Ad platforms are handing more of the targeting work to AI, but not at the same speed. Here's what each one still lets you choose, and how we build audiences across all of them.
Ad platforms now differ sharply in how much targeting they let advertisers control. Meta relies mostly on its machine-learning system, Andromeda, and on the creative itself to find new customers, and its hand-picked interest targeting has shrunk. TikTok still lets advertisers target specific interests, creators and hashtags, and that targeting updates with what's trending. YouTube lets advertisers place ads on specific channels and videos.
Here's how Meta, TikTok and YouTube compare on audience targeting.
| Platform | Can you target people who like a specific brand, creator or niche? | Strongest at |
|---|---|---|
| Meta | Only established interests; the algorithm finds the rest | Reaching a brand's known audience; finding new high-intent customers; broad age reach |
| TikTok | Many, by name and hashtag; not every option is available | Reaching new audiences through creators, trends and lifestyle interests |
| YouTube | Yes, through their channels and videos | Discovery and channel growth |
For current ad costs on each platform, see Gupta Media's Social Media CPM Tracker and our report on the true cost of social media advertising.
Meta is still the best platform for reaching a brand's known audience: the people who already follow, watch or engage with it on Facebook and Instagram. What has changed is how you reach people who like similar brands.
The limitation: Meta's interest list hasn't kept pace. Many newer brands, creators and cultural niches were never added, and Meta has recently scaled interest targeting back even further. That leaves fewer specific interests to choose from.
Why: Instead of updating that list, Meta is pushing advertisers toward automation, led by its machine-learning system, Andromeda. Andromeda decides which ads are in the running for each person. It reads that person's real-time behavior and matches it to what each ad actually shows and sounds like.
In practice: If someone regularly watches and engages with content from brands in a particular category, Andromeda reads that behavior. It can then serve them a new brand whose ad looks and sounds like a fit, without the advertiser naming any competitors. On Meta, the creative is doing much of the targeting.
The trade-off: You can assume Meta is reaching people who like similar brands, but it won't tell you. Reporting on which affinity groups saw or engaged with the ads is limited, so that part of the strategy runs on trust in the algorithm.
TikTok is the best platform for reaching new audiences by naming the creators, interests and trends they follow. Its targeting updates with what's popular right now, so many newer niches are targetable here when they aren't on Meta.
Engagement is high, which makes TikTok strong for early awareness. The audience skews younger, which can limit reach unless the brand's customers skew young too. For current pricing, see our guide to what TikTok ads cost.
YouTube is the platform where you can choose exactly where an ad appears. Short-form Shorts lead viewers to full-length videos on a brand's channel, turning casual viewers into subscribers.
Channel targeting is the closest substitute for the specific-interest targeting Meta has cut back. YouTube is also the right home for campaigns aimed at growing a brand's own channel.
| Option | What it's good for |
|---|---|
| Strong niche communities, organized by topic, brand and interest. | |
| Twitch | Highly engaged fans of streamers and games, targetable down to individual creators. |
| A high-intent audience, especially strong for women interested in lifestyle topics. | |
| Amazon DSP | Full-screen, sound-on ads on streaming TV, targeted by shopping and viewing behavior. |
| Geotargeting | Audiences built from anonymized mobile device IDs, to reach people who visited specific places or events. |
Yes, but the options are narrower than they used to be. Meta has scaled back its interest list, and many newer brands and niches can't be selected. Meta now leans on its Andromeda system and on ad creative to find people with those interests.
Andromeda is Meta's machine-learning system for choosing which ads are eligible to show each person. It matches a person's real-time behavior to what each ad shows and sounds like. That means the creative itself acts as a targeting signal.
TikTok and YouTube. TikTok lets advertisers target creators, interests and hashtags that update with trends. YouTube lets advertisers place ads on specific channels and videos.
Channel targeting places ads on specific YouTube channels or videos, so they reach people while they're watching that content. It's the closest substitute for the specific-interest targeting Meta has cut back.
Start with your existing audience (followers, engagers and viewers), then lookalikes of that audience, then interest targeting. This order works on every major platform; what varies is how much of each step the platform lets you control.
Only at a high premium. Most TikTok retargeting uses people who engaged with a brand's previous TikTok ads, plus lookalikes built from them.
See how we've built audiences for brands like Fender and the Cleveland Cavaliers in our case studies.
Want updates on the latest changes in targeting? Subscribe to The Thread, Gupta Media's weekly newsletter on what's driving performance in digital marketing.