The Business Plan

01 · Summary

The business in one page

The business. A YouTube series that tests one AI or productivity tool against one real task and reports what happened, failures included. It is meant to earn from YouTube ads first and from clearly labelled sponsorships second. After costs, taxes and my own pay, what is left goes toward hardware testing for Johnny Autoseed, my company working on autonomous tools for food and farming, and later toward land.

Channel snapshot of October 19, 2026, from vidIQ. measured

What the numbers say

1. Sponsorship is a later income, not the engine. Active peer channels with 10,000 to 100,000 subscribers average about 6,100 views per video measured. This channel averages 165. Sponsors who pay by the view need audiences closer to the first number.
2. At this size, ads are likely to earn more than sponsors. With 4 of every 20 videos sponsored, ads bring in 67% to 71% of revenue in every case I modelled estimate. So ads are the base, and sponsored videos stay at about 1 in 5.
3. Land is years away on any realistic path. Putting $1,000 a month into a land fund needs about $1,700 of monthly revenue if I take no pay, or about $3,100 if half goes to my pay estimate. So the plan starts with access to land at no capital, and buys later, if ever.

The plan in four stages

The windows are targets, not deadlines. target

02 · Where it stands

The starting point, measured

Every number in this section comes from vidIQ, pulled October 19 to 21, 2026. The channel is small, the library is unfocused, and the series has not started yet.

Average views is lifetime views divided by video count, Shorts included. One channel with a view count that does not fit its size was left out of the medians. measured

Four readings

1. No topic is working yet. The 25 videos range from radio episodes to sleep audio, a game trailer and AI commentary, with views from 7 to 1,368. The one faint signal is a practical, make-it-at-home Short at about five times the Shorts median. Worth one test, not a strategy.
2. Subscribers are not moving. One gained in 84 days measured. Ad revenue needs 1,000 subscribers sourced, and subscribers follow views.
3. Ad revenue is far off. Even if every viewer had watched every video to the end, the whole library would hold about 159 watch hours measured. YouTube asks for 4,000 inside 12 months sourced.
4. Posting daily is common here, so it is not an edge. 4 of 14 small active peers published 30 or more long videos in 30 days, and the busiest published 79 measured.

A few videos from the channel

From before the series. Each loads from YouTube only when you press play. View counts are from October 21, 2026.

Did Google just say ASI is coming soon?Long video · 295 views

You can print this dopamine fidget yourself at home or order it onlineShort · 1,368 views Claude coded this little farm sim and it’s actually good??Short · 339 views

03 · The product

One tool, one real task, an honest result

“I test one AI or productivity tool against one real task and tell you what happened, including the parts that failed.”

The viewer I have in mind is someone deciding which tool to use for their own work or life. That is a guess, and the first 90 days test it against comments and search terms. The existing library stays up, filed into its own playlists so it does not read as the channel’s identity.

Three formats

The Tool Test scorecard

Filled in during the test, not after, and shown on screen near the end of each video.

• Task: the real task in one sentence, written down before recording.
• Time without the tool, and with it: on a stopwatch, including setup and fixing its mistakes.
• Output quality: 1 unusable, 2 needs heavy rework, 3 usable with edits, 4 good, 5 better than I would do.
• Failures: each thing that broke or was wrong, with the timestamp.
• Cost: the price, and what the free tier stops you doing.
• Verdict: use it, use it if (with the condition), or skip it. Plus the version, plan and test date.

The anchor series: can AI manage my life better than I can?

A documented experiment with the rules fixed before filming. Episode 0 names four metrics, each from a source the AI does not control (tasks done against tasks planned, spending against budget, sleep, exercise), and a pass line. One example of a pass line: the AI weeks beat the baseline on at least 3 of the 4 metrics by at least 10%. Once published, the pass line does not move. It is one person over six weeks, not evidence about AI.

Cadence

Six videos went out in the last 84 days measured. Daily would be about 14 times that, so the plan ramps up in steps and only moves up when the step before has held.

Packaging

• Titles are the question a searcher types, such as “Is Notion AI worth paying for?”, checked against keyword research before recording.
• Every video gets a thumbnail. No thumbnail, no upload. 1280 by 720, three words or fewer, a readable face and the tool on screen, checked at 320 pixels wide, in the site’s colours.
• Every Tool Test gets a Short cut from the same session, for discovery.
• Thumbnail tests wait for traffic. They need a few thousand impressions to mean anything estimate.

YouTube renamed its “repetitious content” policy to “inauthentic content” in July 2025 and applies it to mass-produced uploads sourced. The protection is substance: real screen recordings of real tests, live commentary, varied structure, and no scripts that were generated and simply read out.

04 · The market

Crowded, so the angle has to be specific

AI tool reviews and productivity tutorials are a busy niche. In the peer sample, the median small active channel published 13.5 long videos in 30 days and the busiest published 79 measured. Volume alone does not stand out here.

