Content, E-E-A-T and Editorial Economics

Overview

Course 6 established that only original work survives displacement. This course answers the obvious follow-up: original work is expensive, so how much of it can you actually afford, and how do you keep it honest once it is published?

Three ideas carry the course. First, the 20/30/30/20 ratio (Rule 6) forces a portfolio of content types rather than 30 undifferentiated articles. Second, E-E-A-T is not a score — it is a set of artifacts you can hand to a sceptical reader, an advertiser or a regulator on demand. Third, every published asset is a liability as well as an asset: it carries a recurring verification and maintenance cost that belongs in the Course 0 model from the day it goes live.

7.1 What is the 20/30/30/20 content ratio?

What is it?

Rule 6 of the seven global rules. Your published assets are allocated deliberately across four types: 20% acquisition, 30% original research and data, 30% tools and interactive utilities, 20% commercial and conversion.

Why it matters

Left alone, every publishing operation drifts toward whichever content type is cheapest to make — which is now the type most exposed to displacement. The ratio is a budget constraint that forces you to keep producing the expensive, defensible things. It also spreads risk: if generated answers absorb your acquisition content, 60% of the portfolio is still doing work a summary cannot do.

How to do it

1. Classify every planned asset into exactly one of the four types. If it fits two, it is probably weak at both — split it.
2. Apply the ratio to the 12-asset MVP (Rule 5), not to some future 200-page site: roughly 2 acquisition, 4 research, 4 tools, 2 commercial.
3. Sequence by dependency, not by ease. Tools and datasets take longest, so they start first; commercial pages need the tools to exist before they can convert.
4. Recalculate the actual ratio monthly from what shipped, not from what was planned. Drift is invisible until you count.
5. When the ratio slips, correct it by cutting the over-represented type, not by adding to the under-represented one. Adding breaks the cost model.
6. Review the ratio at each pruning gate (Month 3, 6, 9) alongside the pruning scorecard.

Tools needed

Your 12-asset MVP list from Course 4, the content brief builder, and a spreadsheet.

Lab 7.1

Take the 12-asset MVP for one Phase 1 site and assign each asset a type. Compute the real percentages. If they do not approximate 20/30/30/20, rewrite the list until they do — and record which assets you removed to get there.

Artifact

Content ratio plan (per site)
  Asset | Type (acquisition / research / tool / commercial) | Owner | Build order | Est. hours
  Planned ratio ___/___/___/___  vs  target 20/30/30/20
  Assets cut to reach the ratio, and why

Common mistakes

• Filing an opinion piece as "research" because it contains numbers copied from somebody else's report.
• Planning the ratio across a hypothetical 40-asset site so that today's ten articles look acceptable.
• Building the 20% commercial first because it feels closest to revenue, then having nothing for it to convert.

Pro astuces

• The tools bucket is the one most people under-build and it has the lowest displacement exposure of the four. If you are short on time, protect it before anything else.
• One dataset can legitimately supply a research asset, a tool's default values and the weekly brief. Count it once in the ratio, but plan it as three deliverables.

7.2 How do you treat E-E-A-T as infrastructure?

What is it?

E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness — treated not as a ranking dial but as four categories of evidence that must physically exist on your site and be produceable on demand.

Why it matters

Nobody outside the platforms can tell you how these qualities are assessed algorithmically, and anyone who claims a precise mechanism is selling something. But the evidence itself has value regardless of any algorithm: it is what an advertiser's compliance team, a potential sponsor, a journalist and a sceptical reader all look for. Build it for them, and you are covered either way.

How to do it

1. Experience — publish first-hand accounts: what you tested, when, with what, and what went wrong. Include the failures; they are the least fakeable part.
2. Expertise — name real authors with verifiable credentials on a canonical author page, and link every article to its author.
3. Authoritativeness — accumulate third-party evidence: citations of your data, named reviewers, professional affiliations. You cannot self-declare this one.
4. Trustworthiness — make ownership, funding, affiliate relationships, corrections and contact routes obvious and easy to find. This is the load-bearing one for commercial content.
5. Map every claim of E-E-A-T to a URL that proves it. If there is no URL, the quality does not exist as far as any outside party is concerned.
6. Re-audit the map quarterly; credentials, affiliations and reviewer relationships all go stale.

