Decision Economics, Risk, and Kill Criteria

Overview: why economics comes before SEO

In v1 of this academy, Course 1 was SEO foundations. That order rewarded enthusiasm and punished judgement: students launched ten sites, published hundreds of pages, and discovered the cost of expert review, tool development and content refresh only after the money was spent.

v2 puts a financial gate in front of everything. A website is a capital allocation decision with a long payback period and a real chance of returning zero. Course 0 teaches you to treat it that way — and to write down, in advance, the evidence that would prove you wrong.

0.1 What is decision economics?

What is it?

Decision economics is the practice of pricing a decision before making it. For a publisher it means one number — the fully loaded twelve-month cost of operating one website at the quality level the topic demands — and the confidence interval around it.

Why it matters

Search results in commercially valuable niches are now defended by teams with editors, licensed reviewers, original data and engineering. Entering that market without knowing your budget ceiling is how people spend eleven months of effort and abandon a site in month twelve, one month before compounding would have started.

How to do it

1. List every cost line below, even the ones you plan to do yourself.
2. Price your own labour at a real hourly rate. Free labour hides the true cost of a bad niche.
3. Get three actual quotes for anything you will outsource (writing, expert review, tool development).
4. Split costs into setup (one-off), run (monthly) and variable (per asset).
5. Add a 20% contingency line. Refreshes and reviewer re-reads always exceed the estimate.
6. Produce three scenarios: lean, expected, and quality-first.

Cost lines to price (per site, 12 months)

Domain                       Screenshots/media
Hosting / CMS                Newsletter software
Cloudflare / tooling         Analytics
SEO tools                    PR / link building
Content briefs               Maintenance
Writing                      Refresh cost
Editing                      Contingency (20%)
Expert review                Your own labour (hours x rate)
Legal review
Design
Calculator / tool development

Tools needed

A spreadsheet (or the academy unit economics & budget tool), three written quotes, and your own time log for two typical weeks.

Lab 0.1

Price all sixteen lines for SMB Finance and Banking, the first Phase 1 site. Produce lean, expected and quality-first totals. Then answer in one sentence: what is the largest single cost, and would you still build if it doubled?

Artifact

12-month budget model
- Setup costs
- Monthly run costs
- Variable cost per asset
- Three scenarios (lean / expected / quality-first)
- Budget ceiling and owner

Common mistakes

• Valuing your own time at zero, which makes every niche look profitable.
• Budgeting content but not refresh — dated assets need money every quarter, forever.
• Forgetting expert review in YMYL topics, then quietly dropping it when the invoice arrives.
• Treating tool development as a weekend project when it is the main defensible asset.

Pro astuces

• Budget the first tool before the first article. It is the asset an AI answer cannot replace.
• Track cost per published and reviewed asset, not per draft. Drafts are not inventory.
• Keep a "decisions we did not fund" list. It becomes your Phase 2 pipeline.

0.2 What are unit economics?

What is it?

Unit economics reduce a whole business to the profit or loss of one repeatable unit. For a publisher the units are: one published asset, one subscriber, one lead, and one tool completion.

Why it matters

Portfolio-level revenue hides everything. A site earning money overall can still be publishing assets that will never repay their cost. Unit economics tell you which asset type to make more of — and in the AI era the answer is usually tools, data and commercial pages, not explainers.

How to do it

1. Compute cost per published asset = total variable cost ÷ assets published.
2. Compute cost per expert-reviewed page separately; it is often 2–4× a standard page.
3. Compute cost per subscriber = acquisition spend ÷ net new subscribers.
4. Compute revenue per page, per subscriber and per lead over a rolling 90 days.
5. Payback period = cost of asset ÷ monthly revenue attributable to that asset.
6. Break-even month = the month cumulative revenue crosses cumulative cost.

