Jeff Bullas

Hook: Want AI to teach math step-by-step while forcing the student to think — not just handing over the answer? Use prompts that force scaffolding, checkpoints, and progressive hints.

Why this matters: Students learn when they struggle just enough. That means more retention, clearer diagnosis of misconceptions, and less tutoring time. Small prompt tweaks deliver big teaching wins.

What you’ll need

• Problem statement (clear and exact).
• Student level (e.g., middle school algebra, high-school calculus).
• Teaching style (Socratic, worked-example, exam-prep).
• Allowed hint count (usually 0–3).

Do / Do not checklist

• Do require numbered steps, short checks after each step, and practice questions.
• Do limit hints and make them progressively revealing.
• Do not allow the tutor to display the final numeric answer unless the student types a clear reveal phrase like SHOW ANSWER.
• Do not accept long, unfocused explanations—ask for concise steps (max 2 sentences each).

Step-by-step prompt recipe

1. Open role: “You are a patient math tutor for a [student_level] student.”
2. Structure demand: “List assumptions (1 sentence), then 4 numbered steps. After each step include one yes/no checkpoint question.”
3. Prevent leaks: “Do NOT reveal the final numeric answer unless the student types ‘SHOW ANSWER’. If you provide the final answer, restart and apologize.”
4. Hints: “Offer up to 3 hints labeled HINT 1/2/3. Each hint must be progressively more revealing.”
5. Finish: “End with two short practice problems using the same method but different numbers.”

Copy-paste prompt (robust)

“You are a patient math tutor for a [student_level] student. Given the problem: ‘[paste problem here]’. (1) State assumptions and definitions in one sentence. (2) Provide 4 clear numbered instructional steps that teach the method—after each step include a single yes/no checkpoint question for the student. (3) Do NOT calculate or reveal the final numeric result under any circumstance unless the student types ‘SHOW ANSWER’. If the final answer is given, apologize and restart. (4) Offer up to 3 hints labeled HINT 1/2/3; each hint must reveal progressively more. (5) End with two short practice problems using the same method but different numbers.”

Worked example

Use the prompt above with: “Solve for x: 2(x + 3) = 14”. The tutor should deliver assumptions, four concise steps (e.g., distribute or isolate), a yes/no checkpoint after each, three staged hints, and two practice problems—while withholding the final value until the student types SHOW ANSWER.

Mistakes & fixes

• If AI still reveals answers: add a strict penalty line — “If you reveal the answer, include the phrase: ‘I gave the answer by mistake’ and restart.”
• If steps are too long: ask “Max 2 sentences per step.”
• If hints leak too much: limit to 3 and require each to be shorter than the previous.

1-week quick action plan

1. Day 1: Run 5 problems with the baseline prompt; save outputs.
2. Day 2: Tweak phrasing (concise & penalty); re-run.
3. Day 3: Pilot with 3 learners; note hint and SHOW ANSWER usage.
4. Day 4: Short quiz to measure understanding.
5. Day 5–7: Iterate prompt variants (exam, young learners), finalize best version.

Closing reminder: Start small, measure one metric (hint usage or answer-reveal rate), and iterate. The prompt is your teaching script—tune it like you’d tune a lesson plan.

Your rights-first workflow and clean file naming are spot on. Let’s turn that into a simple, repeatable system you can run every week without friction.

The goal: Build a tiny “factory” that outputs licensable image and music sets with consistent rights, metadata, and formats—so reviewers say yes and buyers find you.

Set this up once (10–20 minutes)

• Folders: Stock > Images > YYYY-MM-DD_theme; Stock > Music > YYYY-MM-DD_mood-bpm
• Filenames: date_subject_v1_RF.ext (add _30s, _60s, _loop, _stems for music; add _1x1, _4x5, _16x9 for image ratios)
• IPTC template (images): Title, Description (benefit-focused), 8–15 Keywords, Creator (your brand), Copyright notice, Usage notes (e.g., “royalty-free, commercial use”), AI-generated flag if required
• ID3/metadata template (music): Title, Album (collection name), Artist (your brand), BPM, Key, Genre, Mood, Comments (license/usage notes), ISRC field empty unless you control it
• Rights Log (spreadsheet): Date, Tool/Model, Prompt, Settings/Seed, File names, License choice, T&C text+URL+timestamp ref, Notes on edits
• Listing text snippets: short license menu (below), usage examples, and a provenance note you can paste

7-minute preflight checklist (images and music)

