What advertising creatives work for technical people?
At BlueDot Impact, we run educational courses that help people develop knowledge, skills and connections to pursue high-impact careers. To find great people for these courses, we’ve been experimenting with paid advertising.
This article explores what creatives worked best in our paid ads for our AI Alignment (June 2024) course. The course focuses on helping people get into technical AI safety research to reduce risk from advanced AI.
While we accept people from a range of backgrounds, our most typical participant:
- is a working professional, usually with several years of experience
- has a background in computer science, machine learning or software engineering
- lives in a developed English-speaking country, often the US or UK
Given this is our most identifiable audience, we focused most of our efforts here. We also advertised a little to ML academics, and some specific AI organisations.[1] We previously advertised to university students but chose not to this round.[2]
Ad creatives
This will be a selection of some of the ads we ran, and what their average cost per click was. There were some minor differences in how we chose to deliver and target certain ads, and there is some randomness in the data. Therefore, you shouldn’t take these figures as exactly what you’d get, or very strong evidence a particular ad is much better than one with a similar cost per click. But where there are significant differences we can probably say one ad is better than another.
This data is solely from LinkedIn, as it had the best analytics capabilities out of the platforms we used and it’s where we spent the most so had more data generally. We’ve also filtered the data down to only those ads which got at least 10 clicks, and to the more basic audiences (people with machine learning skills or technical skills and an AI interest).
We use cost per click as a measure of ad performance. This is a lossy proxy for what we actually want: great people who will actually apply. However, it strikes a balance between relevance to our goal, while being high volume enough so we can make meaningful comparisons. We use cost per click (rather than click-through rate, or total clicks, etc.) because we only pay when someone clicks an ad: so this best represents how effective our ad spend is on different ads.
Lower is better for cost per click (as this means we’re paying less for someone to click the ad, and come to our website).
High performer: Ads with technical references
$1.31 per click
$1.35 per click
$1.37 per click
$1.49 per click. We also had a much higher engagement rate on this ad, people commenting and discussing the ‘puzzle’. This likely boosted its organic reach, although we didn’t verify this.
Moderate performance: Calling out job title
$1.62 per click
Mixed performance: AI risks or malfunctions
These three ads performed well for more general tech professionals with an interest in AI, but comparatively quite poorly for machine learning specific professionals.
$1.36 per click for tech professionals with AI interest
$1.63 per click for ML professionals
$1.38 per click for tech professionals with AI interest
$1.87 per click for ML professionals
$1.46 per click for tech professionals with AI interest
$1.84 per click for ML professionals
Mixed performance: AI generated images
$1.57 per click
$2.03 per click. We debated using this one: it’s really attention grabbing and certainly not something you’d usually see on LinkedIn. But some team members thought it might scare people away. In the end, it performed a lot worse than the slightly more friendly image above.
$1.97 per click. Similar considerations to the scarier image above.
Poor performance: ControlNet images
We thought these ControlNet images would be eye catching, and an interesting use of AI that might also result in higher engagement. Unfortunately, these didn’t work.
$1.84 per click
$1.92 per click
Poor performance: Original community focused ads
This was the best performing ad that we ran for our March 2024 course (where paid marketing was generally underwhelming - we discussed it in our retro on this). We ran this ad alongside our new ads ones as a benchmark, which allowed us to test whether we had actually made our ads better, or something had changed about ad serving since our last marketing campaign.
Its relatively poor performance helped confirm that our new ad concepts above were better.
$2.14 per click
Other areas for improvement
Compared to the last round, we spent a lot more time developing and testing new ad creatives. However, we’ve sent comparatively little time improving the text of the ads themselves. We think this seems like the place to start testing out different ideas.
Also see
If you enjoyed this article, you might also want to see our articles about:
Footnotes
All the platform’s targeting abilities are far from perfect, so implementations are fuzzy. LinkedIn was most specific, where we created 6 audiences:
- Machine learning professionals
- Software engineers, with some machine learning skills
- Tech professionals (software engineers, data scientists, technical product managers etc.) interested in AI, and open to switching jobs
- Machine learning academics
- Specific AI organisations, for example people already in frontier AI companies
- Retargeting, i.e. people who had already engaged with us on LinkedIn previously
University students are much cheaper to market to, but tend to be much lower quality participants than working professionals.
Undergraduates have much worse attendance, submit poorer quality projects, and are less likely to go on to working in the field. They’re also more likely to have alternative opportunities to learn about our topic areas, for example through university groups and societies.
Of course there are exceptions, and we are keen for great students to apply! But previously we’ve seen that the low quality of student applications is more significant than the reduced cost of marketing to them, so decided not to focus on them this round.