Unit economics sheet

Cost per published asset          Revenue per page
Cost per expert-reviewed page     Revenue per subscriber
Cost per subscriber               Revenue per lead
Cost per tool                     Payback period (months)
Cost per data asset               Break-even month

Tools needed

Spreadsheet, analytics revenue export, affiliate/network reports, newsletter platform stats. Use Unit economics & payback to model it before you have real data.

Lab 0.2

Model a site with 12 assets: 2 pillars at high cost, 3 money pages at medium cost, 2 tools at high cost, 2 data assets at medium-high cost, 3 support articles at low cost. Assume a plausible revenue path and find the break-even month. Then halve the traffic assumption and find it again.

Artifact

Unit economics sheet with payback per asset type and two traffic scenarios

Common mistakes

• Averaging revenue across all pages, which flatters weak informational content.
• Ignoring refresh cost, which turns a profitable dated asset into a loss-maker in year two.
• Counting gross affiliate revenue before reversals and returns.
• Assuming subscriber value without ever measuring revenue per subscriber.

Pro astuces

• Report revenue per asset type monthly. It is the single most decision-useful table you will own.
• If cost per subscriber is below three months of revenue per subscriber, spend more on the lead magnet.
• Tools have near-zero marginal cost after launch — model them as capital, not content.

0.3 What is a falsifiable business thesis?

What is it?

A thesis is a written belief about a specific audience, plus the evidence that would prove it false by a specific date. "This niche has high CPM" is not a thesis. It is a hope with a number attached.

Why it matters

Without falsifiability there is no honest way to stop. Every disappointing month gets reinterpreted as "SEO takes time", and the sunk cost grows. A thesis converts opinion into a testable claim, which is also what makes a portfolio sellable.

How to do it

1. Name the buyer precisely (not "small businesses" but "owner-operators with 3–20 staff who handle their own banking").
2. State the belief in one sentence, including the behaviour you expect.
3. List the evidence required by month 6, with numbers.
4. List the evidence that would disprove it, with numbers.
5. Name the owner and the review date.

Example thesis — Site 1 (SMB Finance and Banking)

We believe small-business operators will subscribe to a weekly finance
newsletter if we provide calculators, templates and funding updates.

Evidence needed by month 6:
- X subscribers
- Y tool completions
- Z pages with search traction
- at least N qualified affiliate/lead clicks

Disproof by month 6:
- signup rate below 1% on tool pages after 2 iterations
- fewer than 3 assets with any impression growth
- zero repeat newsletter click cohort

Tools needed

Investment memo template, keyword/impression baseline, competitor newsletter samples, one page of buyer interview notes (three real conversations beat thirty assumptions).

Lab 0.3

Write theses for all three Phase 1 sites. Each must fit on one page and contain at least four numeric disproof conditions. Show one to somebody who will argue with you and record their strongest objection.

Artifact

Site investment memo (one per Phase 1 site)

Common mistakes

• Writing a thesis about traffic instead of about people and payment.
• Setting evidence thresholds after seeing the first data.
• Confusing the method ("SEO works") with the business bet ("this audience will subscribe").

Pro astuces

• Put the disproof conditions in the same document as the budget. Approval means accepting both.
• Date-stamp the memo and never edit it in place — version it, so future you can audit past you.

0.4 What are kill, freeze, and pivot criteria?

What is it?

Thresholds, written before launch, that convert evidence into one of four decisions: continue, pivot the content mix, freeze new investment, or kill and reallocate.

Why it matters

Portfolio returns come from concentration, and concentration requires pruning. Publishers who never kill anything end up maintaining nine mediocre sites and cannot fund the one that works.

How to do it

1. Define three gates: month 3 (infrastructure), month 6 (traction), month 9 (economics).
2. For each gate, list observable checks — not feelings.
3. Attach one decision per gate, with a default action if nobody reviews it.
4. Store the gates where you will see them: the governance dashboard, not a buried doc.

