Analyzing YouTube Channel Growth With Python

"Our subscriber count is going up" is not analysis, it's an observation. Actual growth analysis means understanding what's driving the trend — which videos, which formats, which upload timing — well enough to make a repeatable decision about what to do next, rather than hoping whatever worked last time works again by accident. Setting up a growth dataset You need two things at minimum: a channel's subscriber count tracked over time, and video-level metadata (upload date, view count, video type) for the same period, so you can correlate specific uploads with subsequent growth. def get_channel_info(channel_id, token): return requests.get( "https://ensembledata.com/apis/youtube/channel/detailed-info", params={"channel_id": channel_id, "token": token} ).json()["data"] def get_channel_videos(channel_id, token): return requests.get( "https://ensembledata.com/apis/youtube/channel/videos", params={"channel_id": channel_id, "token": token} ).json()["data"] Poll get_channel_info weekly to build a subscriber-count time series, and pull the video list periodically to catch new uploads with their metadata. EnsembleData's YouTube endpoints, including channel-level detail and video listing fields, are covered in the API documentation. Finding what correlates with growth Once you've got weekly subscriber deltas and a list of what was published each week, a simple correlation pass — plotting subscriber gain against video count, video type (Shorts vs. long-form vs. livestream), and upload frequency — often surfaces patterns that aren't obvious from watching the numbers day to day. This doesn't require anything more sophisticated than a spreadsheet or a basic pandas groupby to start. import pandas as pd def weekly_growth_by_format(uploads_df, subs_df): uploads_df["week"] = pd.to_datetime(uploads_df["published"]).dt.isocalendar().week format_counts = uploads_df.groupby(["week", "format"]).size().unstack(fill_value=0) return format_counts.join(subs_df.set_index("week")["sub_delta"]) Being careful about causation A week with three uploads and strong growth doesn't prove upload frequency drove the growth — it might have been one standout video, an external mention, or an algorithm-driven push unrelated to your own actions. Look for patterns that hold across multiple instances, not conclusions drawn from a single good week. Channel growth analysis is genuinely useful for spotting what's correlated with growth over time, but treat single data points as hypotheses to test, not proven strategy. Where this pays off The real value shows up over months, not days — a channel with six months of tracked data and correlated upload patterns is in a much stronger position to make deliberate content decisions than one relying on gut feeling about "what seems to be working lately."