Where this can be different

• Measured, not described. A stopwatch and a written task, so a verdict is a result and not an impression.
• Failures stay in. Most tool videos cut them. Here they are part of the score.
• Rules set in advance. The scorecard and the anchor series’ pass line are published before the test.
• A builder’s view. I use these tools to build a robotics company, a book and an album, so the tasks are real work.

Who watches is still a guess. The first 90 days answer it with search terms, comments and which tests hold attention, and the decision rules in section 11 say what changes if the answer is “nobody yet”.

05 · How it earns

Five revenue lanes, opening at different times

Left out on purpose: affiliate links, because a commission gives a reviewer a reason to prefer one verdict; coaching and consulting, which this site stopped offering in September 2026; and paid reviews.

Ads against sponsors

With 4 of every 20 long videos sponsored, ads bring in 67% to 71% of revenue across the low, base and high cases estimate. For sponsors to match ads, about 40% of all videos would need to be sponsored, and that would cost the trust the whole thing depends on. That is the numerical reason sponsorship is the upside and not the base.

What a views-priced sponsorship pays

Price per video at a given median of first-90-day views, at three sponsor rates per 1,000 views sourced. Under about $150 a video, the outreach and the contract are not worth the time, which is why this lane waits for about 5,000 views.

Views per video each income level needs

Profit per month before tax, with 20 long videos a month, 4 of them sponsored, about $250 a month of costs, and ad revenue earned on each video over its life estimate. Read it as “what would have to be true”, not a forecast.

For scale: active peers with 10,000 to 100,000 subscribers sit near 6,100 views per video measured. The base case needs 6,250 for $2,000 a month. Reaching that within a year would put this channel above the median of channels with 17 to 165 times its subscribers, so it is a stretch, and the targets in section 11 are built around smaller steps.

06 · Independence

How paid work and honest reviews fit together

The rules I am working to. They are intentions, not contract terms; any actual deal is governed by its own written agreement.

1. Payment buys a labelled placement, not a verdict. A sponsor can pay for a segment or for a video about its business. A tool’s score is not for sale.
2. I can decline to publish a sponsored video and refund the fee. The intent is that no honest review quietly disappears because it was negative.
3. No approval before publishing. A sponsor can flag a factual error in specifications. Opinion, order and verdict stay mine.
4. Paid, gifted and free things are labelled, in line with FTC guidance and YouTube’s paid promotion rules sourced.
5. Errors are corrected in public, with a dated note in the description and a pinned comment.

The disclosure checklist

For any video with a material connection, all five, before upload:

• Said out loud in the first 30 seconds, and again before the sponsored segment.
• On screen for the whole segment, large enough to read on a phone.
• The first line of the description.
• YouTube’s “includes paid promotion” box ticked.
• A pinned comment with the test conditions.

Conflicts of interest

I drafted this plan with Claude, which is made by Anthropic, and the channel is likely to review Claude. Weigh what this page says about AI tools accordingly.

Once there is revenue, the plan is a short monthly report: the number and categories of sponsors, total sponsor revenue, ad revenue, expenses and corrections. Individual sponsor prices can stay private.

07 · Where the money goes

The order the money moves in

Money is spent in this order. The percentages are defaults I have not settled yet, and the tax share needs a CPA’s check.

1. RevenueAds, sponsors, tips
2. ↓
3. Platform and payment fees
4. ↓
5. Tax reserve, about 30% of profitMoved to its own account the day money arrives estimate
6. ↓
7. Running costs, about $250 a monthTools, software, storage, small gear estimate
8. ↓
9. RunwayThree months of costs saved before anything else
10. ↓

11. Then what is left, split

    My payDefault: half

    Land fundDefault: half

Profit is taxed when it is earned, and land is not a deductible expense. So each $1 that reaches the land fund costs about $1.43 of pre-tax profit, or about $2.86 with the default split estimate.

Revenue needed to fund land

08 · Land and Johnny Autoseed

Access first, buying last

Johnny Autoseed needs somewhere to test FarmBot and other hardware. That does not require owning land. The plan gets access first, at little or no cost, and treats buying as a later decision with its own checks.

What land costs, for reference

Asking prices from Butte County listings seen in 2026, not sold prices, and not properties I am pursuing sourced. The middle row is a stand-in estimate. Cash needed is the down payment plus 5% for closing and a first-year reserve, plus a $40,000 placeholder for a well, power and access on the raw parcel estimate. A real utilities quote is the first thing to get on any raw parcel.

A 5% down payment is only possible through the federal loan below, and only if I qualify.

Financing

• Federal route. The USDA Farm Service Agency’s down payment loan can cover up to 45% of the price, with the buyer putting down about 5% and a commercial lender covering the rest sourced. Rates and caps change; FSA has the current ones.
• The catch. FSA’s farm ownership loans generally ask for at least three years of farm management experience, or other experience it accepts sourced. If that applies, the clock should start now, which is the main reason access comes first.
• Commercial route. Usually 20% to 30% down on rural land sourced.