• No faces, logos, brand shapes, or identifiable trademarks
• No “in the style of” living artists or named songs
• Polish: fix artifacts/clicks; images sharp at 100%; music peaks below -1 dBTP; preview around -14 LUFS
• Loop test for music: play start/end seam; add short fades if needed
• Metadata complete and buyer-focused; embed IPTC/ID3 where possible
• Rights Log updated with model/version and T&C snapshot
• Mark AI-generated if the platform requires it

Copy-paste prompts (refined)

• Image series (safe, licensable, multi-ratio): “Create a commercial stock background set on one theme: soft neutral studio textures with warm lighting, no text, no logos, no realistic faces, not in the style of any identifiable artist. Deliver 12 variants with subtle changes in texture and light. Provide three aspect ratios per variant: 1:1, 4:5, 16:9. Target 4000px on the long side, low noise, even exposure.”
• Music bundle (cuts, loop, stems): “Produce a calm corporate ambient piece at 95 BPM for commercial stock use. Instruments: soft synth pads, gentle guitar plucks, light percussion. Avoid recognizable melodies. Deliver: seamless 15s loop, 30s cut, 60s cut, 2-minute bed, and stems (drums, bass, keys, melody). Clean mix, warm reverb. Export WAV (44.1kHz) and MP3 previews.”
• Metadata writer: “You are a stock library keyworder. Based on this asset: [paste your description], produce 1 clear title (≤60 chars), a 2-sentence benefits description (where it fits: web banners/YouTube/presentations/etc.), and 12–15 buyer-oriented keywords. Exclude brand names, people, locations, and artist styles.”
• Listing copy (license menu): “Draft concise listing copy for a royalty-free stock [image/music] asset. Include: what it’s perfect for, what’s included (ratios or cuts/stems), license summary (non-exclusive, no resale as-is), and a short provenance note: ‘Created with an AI tool; no real persons or brands depicted; rights workflow and tool terms logged on [date].’ Keep it friendly, trustworthy, and under 120 words.”

Simple license menu you can paste into listings

• Standard RF: Non-exclusive, perpetual, worldwide, for web/social/presentations/YouTube. No resale or redistribution as a standalone file. No use in logos/trademarks.
• Extended RF: Includes print runs and paid ads (web/social). No resale as stock/templates or in merchandise without additional permission.
• Custom/sync: Pricing by media, territory, and term (contact first). Not available where platform rules forbid custom deals.

Example: one-image set that reviewers like

• Theme: “Warm neutral linen texture”
• Deliverables: 4 colorways (beige, sand, teal, slate) × 3 ratios (1:1, 4:5, 16:9) = 12 files
• Title example: “Warm Linen Texture Background (Neutral, Minimal)”
• Description example: “Soft, neutral fabric texture for websites, presentations, and social posts. Clean and logo-free; designed for versatile overlays and headlines.”
• Keywords: neutral background, linen texture, warm minimal, website banner, presentation slide, social media, subtle fabric, soft pattern, copy space, modern design, beige teal slate, commercial use

Quality bars to hit

• Images: 4000px long side, no halos/banding, even lighting, artifact-free at 100% zoom
• Music: peaks < -1 dBTP, noise/click-free, clean loop seam, WAV master + MP3 preview, clear BPM and key

Extra mistakes to avoid (and fast fixes)

• Exclusivity traps: If a library requires exclusivity, do not upload the same asset elsewhere. Fix: keep an “exclusivity” column in your log.
• Content ID/PRO conflicts: Many libraries don’t allow them. Fix: decide up front—either library-only or your own monetization path—and document it.
• Unclear bundles: Buyers guess what’s inside. Fix: list contents and counts (ratios, cuts, stems) in bullets.
• Too many lookalikes: Algorithm may down-rank duplicates. Fix: vary colorway, framing, texture density, or instrument mix by 10–20%.

Fast action plan (90-minute sprint)

1. Minutes 0–10: Duplicate the folder, filename, and metadata templates. Open your Rights Log.
2. Minutes 10–35: Run the Image Series prompt. Pick the best 4 variants; export 3 ratios each. Embed IPTC.
3. Minutes 35–65: Run the Music Bundle prompt. Check peaks/loop seam. Export WAV + MP3; label cuts and stems.
4. Minutes 65–80: Use the Metadata writer to generate titles/descriptions/keywords. Paste into files and listings.
5. Minutes 80–90: Update the Rights Log with T&C snapshot refs; upload to one marketplace. Mark AI-generated if required.

