SuperAPI AI Directory

case-study-writer

General & other

View repository

Updated 10-16 09:44

About this skill

Case Study Writer

Produces a case study in the Case Study Kit structure — the canonical 12-section format (result-first, scannable in 60 seconds). The case study is a salesperson: it convinces the NEXT prospect with numbers and a sequence of decisions, not adjectives.

What this skill produces

  1. Website case page — full 12-section copy with inline visual markers, ready for CMS.
  2. One-pager outline — condensed sections for the PDF sales attaches to proposals.
  3. LinkedIn handoff — a post-ready summary passed to the post-writing process.
  4. On demand: the SME interview questionnaire (if nothing is written yet), the designer visual brief, and the pre-launch checklist.

Working language: match the user. Case language: the client's market (ask if unclear — usually English). Ask up front which output formats are needed (default = website page).

Hard rules

  • No fabrication. Every number comes from the user or their materials. Missing number → insert a visible placeholder [[уточнити: X]] and list all placeholders at the end. Never invent metrics, quotes, or timelines.
  • No result — no case. No measurable outcome yet → say so and offer Weak-numbers mode (below) instead of faking impact.
  • ≤150 words per section. Must scan in 60 seconds. "2000 words" is NOT the goal — clarity is. If a section does not move the sale, cut it.
  • Outcome ≠ output. Results = business consequences (before → after). Features = what was shipped. Never mix them.
  • Client confidentiality / NDA. Ask whether the client's name/logo may be used. If not, anonymize: "a 120-person MEP engineering firm (US)" — industry + size + geo, no name. Blur/anonymize any NDA data in screenshots. Ask what may NOT be published (numbers, names).
  • Quotes only verbatim from provided materials. Light grammar cleanup allowed, meaning untouched. Minimum 2 pull-quotes (one at the Challenge, one at the Result) + 1 full testimonial.
  • CTA mandatory, ends with a question. A case without a CTA loses the lead.
  • No AI clichés, no negative parallelism ("not X, but Y" — banned; "cutting-edge", "seamless", "game-changing" — banned). Active voice, past tense, concrete. Run the final through anticopywriting-ai.
  • Fact-check every number and name against the SME answers before shipping.

Step 1 — Intake (SME questionnaire → 12 sections)

If the user has notes / a call transcript → extract answers from there first, ask only the gaps. If they have nothing written → output this questionnaire (they run it with the delivery team or the client in one 45–60 min interview, then continue). Ask open questions, always ask for numbers; "don't know the exact figure" → ask for a range/estimate.

  • A · Client context (→ sections 2, 3) — what the company does, market/geo, headcount; how they stand out; who the stakeholder was (role, not name yet); engagement model, duration, our team composition. (Can be self-researched.)
  • B · Challenge (→ section 4) — the business result they needed and why now (deadline, investor, competitor); the hardest technical part (legacy, integrations, load, data); the delivery challenges (timeline, team, unstable requirements, time zones); the cost of NOT solving it; prior attempts / other vendor and why it failed.
  • C · Why us + approach (→ section 5) — why us over a competitor / in-house; how it started (discovery / workshop / audit); key milestones and dates (Week 0 → launch); what changed along the way (scope, priorities).
  • D · Solution (→ section 6) — how EACH of the 3 challenges (business / tech / delivery) was solved; the key architectural/product decisions and why; deliberate trade-offs (scope cut, MVP focus); which screen/flow best illustrates the solution (→ designer).
  • E · Results — NUMBERS REQUIRED (→ section 7) — the main result metric + figure (mandatory); before → after on 3–4 metrics (time, money, conversion, users, % growth); business consequences (revenue, funding, savings, speed, retention); what the client can now do that they couldn't before; any post-launch data (3 / 6 months).
  • F · Output (→ section 8) — the 4–6 key features/modules shipped; what the client values most.
  • G · Tech (→ section 9) — full stack (frontend, backend, infra, integrations, AI/data); why these technologies.
  • H · Quote (→ sections 4, 7, 10) — will the client give an official quote + name/role/logo; one sentence on why they'd recommend us; may the company be named publicly or anonymized (NDA).
  • I · Legal / media — what may NOT be published (numbers, names, screens under NDA); brand assets available (logo, team photo, product screens).

Step 2 — Find the spine

Before writing, state in one sentence: for whom this case sells what. Pick the ONE result that matters to the target reader (the next prospect, not the past client). Everything else supports it. If the project had many outcomes, lead with the one closest to money or risk.

Step 3 — Write (website format · 12 sections, strict order)

Each solution maps 1:1 to its challenge. One narrative — the reader flows pain → why us → what we did → what came out; each section sets up the next. Mark visual placeholders inline (they go to the designer).

