Demand Gen. A Shameless Ode to AI.

Disclaimer: I am not endorsed by Google, I merely use their products to advertise online.

In this article, I’ll focus specifically on Demand Gen, however my analysis of this new campaign type - its strengths and weaknesses - equally apply to Performance Max and other AI ad types. It’s also important to note that my experience with ad platforms and AI are very much my own personal insights.

Demand Gen campaigns were launched with the intention of building brand awareness, generating leads, and nurturing customer relationships through multi-channel advertising and AI-powered targeting across Google's network - which sounds like jargon, but we’ll get there.

Demand Gen was touted as the much-anticipated replacement for Google discovery ads, while the concept for Google discovery campaigns made sense, it was very underwhelming for those who were looking beyond the valley of vanity metrics and into the depths of Google Analytics.

But Demand Gen promised to change that - introducing a consideration layer into the Google ecosystem that could drive incremental reach AND deliver tangible business outcomes by driving conversions with its AI powered infrastructure - HALLELUJAH!

Admittedly, it was a slow start for me with Demand Gen. I tried it multiple times in 2023 and was left feeling a bit disheartened. Impressions? Yes. Clicks? Eh, sometimes but conversions? No.

I would argue there was a lot riding on this for Google, and, despite my internal claims of a pink and orange sunset on the horizon of this AI driven calamity - Google didn’t waiver.

Come mid 2024, Demand Gen was back on my to-do list -  not necessarily by choice but by necessity. Clients wanted something fresh. They had their search campaigns capturing high-intent audiences, YouTube videos ringing out across mobiles and TVs to drive both reach and brand awareness, but they wanted more; impressions, clicks and conversions - “I want it all”.

Challenge accepted.

So I swallowed my pride and jumped back into bed with Demand Gen, expectations low.

It's worth noting that I'm a big advocate for testing. Experimentation is vital for personal and business growth. When conducting these kinds of tests, it is important to have a clear understanding of what you hope to achieve and what you’d be happy to achieve.

It’s not an exact science, but if it looks like shit and it smells like shit, it's probably shit. Cut your losses and run for the hills - no one thinks you're Einstein for running a test that loses a bucket of money.

In this instance, I wanted a CPC of less than $3, I wanted a CPM under $4 and at the very least, ONE conversion for this lead gen client. If it could do this, it meant it was competitive across each layer of the funnel and might get called up to first grade.

How surprised I was - Demand Gen had evolved, Like an AI film on Netflix, the machine was learning - and it was learning QUICK.

I ran a handful of lead-generation campaigns through Demand Gen, each surpassing my targets for impressions and clicks. I didn’t make projections on conversions because I didn’t believe it would happen - but it did. My CPA on that first Demand Gen campaign was only 20% higher than search, and my search CPA was bloody good (if I do say so myself).

The one thing that I cannot stress enough about these campaigns is this: Good data in, good data out.

If you’re using poor quality conversions or no conversions at all (you crazy thing you), the campaigns are going to be shit. This is where you have to meet AI halfway. Give it a north star that's valuable to you.

The one thing that I cannot stress enough about these campaigns is this: Good data in, good data out.

If you’re using poor quality conversions or no conversions at all (you crazy thing you), the campaigns are going to be shit. This is where you have to meet AI halfway. Give it a north star that's valuable to you.

These strong CPAs, CPCs and impressive awareness metrics were consistent across all of my testing, and checked out in GA4 - with one massive upside. I wasn’t limited by search quantities and battlefield-like search auctions with industry titans. I was more or less free and on my own to do as I wished.  Search is intent based and entirely based on the user taking action. Demand Gen on the other hand was an interceptive ad unit that leveraged behavioural data without them having to take action to trigger it.

So Demand Gen had evolved and was doing what Google had said it could do all along. It might not have launched perfectly, but it feels like Google knew the potential this campaign type had, even if so-called experts like myself didn’t necessarily agree at the time. Which led me to a key takeaway: Google had a clear vision and they were dissuaded by delayed adoption. Respect.

Even as I sit here writing about my experience with Demand Gen, I glance over to my second monitor and see Demand Gen up and running for a lead generation client - effortlessly driving a mass of impressions, clicks and the cherry on top, lead form submissions at a CPA only 3% higher than my generic search campaigns. I would add here, the lead quality isn’t quite on par, but I expected that given its not intent based, its behavioural. Regardless, I was impressed. Well played Google, well played.

While this is all well and good, it does raise a question about the elephant in the room.

How?

Now unless you’ve been living under a rock, you’d know that almost every ad platform in 2025 offers some form of AI-backed solution. All of which seems to sit behind a veil, classified as “proprietary tech”. As advertisers, when we use these kinds of tools, we very rarely get a glance under that veil. Which, at first, really rubbed me the wrong way. How can I optimise something if I don’t even know what it's actually doing to get the result?

Which I suppose raises an even bigger question, one I’ve been asking myself a lot lately and the inception for this article… Does it really matter how these campaigns work?

While that sounds dismissive, bear with me. I firmly believe every digital marketer should have some idea for how these campaigns work, how to optimise them and how to have clarity on when to use them and when not too. But do you really, deeply, need to understand every single possible placement and audience combination for your new Demand Gen campaign? Probably not.

People have been pumping millions of dollars into YouTube for years without checking so much as a placement report so kicking up a stink about these campaigns now feels a bit how ya goin.

It sheds light on one of the most overlooked qualities of a good marketer. Focus. I don’t mean to say that I’m more focused than you and as a result I’m better. I simply mean that too many marketers focus on the wrong things. It's not about working harder or drinking a bucket of lion's mane mushroom extract to get into a flow state. It’s about understanding on what plane you need to be thinking, on what matters and what doesn’t. Focus is everything.

It's a paradigm shift for marketers - where the requirement to spend hours and hours in-platform optimising campaigns is decreasing. But before you rip me in half for devaluing digital ads specialists, let me be clear: this is a good thing. It gives marketers time back to do what they should be getting paid for - understanding the role of each channel, analysing the results and determining what they mean for the overall business objective, not just comparing them to last week's numbers and handing back a vague collection of numbers that really mean nothing to C-suiters or business owners.

Products like Demand Gen don’t just fill gaps between typical campaign objectives like awareness and conversion - they give marketers time back so they can focus on what's important: business growth.

AI is scary; one week we’re warned it's coming for our jobs; the next, we’re using it to decline a meeting that we would rather avoid. But by embracing AI backed solutions, like Demand Gen, we can zoom out from the micro monotony and become dictators of scale.