Provenance bundle (keep this zipped per project)

• PDF or screenshot of tool T&Cs + copied text and URL
• Your exact prompts/settings and model version
• Master files (TIFF/PNG for images; WAV masters and stems for music)
• Final deliverables, previews, and a text file with your license menu

Start with one tidy set, log it, and ship. The system—not a single hit—builds acceptance, rankings, and revenue.

Nice call on the two-cell check — that’s the fastest way to get a strong signal. I’ll add a compact, practical checklist and a worked example you can run in an afternoon to move from signal to confident next steps.

What you’ll need (quick)

• CSV with visitor_id, variant (A/B), outcome (0/1 or value), and 2–4 attributes (device, new/returning, source).
• Spreadsheet (Excel/Google Sheets) or any table tool you already know.
• 10–60 minutes to slice and run two validation checks.

Step-by-step — do this now

1. Clean: drop duplicates and rows with missing variant or outcome.
2. Compute: overall conversion rate for A and B (conversions / visitors).
3. Slice: compute the same rate and N within each attribute level (mobile/desktop, new/returning, source).
4. Flag: mark segments where absolute lift > overall lift × 2 and N ≥ 100.
5. Quick stats: in a sheet, use =Z.TEST or manual difference-in-proportions z formula to get a p-value for that subgroup.
6. Validate: check balance (variant distribution by segment) and instrumentation (event fires and assignment logs) for the flagged segment.

Worked example (copy numbers into your sheet)

• Overall: A=12% (n=20,000), B=14% (n=20,000) — absolute lift = 2%.
• Mobile: A=10% (n=8,000), B=16% (n=8,200) — mobile lift = 6% (3× overall), N OK.
• Action: focus on mobile hypotheses, run balance & instrumentation checks, then replicate on mobile-only sample (calculate required N for desired power—rule of thumb: for 25% relative lift at baseline 10% aim for ~6–8k per arm).

Do / Do not (quick checklist)

• Do prioritize segments with substantial N and large absolute lift.
• Do validate instrumentation before acting.
• Do not chase tiny subgroups or uncorrected multiple slices as if they’re proven causes.
• Do not skip a small replication if the subgroup N is near your minimum threshold.

Copy-paste AI prompt (use after you have counts & rates)

“You are a data-savvy product manager. Summary: overall A: 12% (n=20,000), B: 14% (n=20,000). Mobile users A: 10% (n=8,000), B: 16% (n=8,200). Desktop users A: 13% (n=12,000), B: 12% (n=11,800). Provide: 1) three prioritized causal hypotheses for the mobile lift, 2) two quick validation checks I can run in data (exact queries or spreadsheet formulas), 3) an A/B follow-up design to confirm causality for mobile-only (suggest sample sizes and stopping rule).”

Common mistakes & fixes

• Small-N false positives — fix: don’t take action until you replicate or N≥100–200 per arm in that segment.
• Multiple comparisons — fix: pre-specify top 2 hypotheses or adjust thresholds (Benjamini-Hochberg or simpler: require larger effect).
• Broken instrumentation — fix: check event-fire timestamps and variant assignment logs for that timeframe.

1-week action plan (practical)

1. Day 1: run slices, compute rates & Ns and flag candidates.
2. Day 2: run balance & instrumentation checks; feed summary to the AI prompt above.
3. Day 3–4: design mobile-only replication (or targeted experiment) with suggested N.
4. Day 5–7: run replication, monitor, and decide to scale or iterate.

Keep it small, fast, and repeatable — find signals, validate them, then scale. That’s the practical path from AI hints to confident causes.

Quick win (under 5 minutes): Paste this prompt into ChatGPT and ask for 3 headline variants and a 120-word landing blurb. Put the best headline on a one-page landing and link a simple payment button. You’ll have a testable page in minutes.

Why this matters: AI gets you from idea to test fast. But the only true signal of product–market fit is people who pay and keep coming back. Use AI to design tests, not to replace conversations or cash.

What you’ll need:

• AI access (ChatGPT or similar)
• One-page landing builder (Carrd, Squarespace, or similar)
• Email tool and a payment processor (Stripe/PayPal)
• Short survey or booking tool (Typeform/Calendly)

Step-by-step (do this):

1. Pick 2–3 audience segments to test (e.g., mid-career marketers; founder-operators; freelance designers).
2. Use AI to generate 3 headline variants, a 120-word outcome-focused landing blurb, and a 5-question presale survey.
3. Build a single landing page per segment with a clear CTA: paid pre-sale, refundable deposit, or join-waitlist-with-commitment.
4. Drive 100–300 visits per page (email your list, post in niche groups, run a small ad test).
5. Track conversions: visitor→opt-in, opt-in→paid, and interview 10–20 paid or interested people per segment.

