The short version
An audience view for creators who have never connected an account
Streamforge's audience analysis is AI-based and does not require the creator to authenticate through OAuth. The model was trained on first-party audience data, and in Streamforge's tests its results are extremely close to the platforms' direct metrics. It reads the comments and profiles of the people around a creator and returns demographics, psychographics, interests, authenticity and community health.
What goes into the analysis
Audience evidence
Comment data and available profile data from the people who follow or interact with a creator, including the language they write in and the interests they show.
Creator and platform context
Creator identity, content, platform behavior, and cross-platform connections help interpret the audience evidence in the right context.
Audience outputs
Estimates for demographics, psychographics, interests, community health, loyalty, cohesion, toxicity, and authenticity.
From comments and profiles to an audience view
- 01
Collect the audience evidence
Gather comment data and available profile data from the people who follow or interact with the creator.
- 02
Separate real engagement from noise
Streamforge analyzes the comments on a creator's posts, videos and streams using AI to detect bot behavior, writing styles, and repetitive phrases or comments that do not engage with the content.
- 03
Model the audience
A model trained on first-party audience data turns that evidence into demographic and psychographic estimates, interests and community traits.
- 04
Show it beside the creator
The audience view sits on the creator's profile beside their content and performance, so fit is judged in one place. In the dashboard it is generated on demand, when a profile is opened.
Why it matters that no creator has to log in
There are two ways to learn about a creator's audience. One is to ask the creator to connect their account and read the analytics the platform gives them. The other is to work it out from what the audience does in public. Connected accounts give exact figures, but only for creators who agreed to connect, which usually means creators you are already talking to.
That is the wrong way round for discovery. Audience data exists to help you decide who to contact in the first place, across hundreds of candidates who have never heard of you. Streamforge's audience analysis is AI-based and does not require the creator to authenticate through OAuth, so it works on a creator you found a minute ago the same way it works on one you have partnered with for years.
It also means the method is not limited to creators who opted in to a network or a marketplace. Audience analysis is supported across YouTube, Twitch, TikTok, Instagram, and X.
How the model was trained and tested
The model was trained on first-party audience data, so it learned from known answers about real audiences instead of assumptions about who watches what.
It was then tested the same way. Streamforge has benchmarked its audience-analysis technique against first-party audience data from the platforms and found it to be highly accurate, and in Streamforge's tests its results are extremely close to the platforms' direct metrics.
Buyers have tested it too. Fortune 500 companies have tested Streamforge's creator data side by side against other vendors. Streamforge came out ahead in every completed evaluation.
What an audience report tells you
Demographics describe who the audience is: the countries they are in, their age groups and the gender split. Psychographics and interests describe what they care about and what else they follow. Authenticity and community health describe how real and how engaged the people behind the numbers are.
Henry + Baloo
@henrythecoloradodog2.3M followers
Audience demographics
AI-enriched audience analysis for this creator.
Core Demographics
Gender
- Female57.0%
- Male33.0%
- Non-Binary10.0%
Age Ranges
18-24
25-34
35-44
45-54
55-64
65+
Countries
- United States83.3%
- Canada5.6%
- United Kingdom3.3%
- Brazil3.3%
Use them in that order. Start with the hard requirements, such as the countries a campaign can serve, to decide who is in scope. Use age, gender and interests to rank the creators that pass. Then check authenticity and community health before you commit, so a strong-looking audience is also a real one.
A creator with very little audience activity gives the model less to read. When there is not enough evidence to support a field, the field should read not enough data rather than show a value with nothing behind it.
How to test it on creators you already know
The quickest way to judge audience data is to check it against something you already know. Pick five to ten creators you have worked with, ideally ones who shared their own analytics. Look each one up and compare the audience view with what you saw: the countries, the age and gender split, the interests.
Then run it the other way. Describe the audience you need and see whether the creators that come back are ones you would have picked, along with ones you had not found.
After a campaign, close the loop. Compare the audience view you recorded beforehand with where the traffic and the customers came from. If the creator shown with the most US audience sent the most US traffic, the data is doing its job.
We will run this with you. Bring the list to a call and we will go through it creator by creator.
