
    2. Andrew says: "it is very different as an experience to the user, it is a feedbased experience so viewers have more user comparability to Reels, TikTok, etc" - this also matches what people like Cory Henke have said about treating Shorts like "Entertainment"
    3. Olivia responded"that's a great point, Shorts could be the low barrier to entry, lower cost option for brands to get profitably started on YouTube"

1. For smaller brands:
   1. stop trying to figure out incrementality, stay more single channel so you can actually see what happens with the numbers.
   2. Use last click data to determine FB (if Meta says you've 5xed your revenue, but no last click data increase in sales, then you should investigate that), not to tell you the whole story, but because it's a stable form of measurement that never lies. It's directional in nature, NOT the whole story
   3. As Google is layered in, things get VERY complex because of the difference in intent.
   4. Andrew wants this:
      1. the intro to this incrementality problem for Google Ads
      2. The Google system is a fraction of the spend of Meta, but a big question mark
   5. Haus is kicking the tires on an ala carte model for SMBs, but it's very complex. Incrementality testing needs a lot of service and consulting on applying the answers correctly.
      1. BUT IMPORTANTLY: "Doing incrementality testing poorly can be more destructive than not doing it at all"
2. You can use brand search revenue growth to determine if upper funnel is working. But ultimately "forecast your P&L and keep Ad spend a certain % of revenue"
   1. "if you're P&L is working it's working"
   2. Olivia: ultimately, if you don't know if Google will work, shut it off and see what happens. If there is no obvious impact on sales, then it's prob not helping you much.
   3. Andrew: there are also non-brand "almost brand" terms that are a search keyword basically created by a brand where their clients only search for those words to find THEM, that's basically brand. (Kirk caveat: just be cautious before shutting these off in case you have competition in the space!)

‍

‍

Big thanks to Andrew Faris & Olivia Kory for sparking such an insightful discussion! 🚀

Big thanks to Jack Hepp of Industrious Marketing, one of our ZATO freelancers, who hopped in there to help us think as well!

What do you think? Have you found a better way to measure YouTube’s true influence? Drop a comment on social media (@PPCKirk)—we’d love to hear your take!

‍

‍