Example output (copy-pasteable sample):

• Headlines: 1) “Weekly Growth Experiments That Move the Needle” 2) “Practical Playbooks for Mid‑Career Marketers” 3) “Run Small Tests. Get Big Results.”
• 120-word landing blurb: This short weekly newsletter delivers three actionable growth experiments, templates, and a 10-minute playbook you can run this week. Designed for mid-career marketers who need measurable wins, each issue includes step-by-step setup, expected impact, and tracking metrics. Limited to 100 founding members at an introductory price of $7/month or $60/year. Join the first cohort and get access to a private discussion thread and one group Q&A. No fluff — just tests you can run in a day and results you can measure.

One robust copy-paste AI prompt:

“You are a marketing copywriter. I run a paid weekly newsletter/short online course for [audience: e.g., mid-career marketers who want to run growth experiments]. Write: 1) three headline variants; 2) a 120-word landing page description that emphasizes outcomes and includes price and limited spots; 3) a 5-question pre-sale survey focused on pain, current solutions, willingness to pay, and ideal outcomes; 4) a 3-line follow-up email to send after sign-up asking to book a 15-minute call. Keep tone direct, credible, and non-salesy.”

Mistakes & fixes:

• Mistake: Relying on likes or AI praise. Fix: Require payment or a refundable deposit.
• Mistake: Overlong surveys. Fix: Ask 3–5 tight questions and invite a short call.
• Mistake: Testing one audience. Fix: Run 2–3 segments in parallel and compare conversion rates.

7-day action plan:

1. Day 1: Define segments, run the AI prompt, pick best headline and blurb.
2. Day 2: Build landing + payment flow; create presale survey and email sequence.
3. Day 3: Launch to your email list and post in 3 niche places; start a small ad test if you plan to scale.
4. Day 4–6: Drive traffic, run headline A/B tests, book calls with respondents.
5. Day 7: Analyze: conversion %, paid pre-sales, interview notes. Decide: iterate, pivot, or scale.

What to expect: In 2–4 weeks you’ll have clear signals: whether people will pay, common objections, and which segment is strongest. Use those signals — not AI vanity — to decide your next move.

Quick win: In 5 minutes create a neutral 4000×3000 image or a 10–15s music loop, export it, and save a one-line title plus three keywords. That tiny test proves the whole cycle: create → label → store.

Why this matters

AI makes content fast. The gap that kills sales is the business steps: clear metadata, compliant licensing, and a saved audit trail. Do those and you turn experiments into repeatable income.

What you’ll need

• An AI tool that permits commercial use (confirm and save the exact T&Cs).
• Basic editor: photo editor for images; a simple DAW or audio editor for music.
• Accounts on 1–3 marketplaces or your own storefront.
• A generation log: date, tool, prompt, settings and the license terms you relied on.

One small correction: don’t rely on a screenshot alone for the tool’s terms. Export or copy the T&C text, save the URL, and keep a timestamped screenshot or PDF. That combination is more defensible than a single image.

Step-by-step — Images

1. Pick demand: neutral backgrounds, simple textures, props without brand marks.
2. Prompt and generate 6–12 variants; pick 3 best.
3. Edit lightly: crop, color-correct, remove artifacts and any recognisable people or logos.
4. Add metadata: clear title, 8–15 keywords, short usage note (e.g., “royalty-free web banner”).
5. Upload with the license you choose and attach your generation log to your records.

Step-by-step — Music

1. Decide mood, tempo, length and use (loop, bed, sting).
2. Generate 4–8 short loops or stems; pick and combine the best parts.
3. Edit: normalise levels, check loudness (platform LUFS if required), export WAV and an MP3 preview.
4. Metadata: tempo, key, mood tags and permitted uses (e.g., sync, royalty-free).
5. Upload and save the generation log and chosen license details.

Copy-paste prompts

• Image prompt: “Create a high-resolution 4000x3000px seamless neutral studio background: soft warm light, subtle texture, desaturated teal and beige tones, minimal shadows, no text, no logos, no realistic faces, commercial-use allowed.”
• Music prompt: “Produce a 15-second instrumental loop, 95 BPM, mellow corporate ambient, soft synth pad, gentle guitar arpeggio, warm reverb, clean mix, royalty-free use. Export 44.1kHz WAV and MP3 preview.”