Tools needed

Your about, authors, editorial-policy and corrections pages, plus the reviewer register.

Lab 7.2

Build the evidence map for one site: four rows, and for each a live URL that proves it. Every row without a URL becomes a task with a date. Count how many of the four you can currently evidence.

Artifact

E-E-A-T evidence map (per site)
  Component        | Evidence artifact            | Live URL | Last verified | Gap owner / due date
  Experience       | dated test log               |          |               |
  Expertise        | named author page            |          |               |
  Authoritativeness| third-party citation list    |          |               |
  Trustworthiness  | ownership + funding + policy |          |               |
  Score: ___ / 4 components evidenced with a live URL

Common mistakes

• Adding an "About the author" box containing a stock photo and no verifiable credential, which reduces trust rather than raising it.
• Claiming authoritativeness in your own copy. It is the one component that can only be conferred by others.
• Burying the affiliate and ownership disclosures where only a determined reader would find them.

Pro astuces

• Publishing what went wrong in a test is the single strongest Experience signal available to a small publisher, and almost nobody does it.
• Keep the evidence map in the repository next to the entity definition sheet from Course 6.3. Both are infrastructure and both drift silently.

7.3 How does the 90-day claim register work?

What is it?

Rule 1 made operational. Every factual claim about a platform, a price, a product capability or a regulation is logged with its source, the date it was verified, and an expiry 90 days later. When it expires it is re-verified, amended, or removed.

Why it matters

Publishing is easy; staying correct is the hard, unglamorous, recurring part — and it is what separates a publication from a content farm. Prices change, features are discontinued, guidance is updated. A page that was accurate in March and wrong in June does more damage than a page that never existed, because a reader acted on it.

How to do it

1. Log claims, not pages. One page may carry twelve claims with twelve different expiry dates.
2. Record for each: the claim as published, the exact source URL, who verified it, the verification date, and the 90-day expiry.
3. Classify volatility. Prices and platform behaviour are high-volatility; regulatory text and arithmetic are low. High-volatility claims may need a 30-day cycle instead.
4. Show the reader a visible "last verified" date near volatile claims. This is a trust signal and it costs nothing.
5. On expiry, take one of three actions and record which: re-verify (new date), amend (with a correction note), or remove.
6. Budget the re-verification hours in the cost model. A register with 200 claims on a 90-day cycle is roughly 800 verifications a year — that is a real job, and it caps how much you can publish.

Claim verification register — one row per claim
  ID | Asset URL | Claim as published (verbatim)
     | Volatility (high 30d / medium 90d / low 180d)
     | Primary source URL | Verified by | Verified on | EXPIRES ON
     | Action at expiry (re-verify / amend / remove) | Action date | Correction logged? Y/N

Tools needed

The claim verification register, your corrections page, and a recurring weekly calendar block.

Lab 7.3

Take three published assets. Extract every factual claim, verbatim. Log each with a source URL, a volatility class and an expiry date. Then count the total and multiply by the annual verification minutes — that number is your real editorial maintenance load for just three assets.

Artifact

Populated claim register (3 assets)
  Total claims: ___  | High-volatility: ___  | Annual re-verifications: ___
  Estimated annual maintenance hours: ___
  Implied maximum publishable assets at my available hours: ___

Common mistakes

• Registering pages instead of claims, which makes re-verification vague and therefore skipped.
• Citing a secondary source — someone else's blog post about the primary source — which inherits their error and their staleness.
• Silently editing an expired claim with no correction note, which destroys the audit trail you built the register for.

Pro astuces

• Sort by expiry ascending and work the top of the list for thirty minutes a week. The register only fails when it becomes a wall of overdue rows.
• A visible "Last verified: [date]" line next to a price is one of the cheapest trust signals in publishing, and it also disciplines you into actually doing it.

7.4 What does expert review actually cost?

What is it?

Reviewer economics: deciding which assets require a qualified human reviewer, what that review costs in money and calendar time, and how the relationship is documented so the credential is real.

Why it matters

All three Phase 1 niches — SMB finance, B2B SaaS and SMB cybersecurity — publish content that people act on with money or with their security posture. Getting it wrong is not a ranking problem, it is a harm problem. Review is the control. But review is also the single most under-budgeted line in independent publishing: people plan the writing and forget that a qualified reviewer costs real money and takes a week of calendar time.