  1. Title + headline metric — [what we did] + [main number]. Not "How we built X". Example: "Released MVP in 8 weeks → first 100 paying customers & $30K accelerator funding". Number is mandatory.
  2. At-a-glance snapshot [SNAPSHOT] — scannable panel right under the title: Industry · Location · Team size · Engagement model · Duration · Stack · Headline result.
  3. About the Client — 2–3 sentences: business, industry, market position, headcount. No water. Who they are and why to take them seriously.
  4. The Challenge — three sub-blocks:
    • Business Challenge — the business problem being solved.
    • Technical Challenge — the hardest thing technically.
    • Delivery Challenge — team / process / timeline challenges. Insert the first pull-quote about the pain here [QUOTE]. State the cost of inaction.
  5. Why us + Approach — why they chose us + how we worked: discovery → sprints → key milestones. Short timeline strip (Week 0 / 2 / 6 / 8). Removes "why trust you".
  6. The Solution → mapped 1:1 to Challenge — each solution explicitly names the challenge it closes:
    • Solution → Business Challenge: … [SCREEN]
    • Solution → Technical Challenge: … [DIAGRAM]
    • Solution → Delivery Challenge: … Features are woven in here by the way, not as a detached list. The technical steps live here too — there is no separate "Approach" section competing with this one. This is the largest section of the case (400–500 words).
  7. Results & Business Value — before → after with numbers [METRIC], not "built a dashboard". Table: Metric · Before · After. Then the second pull-quote about the result [QUOTE]. Timeframe explicit.
  8. Features Delivered — short list of what shipped (4–6). Output ≠ outcome — do not confuse with Results.
  9. Tech Stack [STACK] — frontend, backend, infra, integrations (logos/list).
  10. Full Testimonial [QUOTE] — 2–4 sentence quote + name, role, photo/logo. Strongest one is about ROI or trust; placed after the result.
  11. CTA — one action for a reader with the same problem, ends with a question. "Facing a similar challenge? → Book a call."
  12. SEO / meta — target keyword (in H1 + first 100 words), meta title ≤60 chars, meta description ≤155, alt tags on all visuals, internal links to adjacent cases/services.

Voice: dry, specific, zero superlatives.

Length is a gate, not a preference. Body 1500–2000 words. Under 1200 it is a project card, not a case study — go back to the SME for facts instead of shipping it. Per-section budget:

Section Words
Title + number 25
Snapshot 60
About the client 120
Challenge 250
Why us + approach 250
Solution 400–500
Results before → after 200
Features delivered 120
Tech stack 80
Testimonial 80
CTA 40

Two inline pull-quotes add ~60 more. Solution must be the largest section. If it is shorter than Challenge, the case describes the problem instead of the work. A 40-word section reads as underbaked; fill each one with real decisions and context, cut only water.

Weak-numbers mode

If the client cannot share hard numbers (NDA, no baseline measured):

  • Lead with the operational change: "from 3 tools and manual copy-paste to one pipeline", "first 5 SQLs from a channel that produced zero".
  • Use ranges or ratios agreed with the client ("~2× faster", "under 4 weeks").
  • Never dress qualitative claims as metrics.

Step 4 — Derivative formats

  • One-pager: Headline · 3 stat tiles · Problem (2 lines) · Solution (3 bullets) · Result (2 lines) · Quote · Contact. Fits one page.
  • LinkedIn handoff: 5–7 line summary (hook candidate + core numbers + one surprising detail). Explicitly mark it as INPUT for the post-writing process, not a finished post.

Step 5 — Designer brief (visual markers)

Every marker left in the draft is a visual to produce:

Marker What to make
[HERO] Hero visual on top: product in context + headline metric as a badge
[SNAPSHOT] At-a-glance panel (icons: industry / team / duration / stack)
[BEFORE/AFTER] Before/after comparison (metrics or UI)
[DIAGRAM] Architecture / flow / process
[SCREEN] Product screens in a device frame, clean, retina
[METRIC] Big number tiles for the Results section
[STACK] Technology logos in a row
[QUOTE] Pull-quote card with client photo/logo

Design requirements: agency brand colors/fonts, white background under screens; real project screens (blur/anonymize under NDA), not stock mocks; mobile contrast & readability (most traffic is phone); alt text on every visual; export WebP/optimized, lazy-load, ≤200KB where possible.

Step 6 — Pre-launch checklist

  • Number is in the headline
  • Snapshot panel filled
  • Each Challenge has a mapped Solution (1:1)
  • Results in before→after format with numbers
  • Body ≥1500 words and Solution is the largest section
  • 2+ pull-quotes + 1 full testimonial
  • CTA present, clickable, ends with a question
  • All [markers] replaced with final visuals
  • Legal approval on numbers / name / screens (especially NDA)
  • SEO: meta title/description, keyword, alt tags, internal links
  • Run through anticopywriting-ai (no AI trace, no negative parallelism)
  • Mobile check
  • Every number sourced or marked [[уточнити]] — placeholder list printed at the end
  • Remove any internal / references section from the public version