Common mistakes & fixes

• Bad metadata → poor discoverability. Fix: spend 5–10 minutes per asset on 8–15 targeted keywords.
• Only a screenshot of T&Cs → weak record. Fix: copy text, save URL and timestamped screenshot/PDF.
• Poor audio levels → rejections. Fix: normalise and set LUFS per platform guidance.

One-week action plan

1. Day 1: Run the 5-minute quick win; save prompt, settings and T&Cs (text+URL+timestamped screenshot).
2. Day 2–3: Produce 10 image variations, add metadata, upload to one stock site.
3. Day 4–5: Produce 10 music loops, export WAV+MP3, upload to one music library.
4. Day 6: Review acceptances; tweak prompts and edits.
5. Day 7: Set a weekly target (e.g., 5 assets/week) and track acceptance and revenue.

Reminder: Start small, log everything, and refine. The first few assets teach more than any theory—so make one today and learn fast.

Yes to the tiny loop. Your plan nails the essentials: small topic, active-recall cards, 15–20 minutes a day, track one metric. Let’s add a plug-and-play toolkit so you can move faster, make better cards, and keep your review load under control.

What you’ll need

• One SRS (Anki, Quizlet, or a simple Sheet/Notion table)
• Your notes (text, highlights, or voice-to-text)
• A conversational AI for the prompts below
• Optional: a 20-minute timer and a simple weekly tracker (date, time spent, retention %)

Step-by-step (practical and fast)

1. Capture: Pick one tight topic and list 5–15 facts or ideas. Keep each item to one idea.
2. Draft with AI (Pass 1): Use the primary prompt below to turn notes into active-recall Q/A cards.
3. Refine with AI (Pass 2): Run the refinement prompt to tighten wording, remove duplicates, and add mnemonics.
4. Import: Add to your SRS. Tag by topic and difficulty. Cap new cards at 10–20 per week.
5. Review daily: One 15–20 minute block. Mark ease honestly. If manual, start with 1, 3, 7, 16, 35 days.
6. Autopace weekly: Use the autopacer prompt with your simple stats (retention %, minutes/day) to set next week’s new-card cap and, if needed, longer/shorter intervals.
7. Trim monthly: Suspend or rewrite “leeches” (cards you miss ≥3 times). Merge or split any cards that still feel clunky.

Robust copy-paste AI prompt (use as-is)

“You are a tutor. Convert the notes below into 12 active-recall flashcards. For each card, output: 1) a concise recall question (no true/false, no multiple choice), 2) a one-sentence correct answer, 3) a short mnemonic if helpful, 4) a difficulty tag (easy/medium/hard), 5) suggest an initial interval in days (choose from 1, 3, 7). Keep language simple and precise. Avoid duplicates and make each card test one idea. Number the cards. Notes: [paste notes here].”

Refinement prompt (tightens and fixes)

“Review these cards for clarity and cognitive load. For each card: shorten the question to 12 words or fewer, keep the answer to one sentence, add or improve the mnemonic, and flag any recognition-style items to rewrite as recall. Combine duplicates; split any double-barrel questions into two cards. Return the improved list, numbered.”

Cloze prompt (great for facts, formulas, vocab)

“From the notes below, create 8 cloze-deletion cards. Each card should hide one key term or number only (use brackets like [term] to indicate the hidden part). Include a one-line hint under each card. Keep wording simple. Notes: [paste notes here].”

Weekly autopacer prompt (prevents overload)

“Here are my stats: retention on first review = [percent]%, average minutes/day = [minutes], number of new cards added last week = [count]. Suggest: 1) a new-card limit for next week, 2) whether to lengthen or shorten intervals slightly, and 3) which 3 cards to suspend or rewrite (if any). Keep it practical and brief.”

Insider trick: card types by purpose

• Cloze for stable facts: dates, definitions, steps, formulas.
• Q/A for understanding: “Why…?”, “How…?”, “What happens if…?”
• Scenario cards for real application: short situations that force a decision.