How to do it

1. Define your review tiers explicitly. Not everything needs a reviewer; deciding which things do is the whole exercise.
2. Recruit reviewers with verifiable credentials and a real public identity. An anonymous reviewer provides no Authoritativeness at all.
3. Agree scope in writing: what they are certifying, what they are not, turnaround time, and rate per asset or per hour.
4. Publish the reviewer's name, credential and review date on the asset, and link to their canonical page.
5. Record every review in the reviewer register with the date, the scope and the outcome, including changes they required.
6. Put review cost into the per-asset cost model in 7.5 and into the 12-month budget. Re-review on a schedule for high-volatility subject matter.

Tools needed

Reviewer register, your editorial policy page, and the unit economics model.

Lab 7.4

Assign a review tier to all 12 MVP assets for one site. Identify every Tier 2 asset, then find and approach one real qualified reviewer: agree scope, turnaround and rate in writing. Add the total annual review cost to your cost model and check whether the model still passes.

Artifact

Reviewer register & rate card
  Reviewer | Credential (verifiable, with URL) | Subject scope | Rate | Turnaround | Conflicts declared
  Asset | Review tier | Reviewer | Reviewed on | Changes required | Re-review due
  Annual review cost: ___  | Cost model still passes? Y/N

Common mistakes

• Publishing "medically/financially/technically reviewed" badges with no named reviewer behind them. This is a fabricated trust signal and it is the fastest way to lose an advertiser.
• Budgeting reviewer money but not reviewer calendar time, then publishing unreviewed because the deadline arrived first.
• Never re-reviewing, so a Tier 2 asset carries a two-year-old review date on volatile subject matter.

Pro astuces

• A retainer for a fixed number of assets per month is usually cheaper and far more reliable than per-asset commissioning, and it fixes the calendar problem.
• Ask reviewers to record what they changed. Those notes are the best training material you will ever have for your own writing.

7.5 What is the true cost per asset?

What is it?

The fully loaded cost of one published asset over twelve months: research, production, review, media, publishing, and the recurring verification and maintenance it commits you to for as long as it stays live.

Why it matters

Almost everyone costs content as "hours to write it". That number is typically less than half the truth, which is why content budgets overrun and why portfolios grow past the point of maintainability. You cannot apply Rule 2's kill criteria to something whose cost you have never measured, and you cannot honestly compare an asset against an ad spend or a sponsorship until you can state its cost as a number.

How to do it

1. Track one real asset end to end with a timer. Not an estimate — a measurement. Estimates are consistently and substantially low.
2. Include the invisible line items: research and reading, source verification, image or diagram production, structured data, internal linking, publishing and QA.
3. Add the review cost from 7.4 and the annual re-verification cost from 7.3.
4. Value your own hours at a real rate. Unpriced founder time is how a project appears profitable while quietly destroying value.
5. Compute cost per asset for year one, then the recurring annual cost from year two onward. They are different numbers and both matter.
6. Divide the annual content budget by the loaded cost to get the number of assets you can genuinely afford. That number, not ambition, sets the plan.

Cost-per-asset model — one column per asset type
  PRODUCTION (one-off)
    Research & source gathering        hrs × rate
    Original data collection            hrs × rate      (research assets only)
    Drafting                            hrs × rate
    Diagrams / images / screenshots     hrs × rate + tool fees
    Build / interactivity               hrs × rate      (tools only)
    Structured data, links, QA          hrs × rate
    Expert review (tier 2/3)            fee
  RECURRING (annual, from launch)
    Claim re-verification               claims × cycles × minutes
    Refresh / rewrite allowance         hrs × rate
    Reviewer re-review                  fee × frequency
  = YEAR 1 LOADED COST ___    = ANNUAL RECURRING COST ___
  Annual content budget ___ ÷ loaded cost ___ = assets I can afford: ___

Tools needed

Unit economics model, a timer, and your claim register counts from 7.3.

Lab 7.5

Time one asset from first research minute to published, honestly, including every interruption. Fill in the model. Compare the measured number to what you would have estimated before starting, and write down the gap as a percentage.