Month 3 gate                     Month 6 gate
- Site technically indexed        - Search impressions growing
- Newsletter live                 - Email conversion above target
- First tool live                 - At least one monetization signal
- 12 assets published/scheduled   - One linkable asset campaign done

Month 9 gate
- Revenue per published page evaluated
- Subscriber acquisition evaluated
- Keep / freeze / pivot / sell decision made

Tools needed

Calendar reminders on the gate dates, the portfolio pruning scorecard, and one named decision owner.

Lab 0.4

Write gate criteria for one Phase 1 site with explicit numbers, then write the sentence you will say to yourself if the month 6 gate fails. Rehearsing the stop is what makes it possible.

Artifact

Kill/freeze/pivot criteria document with gate dates and default actions

Common mistakes

• Gates with no dates, or dates with no owner.
• Killing on traffic alone, when subscribers and tool usage are growing.
• Freezing a site without cancelling its recurring costs.

Pro astuces

• "Freeze" beats "kill" when a site has a small but engaged list — stop publishing, keep sending.
• A failed gate should trigger a written post-mortem of one page maximum. Longer ones never get read.

0.5 What is build vs buy?

What is it?

The choice between starting a site from zero and acquiring an existing one. There is a legitimate version — buying a real operating website with traffic, revenue and a list — and a risky version: buying expired domains to manipulate authority.

Why it matters

Buying can compress two years of ramp into one transaction. It can also transfer somebody else's manual actions, unlicensed lead-gen liabilities and non-transferable affiliate accounts to you. The difference is due diligence.

How to do it

1. Request read access to analytics and Search Console — not screenshots.
2. Reconcile revenue against network dashboards and bank statements for 12 months.
3. Check the backlink profile for paid or networked patterns you would have to disavow or live with.
4. Ask directly about manual actions, and verify in Search Console yourself.
5. Check whether affiliate accounts, reviewer contracts and licences transfer at all.
6. Price the refresh debt: how many assets are stale, and what will re-review cost?

Due diligence checklist
Traffic history        Content quality
Revenue proof          Legal liabilities
Email list quality     Affiliate account transferability
Backlink profile       Analytics access
Manual actions         Ownership/IP rights
                       Reason for sale

Tools needed

Search Console and analytics read access, a backlink tool, an escrow service, and a lawyer for the asset purchase agreement.

Lab 0.5

Find one publicly listed content site in a Phase 1 adjacent niche. Run the eleven-point checklist using only what the listing discloses, then write the five questions that would decide the deal. Do not contact the seller unless you are genuinely buying.

Artifact

Build-vs-buy decision checklist with a build/buy/walk recommendation

Common mistakes

• Believing screenshots. Always verify inside the platform.
• Ignoring the email list's consent status — an unconsented list is a liability, not an asset.
• Underpricing refresh debt on dated content.

Pro astuces

• Ask for the seller's refresh log. Its absence tells you the true content age.
• Weight a real newsletter list far above raw sessions; it survives ranking volatility.

0.6 What is governance?

What is it?

The small set of registers and rituals that keep a portfolio honest: who owns what, which claims are still verified, which risks are live, and what gets reviewed each month.

Why it matters

Publishing is a slow-feedback business. Governance is how you notice a problem in month 4 instead of month 14 — an expired affiliate term, a reviewer who stopped replying, a single program that quietly became 70% of revenue.

How to do it

1. Create four registers: claim verification, risk, reviewer, and content approval.
2. Give every register an owner and a review cadence (monthly is enough).
3. Adopt a versioning protocol: every SOP and claim has a verified date and an expiry date.
4. Set the 90-day recheck rule for all platform and commercial claims.
5. Run a 60-minute monthly portfolio review with a fixed agenda.

Risk register fields
Risk | Category | Likelihood | Impact | Leading indicator |
Owner | Mitigation | Trigger | Decision | Review date

Categories: platform, AI-search, Google policy, compliance,
reviewer/key-person, revenue concentration, traffic concentration,
cash burn, security/privacy, acquisition

Tools needed

A spreadsheet or database, calendar recurrence, and the academy register tools. Nothing heavier — governance dies when it needs software adoption.