Example (project management basics)

• Q: What is the critical path? A: The longest sequence of dependent tasks that sets the minimum project duration. Mnemonic: “Longest chain = longest time” [Medium] (3 days)
• Q: How does a risk differ from an issue? A: A risk is uncertain; an issue is happening now. Mnemonic: “Risk maybe, issue is.” [Easy] (1 day)
• Cloze: The triple constraint balances [scope], [time], and [cost]. Hint: Iron triangle. [Easy] (1 day)
• Scenario: Stakeholder asks for extra features without time/budget change—what do you do? A: Start change control: assess impact, get approval, adjust plan. Mnemonic: “Ask, assess, approve.” [Medium] (3 days)

Quality checklist (use this before importing)

• One idea per card (atomic)
• Question forces recall (no cues, no lists on the front)
• Answer is one sentence or one number
• Simple words, no fluff; add an example if concept-heavy
• Tag by topic and difficulty; cap new cards

Common mistakes & quick fixes

• Making recognition cards: Fix by asking “What/Why/How” and hiding cues.
• Overloading cards: Split lists into separate clozes; one hidden item per card.
• Adding too many new cards: Keep it to 10–20/week. If retention < 70%, halve new cards.
• Never suspending leeches: Suspend after 3 lapses; rewrite as scenario or cloze.
• Ignoring weekly data: Use the autopacer prompt to tune your intake and intervals.

Lightweight spreadsheet setup (if you skip apps)

• Columns: Date Added, Question, Answer, Mnemonic, Difficulty, Next Review, Ease (1–4), Lapses, Tag.
• Start new cards at Next Review = today + 1 day; after a correct recall, bump to 3, 7, 16, 35 days.
• Use a filter on “Next Review = today or earlier” for your daily session.

7-day starter plan

1. Day 1: Pick one topic; list 10 items; run the primary prompt; import.
2. Day 2: 15–20 minutes of review; mark ease honestly; note minutes.
3. Day 3: Add 5 cloze cards for the most factual items.
4. Day 4: Rewrite any hard cards using the refinement prompt.
5. Day 5: Add 2–3 scenario cards for application.
6. Day 6: Clean up tags; suspend any leeches (≥3 misses).
7. Day 7: Run the autopacer prompt with your weekly stats; set next week’s new-card limit.

What to expect

• Session time: 15–20 minutes; setup adds 5–10 minutes in week 1.
• Week 1: 15–40 solid cards and a repeatable routine.
• Week 4: Stable daily reviews, higher retention, fewer “surprise” lapses.

Micro-action now

• Paste one paragraph into the primary prompt above and generate 8–12 cards.
• Import and do one 10-minute review today. Tomorrow, run the refinement prompt on any card that felt fuzzy.

Keep it small, keep it daily, and let the data steer your pace. That’s how spaced repetition compounds.

Spot on: testing one change at a time turns guesswork into signal. I’ll add a practical system you can run in 30 minutes to get stronger headlines fast — and the insider tweak that lifts conversions without keyword stuffing.

Why this works: people skim. Your headline has to tell a specific audience what outcome you create and show one credible proof — in 6–10 words or under ~120 characters for scannability.

What you need before you open AI:

• One audience (who you help)
• One outcome (what changes for them)
• One micro‑proof (tool, result type, niche credential)
• Preferred tone (direct, friendly, bold)

Insider trick: build a 3‑part “headline stack”

• Primary role + audience: who you are for whom
• Outcome verb + benefit: the result you drive
• Micro‑proof token: tool, niche, or soft proof

Format it with a dash or pipe for easy scanning. Example: “Email strategist for DTC — grow repeat sales | Klaviyo.”

Step‑by‑step to craft and test

1. Draft a 20‑word positioning line: audience + outcome + proof. Example: “Help subscription apps activate new users faster with onboarding audits using GA4 and UX heuristics.”
2. Generate options with patterns (prompt below). Ask for 8–12 variants across patterns and tones. Shortlist 2: one benefit‑led, one personality‑forward.
3. Compress for scan speed: keep 6–10 words or under ~120 characters. Remove filler (of, and, the), replace nouns with verbs (reduce, accelerate, cut).
4. Add one micro‑proof token: a tool, niche, or soft proof (“Klaviyo,” “Fintech,” “ex‑Agency,” “GA4”). One token only — the rest belongs in About/Skills.
5. Run two checks:
   • Read‑aloud test: can you say it in one breath and understand it instantly?
   • Search support: ensure 1 role keyword (e.g., “Copywriter”) and 1 niche keyword (e.g., “SaaS”). Put extra synonyms in About, not the headline.
6. Implement and measure: Run variant A for 14–21 days, then B. Track profile views, invites/messages, and replies (Upwork: job invites, messages, interview rate; LinkedIn: views, connection requests, reply rate).
7. Iterate: Keep the winner. Next test: audience vs benefit vs tone — change just one.