Artifact

Cost-per-asset model
  Measured asset: ___  | Estimated before: ___  | Actual: ___  | Gap: ___%
  Year 1 loaded cost by type: acquisition ___ research ___ tool ___ commercial ___
  Annual recurring cost per asset: ___
  Assets affordable per year at current budget: ___

Common mistakes

• Costing only the drafting hours, which is the single most common financial error in independent publishing.
• Treating founder time as free, which makes every asset look profitable and makes the kill criteria unusable.
• Ignoring the recurring cost, so year three arrives with a maintenance bill larger than the entire year-one budget.

Pro astuces

• Tool assets usually have the highest build cost and the lowest recurring cost. Research assets are the reverse. Knowing which is which changes your build order.
• Once you can state cost per asset, sponsorship and advertising conversations change completely: you can price a slot against a real production cost instead of guessing.

7.6 How do you run the refresh, correct and retire loop?

What is it?

The standing editorial maintenance cycle: scheduled refreshes for assets that still earn their keep, a public corrections process for what you got wrong, and a retirement route for assets that no longer justify their recurring cost.

Why it matters

Publishing without a retirement route is how a twelve-asset MVP becomes an eighty-asset liability. Every live asset consumes verification time, carries a compliance obligation and dilutes the reader's sense of what the site is for. The loop is what keeps the portfolio the size your budget said it should be — and it is the mechanism that feeds the Month 3/6/9 pruning gates with evidence instead of opinion.

How to do it

1. Give every asset a scheduled review date at publication, derived from its volatility class in 7.3 — not "when we get round to it".
2. At each review, take one of four decisions: keep (verified, unchanged), refresh (update and re-date), merge (fold into a stronger asset with a redirect), or retire (remove and redirect).
3. Run corrections publicly. A dated corrections log with what was wrong, what it is now, and when it changed is a strong Trustworthiness artifact.
4. Never silently rewrite a substantive claim. Amend visibly and log it — this is exactly what the register exists to make possible.
5. Retire on cost, not on sentiment: if annual recurring cost exceeds the asset's contribution to revenue, subscribers, citations or internal linking, it goes.
6. Feed every decision into the pruning scorecard so the Month 3, 6 and 9 gates see the pattern across the whole portfolio.

Tools needed

Your claim register, pruning scorecard, corrections page, and a 12-month calendar.

Lab 7.6

Build the maintenance calendar for twelve months across all current assets: every review date, every re-verification block, every scheduled re-review. Then total the hours per month. If any month exceeds your available editorial hours, cut assets now rather than discovering it in month seven.

Artifact

Editorial maintenance calendar (12 months)
  Month | Assets due for review | Claims due for re-verification | Re-reviews due | TOTAL HOURS
  Available editorial hours per month: ___
  Months over capacity: ___   Assets cut to fit: ___
  Retirement rule (written): retire when annual recurring cost exceeds ___

Common mistakes

• Refreshing by changing the date and adding a paragraph, without re-verifying a single claim. This is worse than doing nothing because it launders staleness as freshness.
• Keeping assets out of attachment — "it took three days to write" is a sunk cost, not a reason.
• Handling corrections privately, forfeiting the strongest trust artifact available to a small publisher.

Pro astuces

• Build the maintenance calendar before you publish the twelfth asset. It is the only way to see the ceiling before you hit it.
• A well-run corrections page attracts more respect than an unblemished one attracts suspicion. Readers know that nobody publishing volatile facts is never wrong.

Scenario assessment: the freelancer who is cheaper than the reviewer

The SMB Cybersecurity site has 14 live assets. A freelance writer offers 4 assets a month at a price well below the measured cost per asset from 7.5 — but the quote excludes expert review, and the Tier 2 reviewer costs more per asset than the writer does. The operator's plan is to publish the four monthly assets now, mark them "reviewed" using an advisor's name who has agreed "in principle" but has not read anything, and add real review later once revenue covers it. Two of the fourteen live assets already have claims 40 days past expiry, and the maintenance calendar from 7.6 shows month five already at 130% of available hours.

Final project: a costed, evidenced editorial system

Turn content from an activity into a system with a ratio, an evidence base, an expiry mechanism, a measured unit cost and a maintenance ceiling.