Lab 0.6

Populate the risk register with at least one live risk per category (ten rows), each with a leading indicator and a trigger. Then populate the claim register with five claims you already believe about ad networks or affiliate terms, and mark how many you can actually source today.

Artifact

Portfolio governance dashboard: four registers + monthly agenda + owners

Common mistakes

• Registers with no owner — they become archaeology within two months.
• Risks without leading indicators, which means you only learn about them after the damage.
• Copying compliance claims from blog posts instead of primary sources.

Pro astuces

• Put the expiry date column first. Sorting by it turns the register into a to-do list.
• Log the decision, not just the risk. Future you needs to know what was chosen and why.

Scenario assessment: the enthusiastic ten-site plan

Scenario assessments replace most quizzes in v2. There is rarely one correct answer — there is a defensible one, supported by numbers you can show.

A student has 18,000 of capital and 20 hours a week. They plan to launch all ten niches at once, publishing 30 pages per site in the first quarter, using AI drafts with light editing. Three of the niches (insurance, mortgage, health benefits) are YMYL. They have no reviewers, no newsletter and no tools planned. They ask you to sanity-check the plan.

Final project: the 3-site portfolio investment memo

Produce one document an investor (or a sceptical spouse) could approve or reject.

Investment memo contents
Why these 3 sites            Traffic assumptions
Budget                       Kill criteria
Expected cost per site       Risk register
Expected cost per asset      Month 9 pruning plan
Reviewer requirements
Revenue assumptions

Pass standard: every number is traceable to a quote, a measurement or a stated assumption; every assumption has a disproof condition; the memo fits in five pages.

Course 0 checklist

Ticks are saved in your browser with IndexedDB.

• 12-month budget model complete with three scenarios and a contingency line
• Own labour priced at a real hourly rate
• Unit economics sheet with payback per asset type
• Falsifiable thesis written for each Phase 1 site
• Month 3, 6 and 9 gates dated, with numbers and an owner
• Build-vs-buy checklist applied to one real listing
• Risk register populated across all ten categories
• Claim verification register started, with 90-day expiry dates
• Monthly review scheduled with a fixed agenda
• Investment memo written and reviewed by one critic

AI-agent prompt for Course 0

Act as an editorial economist and portfolio risk analyst for an independent web publisher.

Context I will paste: my niche, my capital, my weekly hours, my quotes for writing, editing and expert review.

Do this:
1. Build a 12-month cost model with setup, run and variable-per-asset lines, plus a 20% contingency and my own labour priced at the rate I give you. Produce lean, expected and quality-first scenarios.
2. Derive unit economics: cost per published asset, cost per expert-reviewed page, cost per subscriber, revenue per page, revenue per subscriber, revenue per lead, payback period and break-even month.
3. Write a falsifiable thesis for the site: one belief sentence, four numeric evidence conditions by month 6, and four numeric disproof conditions.
4. Propose month 3, 6 and 9 gates with observable checks and a default action if the gate is missed.
5. Populate a risk register with one risk per category: platform, AI-search, Google policy, compliance, reviewer/key-person, revenue concentration, traffic concentration, cash burn, security/privacy, acquisition. Include leading indicator, trigger and owner.
6. List every assumption you had to invent, and mark each one as "needs primary source" with the type of source required.

Rules:
- Never state a platform, ad-network or affiliate term as fact. Output it as a claim row with Claim / Source URL / Verified / Owner / Expires / Status and leave the URL for me to fill.
- Assume the MVP is 12 high-value assets, not 30 articles.
- Assume 3 sites launch first, not 10.
- Flag any recommendation that would require legal counsel before launch.
Output as tables I can paste into a spreadsheet.

Primary sources for this course

Verify before you publish anything derived from them, and record the verification in your claim register.

• Google Search Essentials
• Google spam policies — read before considering any acquisition for link equity
• FTC endorsement, influencer and review guidance