Headline pattern bank (copy these shapes)

• Role for Audience — Outcome | Proof
• Outcome for Audience — Role | Tool
• Turn X into Y for Audience — Role
• Audience, get Outcome — Role | Proof
• Fix Problem for Audience — Tool | Role
• Outcome without Pain — Role for Audience
• Role — Outcome in Niche | Tool

Copy‑paste AI prompt (generation)

“You are a headline editor. Create 12 LinkedIn/Upwork headline options, each under 120 characters, using these inputs: Role = [role], Audience = [audience], Outcome = [benefit], Proof/Tool = [proof], Tone = [tone]. Use these patterns: 1) Role for Audience — Outcome | Proof, 2) Outcome for Audience — Role | Tool, 3) Turn X into Y for Audience — Role, 4) Audience, get Outcome — Role | Proof, 5) Outcome without Pain — Role for Audience, 6) Role — Outcome in Niche | Tool. Make 8 benefit‑led and 4 personality‑forward. No buzzwords, no hard promises. Deliver as a numbered list.”

Copy‑paste AI prompt (refine and compress)

“Take headline #[number]. Produce 3 versions: 1) 80 chars max, 2) 100 chars max, 3) 120 chars max. Keep the same meaning, include exactly one micro‑proof token, remove filler words, keep verbs active.”

Worked examples

• Before: “UX Writer | Content Strategist | Fintech”After: “UX writer for fintech — cut user errors | UX audits”Why: audience named, outcome verb, one proof token.
• Before: “Google Ads Specialist | E‑commerce | Data‑Driven”After: “Google Ads for DTC — scale ROAS, tame CPA | PMAX”Why: clear who, clear result type, specific tool.

What to expect

• Fast ideation: 10–12 usable drafts in minutes.
• Two rounds of edits: first for clarity, second for tone.
• Early signals within 7–10 days; aim for a modest lift first (+10–20%), then compound with further tests.

Common mistakes and quick fixes

• Identity first, value second: swap to outcome first (e.g., “Reduce churn for SaaS — onboarding”)
• Stuffing keywords: move extra keywords to About/Skills; keep the headline clean.
• Vague superlatives (“expert,” “world‑class”): replace with a concrete outcome or tool.
• Over‑claims: use soft framing (“help reduce,” “aim to improve”) unless you have verifiable proof.
• Emoji overload: 0–1 tasteful symbol max; too many hurt readability and search.

30‑minute action plan (do this today)

1. Write your 20‑word positioning line (audience + outcome + proof).
2. Run the Generation prompt; shortlist 2 headlines (A = benefit‑led, B = personality).
3. Use the Refine prompt to get 80/100/120‑char versions; pick the clearest.
4. Update your headline (variant A). Capture baseline metrics for the past 14 days.
5. Promote lightly (10 targeted connection requests or 5 Upwork proposals) to create signal.
6. After 14–21 days, switch to variant B. Compare percentage changes and keep the winner.

Pro move: pair the headline with the first 80 characters of your About/Overview repeating the same audience + outcome. Consistency boosts trust and clicks.

Start small, measure cleanly, and let the data nudge your wording. One crisp line, tested well, can quietly raise reply rates — and that’s where the wins compound.

Let’s turn your plan into a working policy in under two hours. AI writes the first draft; you supply the facts and a few clear decisions. The result: a policy people will actually read — and a terms summary that protects you without scaring users off.

What you’ll bring to the table (15 minutes)

• Data by action: newsletter signup, purchase, contact form, analytics.
• Vendors: email tool, payment processor, analytics, hosting, form tool.
• Retention defaults you’re comfortable with.
• Business location, contact email, and who your site serves (US, EU, both).
• Whether you market to minors; whether data crosses borders.

Insider trick: Map data by user action first. It keeps the policy honest and simple because it mirrors how visitors actually use your site.

1. Create a 3-minute data map
   • Newsletter signup: name, email.
   • Purchase: name, email, address, payment via Stripe (card handled by Stripe).
   • Contact form: name, email, message.
   • Analytics: IP, device info, pages viewed, cookies.
2. Pick retention defaults
   • Newsletter: until you unsubscribe or we’re asked to delete.
   • Transactions: 7 years for tax/accounting records.
   • Analytics: 26 months, then aggregate or delete.
   • Support emails/forms: 24 months.

