Competitor Video Research With Summaries

Competitive research on video is usually done badly, because the honest version is expensive. People watch three videos, form an impression, and call it analysis.

Summaries change the economics enough to do it properly — and properly means looking at twenty videos rather than three.

What summaries reveal that watching does not

Patterns across a catalogue are invisible while watching, because by video six you have forgotten what video one covered.

Read twenty chapter maps side by side and the structure of a channel becomes obvious: the topics they return to, the ones they avoid, the order they build an argument in, and the claims they repeat across every video.

That last one is particularly useful. Every channel has three or four positions they restate constantly. Those are their actual thesis, and they are much easier to spot in text than in forty hours of footage.

Picking which videos to include

The selection shapes the conclusion, so it is worth being deliberate rather than taking whatever the search results offer.

Take each channel's most-viewed and most-recent videos, which answer different questions: what worked for them historically, and where they are heading now. A channel's direction is often visible in the recent set before it is visible anywhere else.

Avoid sampling only the popular ones. Those tell you what an audience rewarded, which is useful, and they systematically exclude the experiments — and the experiments are where you learn what a competitor is trying to become.

The gap is the output

The point of this exercise is not to know what competitors said. It is to find what nobody said.

Summarize the top videos on a topic across several channels and the shared skeleton emerges quickly — everyone covers roughly the same points in roughly the same order, because everyone researched the same way.

What is missing from all of them is the interesting brief. That judgement requires seeing them together, which is exactly what watching sequentially prevents.

What it cannot tell you

This is worth being clear about, because it is easy to over-read a set of summaries.

Why something performed. Delivery, editing, thumbnail, timing and audience relationship do not appear in a transcript. A video that outperformed yours may have covered less and simply been better made.

Tone and personality. The thing audiences actually follow a creator for is almost entirely absent from a summary, which is a reason not to conclude that a competitor's content is thin just because their summary is.

Production quality. Invisible. Two channels with identical summaries can be very different experiences.

A practical process

1. Pick fifteen to twenty videos across three or four channels — their most-viewed plus their most recent.
2. Summarize each at chapter-map depth and export as you go.
3. Read the maps together, in one sitting, looking for repeated structure rather than individual points.
4. Write down the shared skeleton — the points everyone makes.
5. List what is absent. That list is the research output.

On a five-a-day free tier that is three or four days of work spread over a week, which is still dramatically cheaper than watching forty hours.

Where the line is

Researching what exists is normal practice. Using a summary to remake someone's video is not, and it produces worse work anyway — you get the claims without the reasoning that produced them.

There is a specific risk worth naming: a summary drops hedges. If a competitor carefully qualified a claim and the summary did not, building on that summary means publishing the unqualified version under your own name.

So treat summaries as a map of the territory rather than as source material. Anything you intend to actually use, open the timestamp and hear it said — the argument we make in why every summary should carry timestamps.

When their videos have no captions

You will hit this occasionally and it is worth knowing why rather than assuming the tool failed.

YouTube's documentation on automatic captions notes they depend on audio quality, accent and background noise, and that some videos simply do not get them — heavily music-led content and very recent uploads especially.

On an established channel it is uncommon, since captions help their own reach. When it happens, work around the gap rather than treating the catalogue as unanalysable.

Reading summaries in sets, not singly

One more point about method, because it decides whether this exercise works.

Reading twenty summaries one at a time, as you generate them, produces twenty impressions and no analysis. The patterns only appear when you read them together, in one sitting, after they are all done.

So generate first, export everything, and do the reading as a separate task. It feels less efficient and is the only version that produces an insight rather than a vague sense of what competitors are up to.

Tracking a channel over time

The version of this that compounds: summarize competitors' new uploads as they appear, rather than auditing a catalogue once.

Five minutes a week builds a text record of what an entire space is talking about, searchable, going back years. Six months in, you can answer "when did everyone start saying this?" — which is not answerable any other way.

Doing this on yourself

The exercise is more uncomfortable and more useful pointed inward.

Summarize your own last twenty videos and read the maps together. You will find repetition you did not know about, a thesis you state constantly without noticing, and topics you have circled without ever addressing directly.

Most creators have never seen their own output as a set, because watching your own back catalogue is unbearable. Reading it is not, and the patterns are as visible in your work as in anyone else's.

Frequently asked questions