Before any offer

Written answers on zoning and what it allows, a well test, a power line quote, legal access, wildfire risk and an insurance quote, and whether the county would permit workshops or visitors. Plus sold prices from a local broker and a lender’s pre-qualification.

Two photos from my Textures series, shot on a Canon Rebel T3. For illustration. Not a specific property.

09 · Operations

How a week runs

Batching is what makes five a week possible: recording two or three tests in one sitting costs far less than five setups.

Time per Tool Test

An estimate until ten videos are logged; then the measured median replaces it. estimate

Before every upload

• One target search query, written down
• The task written down before recording, with the date
• A real recording of a real run, failures kept in
• A thumbnail, checked at 320 pixels wide
• Title under 60 characters, the query in the first 40
• First line of the description: what was tested and the verdict
• Disclosure check done, or an honest “no one paid for this” line
• A Short cut and scheduled
• The tracking sheet row filled in, hours included

Guardrails

• A buffer. Five finished videos banked before going to five a week; seven before daily. Nothing goes out the day it is finished.
• The drop rule. Any week that starts with fewer than three banked videos drops to three long videos. Volume gets cut, not the thumbnail or the disclosure check.
• No announced streaks. State the rhythm, not a count.
• One day off a week.
• Separate money. The channel’s money lives in its own account from the first dollar.

The weekly review

Fridays, ten minutes: videos shipped against plan, median hours per video, videos banked, best and worst video at 7 days and why, median views at day 30 of the last ten long videos, subscribers, sponsor conversations, open corrections, money in and out, and one decision for next week.

10 · The first 90 days

October 26, 2026 to January 24, 2027

All dates and counts here are targets. target

11 · Targets and decision rules

How I will know if it is working

The targets are tests of whether the plan is working, not forecasts. “Day-30 views” means the median views at day 30 of the last ten long videos. target

If this happens, then

Decided in advance, so the answer is not argued in a bad week.

12 · Risks

What could go wrong, and the early warning

Likelihoods are my judgment, except where marked.

13 · Open questions

What I do not know yet

Not measured yet

• Watch hours in the last 12 months, impressions click-through rate, average percentage viewed and audience geography. These come from YouTube Studio and change several numbers above.
• Real hours per video, which replaces the estimate in section 9 after ten videos.

Not decided yet

• How the money left over is split between my pay and the land fund.
• Whether sponsored videos for small shops and the AI tool series stay on one channel. For now, one channel with two lanes, revisited at 2,500 subscribers.
• Which entity receives the channel’s income, and how money moves from the channel to Johnny Autoseed.

Still to confirm

• Several guides say YouTube’s watch-hour requirement rises to 8,000 on February 1, 2027. I have not confirmed it on YouTube’s own pages. If it does, the views per video needed to qualify inside a year roughly double, from about 208 to about 417 estimate.
• What FSA counts as farm experience, and whether a pending bill to shorten the requirement passed.
• California’s minimum franchise tax for an LLC and whether any first-year exemption applies. Sources disagree.
• The exact disclosure rules, checked against the FTC’s Endorsement Guides and YouTube’s own help pages.

14 · Sources and method

Where the numbers come from

Measured figures come from vidIQ: this channel’s statistics and video list, and a search of peer channels, pulled October 19 to 21, 2026. Every table on this page is generated from one small model, so a changed assumption changes every table at once. Outside figures come from the sources below. I have not checked each against its original page yet, and several are vendor blogs that disagree with each other, so treat them as approximate.

• Sponsorship rates: Kingy AI, vidIQ, OutlierKit
• Ad revenue per 1,000 views: Unilink, Circle
• Partner Program requirements: AIR, vidIQ
• Disclosure: 16 CFR Part 255, Seyfarth, YouTube Help
• Inauthentic content policy: Plagiarism Today, Search Engine Journal
• Land prices: 20.77 acres, Chico, 18.55 acres, Forest Ranch, Land.com
• Farm loans: FSA beginning farmer fact sheet, 7 U.S.C. 1922
• California LLC tax: KDA; the Franchise Tax Board is the primary source

Main assumptions estimate: ad revenue of $5, $8 or $12 per 1,000 views (low, base, high); sponsor rates of $15, $30 or $45 per 1,000 views; a video’s lifetime views at 1.5 times its first 90 days; videos about 8 minutes long with 40% watched; $250 a month of costs; a 30% tax reserve.

15 · The founder

Christopher Tavolazzi

Developer. Creator. Musician. Poet. Somewhere between the first and third cup of coffee. Two decades in marketing, a journalism degree, a former barista. Currently building a startup, a book, an album and a photo series in public.