   Adjust to your reality. The key is to be explicit and consistent.

3. List vendors and purposes
   • Stripe: payment processing; we do not store full card details.
   • Google Analytics: site performance and usage measurement.
   • Email provider (e.g., Mailchimp/Brevo): newsletters and updates.
   • Hosting (e.g., Squarespace, Webflow, WordPress host): site delivery.

   Note if data leaves your region (e.g., US-based processing for EU users).

4. Generate the draft with AI using the prompt below, then paste in your real details.
5. Add a plain-English top summary and a mini-FAQ so 80% of visitors get answers in 30 seconds.
6. Create your cookie banner + preferences panel (copy below). Keep it short and action-oriented.
7. Publish and link in your footer as “Privacy” and “Terms.” Add anchors for each section for easy scanning.
8. Log changes and set a 6-month review reminder. Update when you add a new tool or collect new data.

Copy-paste AI prompt (US-focused; adapt region as needed)

“You are a privacy-savvy writer. Draft a clear, plain-language Privacy Policy and a 3-sentence Terms of Use for a small website. Business location: [Country/State]. Audience: [Regions served, e.g., US + EU]. The site collects: [list by action]. Vendors: [Stripe for payments, Google Analytics, Email provider, Hosting]. Include: 1) a one-sentence ‘What we collect and why’ at the very top, 2) how we use data, 3) retention periods per data type, 4) third-party sharing and subprocessors, 5) cookies (essential, analytics, marketing) and a simple cookie banner + preferences text, 6) user rights (access, correction, deletion; include Do Not Sell/Share for California if applicable), 7) contact details, 8) international transfers note, 9) a 3-question FAQ (opt out, retention, who to contact), 10) a short change log section. Keep headings short, tone friendly, and avoid legalese. Assume payments via Stripe (we do not store full card numbers). Output with clear section headings I can paste into a website.”

EU add-on line (if you serve EU/UK users)

“Include lawful bases (consent, contract, legitimate interests), controller contact, and how users can withdraw consent. Note cross-border transfers and standard safeguards. Keep it simple and human-readable.”

Example snippets you can reuse today

• Top one-sentence summary: We collect basic contact details and usage info to run this site, deliver purchases, and send updates you request; we share data only with trusted providers who help us operate (like payment and email services).
• Cookie banner (basic): We use cookies to run the site and measure what works. Choose Accept All or set preferences. You can change your choice anytime.
• Cookie preferences (options):
  • Essential: Always on — helps the site work.
  • Analytics: Helps us improve content and performance.
  • Marketing: Helps us show relevant offers.
• Terms summary (3 sentences): By using this site you agree to our terms. You may use our content for personal, lawful purposes; do not misuse or attempt to break the site. Digital product sales are final except where required by law; contact us if there’s an issue.
• Change log: March 2025 — added Stripe as payment processor; set analytics retention to 26 months.

High-value checks before you publish

• Readability: ask AI to “rewrite at 8th-grade reading level, shorter sentences, no jargon.”
• Specifics: replace every placeholder with real names, emails, and retention periods.
• Consistency: your cookie banner choices should match the policy (e.g., if you offer opt-out for analytics, make sure it works).
• Scope: if you don’t sell to children, say so. If you accept EU users, add lawful bases.

Common mistakes and quick fixes

• Vague retention: state exact periods; if unknown, say you review annually and remove data you no longer need.
• “We don’t share data” but using vendors: clarify you share data with service providers who act on your instructions.
• Legalese: ask AI to reduce reading level and convert passive voice to active.
• Cookie banner mismatch: align banner options and the policy’s cookie categories.

60-minute sprint plan

1. Minutes 0–10: Build the data-by-action list and retention defaults.
2. Minutes 10–20: List vendors and purposes; decide on regions served.
3. Minutes 20–40: Run the prompt, paste in details, and generate the draft + cookie copy.
4. Minutes 40–55: Replace placeholders; add the one-line summary and mini-FAQ.
5. Minutes 55–60: Publish, link in footer, add a change log, set a 6-month review reminder.

Expectation set: AI gets you to an 80–90% draft fast. For payments, cross-border transfers, or if you serve minors, consider a quick legal review. You’ll still save hours — and end up with a policy people actually understand.

Do this once, then treat it like a living page. Each new tool you add, update one line in the policy and the change log. Small updates now beat big headaches later.