Deliverable
1. Content ratio plan for one site's 12-asset MVP, matching 20/30/30/20, with the cuts recorded
2. E-E-A-T evidence map: all four components, each with a live URL or a dated gap task
3. Claim register populated from 3 real assets, with volatility classes and expiry dates
4. Reviewer register with at least one real reviewer contracted in writing: scope, rate, turnaround
5. Cost-per-asset model built from ONE timed, measured asset — estimate vs actual gap stated
6. 12-month editorial maintenance calendar with monthly hours totalled against capacity
7. A written retirement rule and a written publishing-rate cap derived from the calendar

Pass standard: you can state your loaded cost per asset as a number, your annual maintenance hours as a number, and the maximum number of assets you can afford to keep live — and your current plan does not exceed it.

Course 7 checklist

• All 12 MVP assets classified into exactly one content type
• Actual ratio computed and within reach of 20/30/30/20
• Ratio corrected by cutting, not by adding
• E-E-A-T evidence map built with a live URL per component
• Named author page live, with verifiable credentials
• Ownership, funding and affiliate disclosures easy to find on every site
• Common ownership of the three sites stated publicly
• Claim register populated from 3 assets, claim by claim
• Every claim carries a primary source URL and a 90-day expiry
• Visible "last verified" date shown next to volatile claims
• Review tier assigned to all 12 assets
• One real reviewer contracted: scope, rate and turnaround in writing
• No review badge published anywhere without a logged, named reviewer
• One asset timed end to end; estimate-vs-actual gap recorded
• Own hours priced at a real rate in the cost model
• Annual recurring cost per asset calculated separately from year-one cost
• 12-month maintenance calendar built and totalled against available hours
• Written retirement rule and publishing-rate cap agreed

AI-agent prompt for Course 7

Act as an editorial operations auditor for an independent publisher running three commercially monetised sites in SMB finance, B2B SaaS/AI tools, and SMB cybersecurity.

I will paste: my asset list, my available editorial hours per month, my content budget, my hourly rate, and any existing about/author/policy pages.

Do this:
1. Classify every asset as acquisition / original research / tool / commercial and compute my actual ratio against the 20/30/30/20 target. If I am off target, tell me what to CUT — never what to add.
2. Build an E-E-A-T evidence map (Experience, Expertise, Authoritativeness, Trustworthiness). For each component name the specific artifact and the URL that would prove it, and flag every component I currently cannot evidence. Note explicitly that Authoritativeness cannot be self-declared.
3. Extract every factual claim from the assets I paste, verbatim. Assign each a volatility class (high 30d / medium 90d / low 180d), and produce a claim register with expiry dates.
4. From that register, calculate my annual re-verification load in hours, and tell me the MAXIMUM number of live assets my available hours can support. State that number plainly even if it is far below my plan.
5. Assign a review tier (0 none / 1 factual / 2 expert / 3 legal) to every asset, justify each Tier 2 and 3, and estimate the annual review cost.
6. Build a fully loaded cost-per-asset model: research, drafting, media, build, QA, review, plus recurring re-verification and refresh. Price my own time at the rate I give you. Show year-1 cost and annual recurring cost separately.
7. Produce a 12-month maintenance calendar with monthly hour totals, and mark every month that exceeds my capacity.

Rules:
- Spell it E-E-A-T. Never claim to know how any platform scores it, or that any action guarantees a ranking effect.
- Never suggest publishing a review, verification or expert badge that is not backed by a named, logged, consenting reviewer. Refuse this if I ask for it.
- Do not invent industry benchmark costs, conversion rates or "average" content prices. If I have not measured it, mark it UNMEASURED and tell me how to measure it.
- Where subject matter is financial, security or legal, add a note that professional advice is required and that you are not providing it.
- Prefer cutting scope over increasing budget in every recommendation.
Output as tables, with the affordable-asset count and the over-capacity months called out in plain text above the tables.

Primary sources

• Creating helpful, reliable, people-first content — the self-assessment questions, including the E-E-A-T framing.
• Search Quality Rater Guidelines (PDF) — the primary text where Experience, Expertise, Authoritativeness and Trustworthiness are defined. Read the definitions themselves rather than summaries of them.
• FTC endorsement, influencer and review guidance — applies directly to review badges, comparisons, affiliate content and disclosure of commonly owned properties.
• Article structured data — how author, reviewer and date provenance is expressed to machines.

Every claim above about how any platform evaluates content is a dated claim with a 90-day expiry. Re-verify against the primary source before you rely on it, and log the verification in your own register.