Find and repair silent failures in AI-assisted publishing operations.

Leefan Reports helps teams that are already publishing with AI-assisted workflows but cannot reliably tell what is broken: measurement gaps, weak audit trails, page-quality drift, prompt/card confusion, and reviews that depend too much on runtime self-report.

The engagement ladder is intentionally narrow: diagnostic first, implementation when the repair path is clear, and a capped monthly operations retainer only when there is a real loop to keep healthy. We do not sell links, ranking guarantees, traffic guarantees, or vague AI transformation decks.

The offer ladder

1. Diagnostic Sprint

For teams who are already shipping pages or workflows and need a senior outside read on what is silently broken. The diagnostic looks at the current publishing loop, measurement coverage, card/prompt practice, audit trail, sample page quality, and the gaps that should be fixed first.

• Written diagnostic memo tied to supplied artifacts, not generic best practices.
• Prioritized repair queue with severity, effort, and recommended owner.
• 90-minute walkthrough with the buyer-side operator and decision maker.
• Starter artifacts when useful: card schema, regression DB shape, measurement checklist, decision-log template.

Best fit: “We know something is wrong with our AI content / pSEO / analytics loop, but we do not know what to fix first.”

Wrong fit: the team is not shipping yet, wants a broad strategy deck, or wants guaranteed rankings rather than an operational diagnosis. Quoted after scope memo

2. Implementation Sprint

For teams who have a clear repair map and want the loop installed inside their existing tools. The work product lives in the buyer’s repo, Notion, spreadsheet, GA4/GTM property, CMS, or other agreed system. No new SaaS is added just to make the engagement look bigger.

• Card library scoped to the buyer’s actual workflows, usually 6–12 cards.
• Regression DB or decision-log structure in the buyer’s existing system.
• Measurement checks derived from DiagKit’s GA4 click-event checker posture.
• Page-quality gate for thinness, uniqueness, template smell, and stop-publishing conditions.
• Two named operators trained: one runner and one reviewer.
• Handover pack with owners, recurring checks, and exit conditions.

Best fit: “We need the diagnostic fixed and installed so our own team can run it.” Quoted after scope memo

Out of scope: we do not become a content agency, own the buyer’s tools, buy links, run ads, or promise traffic/revenue numbers.

3. Monthly Ops Retainer

For teams that already have a loop installed and want senior review, repair discipline, and decision support without creating an open-ended agency relationship.

• Recurring review of the regression DB, decision log, and page-quality signals.
• Written “what to stop / fix / scale” memo.
• Defined async review cap for non-trivial decisions.
• Small repair cards, diagnostic updates, or measurement-check updates within a monthly cap.
• Quarterly loop-shape review for retained clients.

Best fit: “The loop works, but we need it to keep improving without daily founder supervision.” Quoted after scope memo

Wrong fit: the team has not installed a loop yet, or wants unlimited daily operational labor.

What every engagement includes (and what it does not)

The engagement contract defaults to “we do not own the buyer’s tools or domains”. Artifacts produced in the engagement live in the buyer’s repo, the buyer’s Notion, the buyer’s GA4 property. There is no Leefan Reports-controlled dashboard that the buyer loses access to when an engagement ends.

Process and timing

1. Inquiry email. You send a four-paragraph email (see homepage §5) to contact@leefan.co.jp.
2. Triage (within 2 business days). We reply with one of three: a short note declining; a short note asking for two or three clarifying questions; a scheduling link for a 45-minute discovery call.
3. Discovery call. 45 minutes, with the operator who would run the engagement. At the end of the call we say one of: Shape A makes sense; Shape B makes sense and Shape A is not needed; neither shape makes sense. No “send a proposal and follow up next week”.
4. Scope memo. One page. What we will do, what we will not do, the price, the start date, the end date, what “done” looks like. Client signs the memo or asks for revisions. We do not start without a signed memo.
5. Engagement. Standing weekly checkpoint, async between checkpoints. The standing checkpoint exists so a 6-week engagement does not silently turn into a 12-week one.

Pricing in plain words

What “from” means: these are floors, not ceilings. A 200-page client portfolio with multilingual variants and 12 regression-DB workflows costs more than a 30-page single-language site. We name the actual number on the scope memo, after the discovery call. We do not negotiate down — the floor is the floor.

What we will not do at any price:

• Guarantee rankings. Search-engine rankings are not under our control or the client’s.
• Guarantee revenue. Revenue depends on the client’s product, market, and pricing far more than on the publishing stack.
• Sell a “premium tier” that includes paid backlinks, paid placements, or any link-buying.

Failure modes the methodology is designed to prevent

1. The buyer cannot field an in-house operator. The card practice requires a named operator on the buyer side; if that person does not exist by week two, we end the engagement and refund the unworked portion of the setup fee.
2. The buyer wants to outsource thinking. The loop produces auditable AI output; it does not produce strategy.
3. Tool sprawl. A team with 14 SaaS subscriptions each owning a slice. The engagement reduces that surface; if the buyer is not willing to reduce it, the loop install becomes a glue project.
4. Agency on the other side. If the buyer already has an SEO agency optimising for ranking and a content agency optimising for word count, the loop conflicts with both. Finish those engagements first.
5. Compliance surprise. Some content categories carry compliance requirements (medical, financial, gambling, age-gated material) that change the diagnostic before we begin.

FAQ

Do you take percentage-of-revenue engagements?
No. Percentage-of-revenue aligns our incentive with a number we do not control. Fixed-fee, defined-scope, no equity.

Do you work with Japanese-language sites only?
No. The loop is language-agnostic. Default operating language is Japanese; English is an option on the inquiry.

Do you sign NDAs?
Yes, before the discovery call if the client prefers. Standard mutual NDA; we will read a different one and sign or red-line.

Will the work product be open-source or proprietary?
The client’s cards, regression DB, and site are the client’s IP. The Leefan Reports card library — the framework — is open and visible on this site.

Can you start next week?
Almost never. We hold one Shape A slot and one Shape B slot per quarter; we say the actual lead time on the discovery call.

Inquire

Inquiries are read by one person. We reply within two business days, or we reply with a short note declining the engagement.

contact@leefan.co.jp

Address shown as plain text; page uses plain text rather than a mailto link. The intake mailbox exists; no mailto link is used.

Please include: what your team currently publishes, the specific bottleneck, the tools involved, whether there is a named buyer-side operator, and the rough timeline.