Competitor video intelligence as a monthly retainer

Marketing teams watch two or three competitor videos a quarter and call it competitive research. Their competitors publish forty. The gap is not analysis, it is reading — nobody has eleven hours a month to watch rival webinars, and so a positioning change goes unnoticed for two quarters. Transcripts close that gap: six channels of video become a few hundred thousand words of searchable text for under a dollar, and the month-on-month difference is a report someone will pay a retainer for. This playbook covers what the report contains, what it costs to produce, and why the hard part is selling it rather than making it.

What you are actually selling

Attention, on a schedule. The client is not buying transcripts and would not know what to do with them. They are buying a four-page document, on the first Monday of every month, that says what their competitors are now telling the market and what they have quietly stopped saying.

Three findings carry the whole product:

• New themes. Language that appears this month and did not appear before. A competitor mentioning “SOC 2” in five videos when they mentioned it in none is a go-to-market change, and it is visible in text before it reaches their homepage.
• Abandoned claims. A phrase that ran through twenty videos and has now disappeared. This is the finding clients react to hardest, because it usually means a product line is being retired or a claim did not survive legal review.
• Named objections. What competitors say about your client, by name or by description. Sales teams use these directly, which is what makes the retainer survive a budget review.

The month-on-month comparison is the product. A one-off snapshot of what competitors talk about is mildly interesting; the delta is what a marketing lead forwards to their VP.

The numbers

The data cost is trivial and the model cost is small. Almost everything you charge is for judgement and for the fact that the report arrives without being chased.

The 99.4% figure is real but misleading on its own, which is why the labour line is in the table. Six hours is the steady state after three or four months, once you know each channel’s rhythm. Report one takes two days.

The month-one backfill is a separate cost and worth pricing separately. Two years of history across six channels is roughly 1,000 transcripts, or $5.00 of input, and it is what makes the first report contain a trend rather than a snapshot. Bill it as a $1,500 setup fee; the client sees an artefact for it.

Three clients at $1,200 is $3,600 a month against about $20 of direct cost and roughly two days of work. That is the realistic shape of this business, and the constraint on it is client count, not capacity.

How it works

1. Agree the competitor set with the client, in writing. Six to eight channels. Let them nominate the list — their view of who they compete with is itself information, and it stops the first report being argued with.
2. Backfill each channel. Run channel-wide extraction with a cap that covers about two years of uploads. Keep the raw output untouched as your archive of record; every later comparison is against it.
3. Store one record per video, keyed by video ID. Title, channel, publication date, duration, view count, full text. A directory of files or a SQLite table both work; do not build a pipeline for six channels.
4. Build the baseline vocabulary. Extract the terms, product names and claims that appear across the backfill, with counts per channel per quarter. This is the reference the monthly delta is measured against.
5. Run monthly, newest-first, capped to the last thirty days. Compare the new terms against the baseline. Anything appearing three or more times that was absent before is a candidate finding.
6. Read the candidates yourself. This is the step that cannot be automated and the reason the report is worth money. Most candidates are noise — a guest speaker’s vocabulary, a new video editor’s intro script. You are looking for the four that are not.
7. Write four pages. What is new, what stopped, what they said about the client, and one paragraph of what you would do about it. Same structure every month, so the client can compare across months without re-reading.
8. Send it on the same day each month. Predictability is a large part of what a retainer buys.

Deliver as a PDF or a document, never as a dashboard. Dashboards get logged into twice and then never again; a document gets forwarded, and forwarding is how the retainer gets renewed.

Where it gets hard

Finding clients. The work is a day of setup and six hours a month; getting three companies to pay for it is the business. Marketing teams have a budget for tools and a budget for agencies, and a $1,200 monthly report fits awkwardly into neither.

What works is the free first report. Pick a company, produce the report about their actual competitors, send it unsolicited with the finding at the top. Your production cost is under $10, so you can afford to do this twenty times. Roughly one in ten converts, which makes the cost of a client about $100 in data and two days of your time.

Four other frictions worth knowing before you start.

Some competitors barely publish. A channel with four videos a year produces no delta, and you will have months with nothing to report. Say this at the start of the engagement, and use the quiet months to go deeper on one competitor rather than padding the document.

The client will ask why you did not catch something. They will see a competitor’s LinkedIn post or press release you did not report, because it was never on video. Set the boundary in the first meeting: this covers what they say on video, which is where the long-form claims live, and it is not full media monitoring.

Month seven is when retainers die. By then the novelty is gone and the finding rate has settled. The counter is a quarterly rollup that shows the year’s trajectory — the thing no single month can show, and the reason to have kept paying.

You are competing with free curiosity. Any marketing manager can watch a competitor video. Your case is not that you can watch video; it is that they will not, month after month, and their competitors’ positioning shifts anyway.

Limitations

• Videos without published captions return nothing. Most business channels caption automatically and are fine. A competitor who uploads without captions is invisible to this method, and no tool can retrieve a transcript that was never published.
• Transcripts lose everything on the slides. Pricing tables, product screenshots, roadmap graphics and comparison charts are frequently the most informative part of a competitor webinar, and none of it is in the text. “As you can see on this slide” marks a hole in your record.
• Text tells you what was said, not what worked. You cannot infer from a transcript whether a message performed. View counts are a weak proxy at best, and they are confounded by promotion spend you cannot see.
• Word frequency invents patterns. A term appearing four times more often this month is usually a scheduling accident or a single long video, not a strategy. Every candidate finding needs a human to check the source before it goes in the report.
• No speaker separation. Webinars with guests, panels and interviews come back as one undivided stream, so you cannot reliably attribute a claim to the company rather than to a guest.
• Channel runs are newest-first and capped. The cap is a sampling decision. State it in the report’s method note, or your trend lines will eventually mislead you.
• Video is a lagging indicator. By the time a positioning change reaches a produced video it has usually been live on the website for weeks. This method catches what page-monitoring misses, not what page-monitoring catches sooner.
• You cannot republish the transcripts. The videos are copyrighted works. Quote briefly with attribution and a link; do not hand the client a transcript archive as a deliverable.

FAQ

How many competitor channels can one report cover?

Six to eight is the working range. Below four there is rarely enough signal for a monthly cadence. Above ten the reading time grows past what six hours supports, and the report becomes a list rather than an argument.

What do I charge for the first month?

A setup fee of around $1,500 for the backfill, then the monthly rate. The backfill is genuinely more work — it is where you build the baseline vocabulary — and charging for it separately stops the first month looking like poor value against the ninth.

Do I need permission from the competitors?

No, and you should not ask. You are reading captions that were published publicly alongside public videos. What you must not do is redistribute the transcripts themselves; quote, attribute, link, and keep the corpus private.

Can I sell the same report to two competing clients?

Do not. The two reports would overlap heavily, both clients would eventually find out, and the credibility loss ends both engagements. One client per competitive set, and say so in the proposal — exclusivity is a reason to pay more.

Is this not something an agency already does?

Some do, inside a broader retainer, and at a price that reflects the broader retainer. The niche here is the narrow, cheap, punctual version: one thing, done monthly, at a price a marketing manager can approve without a procurement cycle.

What if a competitor stops publishing entirely?

That is a finding, and often a substantial one. A channel that published fortnightly for two years and then goes silent for a quarter usually means a team change or a funding problem. Report the silence.

How long before this pays for itself?

The direct costs are recovered by the first invoice. The real investment is the twenty free reports it takes to find three paying clients, which is about two weeks of work spread over a couple of months.