David Baekholm leads growth marketing at Epidemic Sound, a Swedish music technology company with global reach. He’s responsible for finding innovative and efficient ways to grow the company.
Epidemic Sound is a leading provider of royalty-free music and our mission is to soundtrack the world. But how do you find the perfect audience for over 50,000 unique music tracks and 200,000 sound effects? This was the strategic question at the heart of our growth plans, especially with an ambitious 50% growth target for 2025.
Our challenge was that our bandwidth was finite. For example, manually managing search campaigns to connect a specific track with a creator looking for “music for a product reveal” was becoming an impossible task.
We had to find a way to handle this complexity at scal…
David Baekholm leads growth marketing at Epidemic Sound, a Swedish music technology company with global reach. He’s responsible for finding innovative and efficient ways to grow the company.
Epidemic Sound is a leading provider of royalty-free music and our mission is to soundtrack the world. But how do you find the perfect audience for over 50,000 unique music tracks and 200,000 sound effects? This was the strategic question at the heart of our growth plans, especially with an ambitious 50% growth target for 2025.
Our challenge was that our bandwidth was finite. For example, manually managing search campaigns to connect a specific track with a creator looking for “music for a product reveal” was becoming an impossible task.
We had to find a way to handle this complexity at scale to hit our next phase of growth. That’s when we turned our attention to AI.
From shiny new thing to a global rollout in 90 days
On top of my role as head of growth marketing, I also lead an internal company initiative: AIO (AI Operations). Our goal is to champion integrating AI tools into our day-to-day operations. A big part of that is setting realistic expectations about what AI can do — and what it can’t. While many people think AI is always “plug and play”, I understand that the biggest rewards often require a bit of heavy lifting.
In the past, we have experimented a lot with broad match, which, I will admit, we were a little hesitant about at first. But we’ve always believed in testing new things, and with a bit of work, we found our rhythm. AI-powered tools like broad match, Performance Max, and Demand Gen are all now integrated into our marketing processes.
My team and I love new tech and we understand the first-mover advantage. So when I attended Google Marketing Live in May and heard about AI Max for Search, I very much thought: “I want that, and I want it tomorrow.”
In a nutshell, it’s a new suite of features in Google Search that uses AI to generate and optimise ad copy, find new audiences, and bid more effectively. It does this by understanding people’s intent behind a search, not just what they type.
Some of the new features that excited me most included:
- Finds new audiences: Find highly relevant search queries you might otherwise miss. Matching your ads to what users really mean, even in long, conversational searches.
- Creates dynamic ad copy: Uses generative AI to create headlines, descriptions, and calls to action that match user intent. Analyses your website and existing ads to ensure your message is always relevant and effective.
- Directs to the best landing page: Automatically sends users to the most relevant page on your website for their specific search. This creates a smoother experience and helps drive better results.
- Offers more control: Fine-tune your campaigns by looking at specific locations, manage brand suitability by including or excluding certain brands, and review and block any AI-generated ad assets that don’t fit your brand.
- Provides deeper insights: More detailed reports on what’s working. The search terms report shows you how AI is finding new customers. And other reports on keywords, landing pages, and assets give you a full picture of your ad’s performance.
Listening to the product reveal, it seemed to solve our core challenge: scaling globally while staying locally relevant, without micromanaging every single campaign.
Within one week of Google Marketing Live, we began planning our first test. In June, we launched a pilot in the U.S. and Australia — two very different markets. The U.S., because it’s a big market for us, we have a lot of data there, and can get quick learnings. Australia, because the user behaviour is different to the U.S., with a high percentage of users interacting today but not signing up until a few weeks later.
We wanted to see how it would perform in markets that had these different user flows and also different seasonal patterns. And the results were almost immediate. In both the U.S. and Australia we found new efficient traffic sources driving customers at on average lower cost per sign-up.
After just two weeks, we were confident in a global rollout. By mid-July, we had fully shifted all our campaigns over to AI Max, rolling out around ten new markets every day.
10-15% conversion boost with zero added resources
In many markets, we saw a 10-15% increase in subscriptions, with some markets seeing even higher lifts. The upsides were especially significant in countries like Indonesia, where our manual campaigns had traditionally been more simplistic due to language constraints. AI Max found the right users and saved us a lot of time compared to an old-school translation agency-type campaign.
Our blended cost per subscription went down, and the team was freed from the heavy lifting of manual optimisation.
What was most exciting, though, was the data. We started realising there’s beauty in the machinery of actually predicting user intent.
While the team had been manually trying to reach creators searching for “music for a product reveal”, AI Max uncovered counterintuitive queries that challenged our assumptions. We found people were signing up from search terms we wouldn’t have considered relevant and landing on pages we never expected.
That’s because it goes beyond the keywords you’ve manually selected to find other unique but relevant queries.
For instance, a user might search for “how do you know if a song is drm protected”. In the past we would not have chosen this phrase to bid on. But AI Max looks at the user’s intent, serves the perfect ad with copy generated in real time, and points the user at the best page for them.
Within the search terms report we’re then able to see all these connections, and more. The team is able to analyse the data flow between search terms, headlines, and URLs, providing a comprehensive view of customer ad journeys.
We’re also able to see AI-created search terms and their origins. This is ideal when we know users have started to use AI overviews more within search.
This meant we no longer needed to bid on the same keywords as our competitors, but had new insights into untapped searches.
And it showed us AI Max wasn’t just a bottom-funnel tool. It was a way to gain valuable insights that could inform our content strategy and audience building beyond paid search.
All helping us deliver our ambitious growth plans, with finite resources.
Key takeaways
- Embrace AI-powered tools: Be open to testing and integrating new products. AI Max for Search freed the team from manual optimisation and provided deeper insights into unexpected but effective search terms and landing pages, informing their content and audience strategies beyond paid search.
- Prioritise understanding user intent at scale: AI Max for Search helped the team by understanding the intent behind searches, not just keywords. This allows for finding new, relevant audiences and optimising ad copy dynamically.
- Test and roll out strategically: Plan and launch a pilot to gather learnings before a global rollout. This helped the team confirm the effectiveness of AI tools and enabled scaled implementation to be efficient.