Jeff Bullas

Short answer: Yes — AI can write hooks that dramatically improve your odds of stopping the scroll, but it won’t do the whole job for you. The headline alone rarely wins; delivery, visuals and testing do.

Here’s a practical, step-by-step approach you can use today to generate and test attention-grabbing TikTok and Instagram Reels hooks using AI.

What you’ll need

• A clear audience (who are you trying to stop?)
• A one-line video idea or outcome (what will viewers get in 5–30 seconds?)
• An AI tool that can generate text (a chat-based model or similar)
• A simple spreadsheet or notes app to collect variations and test results

Step-by-step

1. Define the angle: Curiosity, shock, benefit, question, or command. Pick one per test.
2. Use this copy-paste prompt with the AI:

Copy-paste prompt: “You are a high-converting social copywriter. For a 15-second Instagram Reel/TikTok about [insert topic], write 12 opening hooks (each 3–8 words) organized by angle: 3 curiosity hooks, 3 shock hooks, 3 benefit hooks, and 3 question/command hooks. Keep them punchy, emotional, and easy to speak. Include one-word delivery notes (soft, urgent, playful) after each hook.”

How to use the output

• Pick 5 hooks that feel authentic to you.
• Record quick takes—same visuals, different hooks—so you isolate the hook’s effect.
• Track views, watch time, and completion rate for each. Keep the winner and iterate.

Example hooks (for a simple productivity tip video)

• Curiosity: “Two-minute habit that changes everything” (playful)
• Shock: “You’re doing mornings wrong” (urgent)
• Benefit: “Get an extra hour daily” (confident)
• Question: “Want less stress in 3 steps?” (soft)

Common mistakes & fixes

• Thinking the hook alone will win — fix: pair with strong first 3 seconds visual and voice energy.
• Using vague hooks — fix: make them specific and promise an outcome or mystery.
• Testing too few variations — fix: test 4–6 hooks per creative batch, not just one.

7-day action plan

1. Day 1: Define audience & 3 video ideas.
2. Day 2: Generate 12 hooks per idea with AI using the prompt above.
3. Day 3: Record 3–5 quick variations per idea.
4. Days 4–6: Post and gather metrics; focus on watch time and retention.
5. Day 7: Keep the top 2 hooks, refine and scale.

Closing reminder

Use AI to speed writing and widen your options, not to replace testing. The quickest wins come from simple experimentation: generate, record, measure, repeat. Keep hooks human, specific and easy to deliver on camera.

Short answer: Yes — AI can turn transcripts into long-form, SEO-friendly articles fast, but you must guide it and edit like a human. Do the heavy lifting where context, structure and authority matter.

Do / Do‑Not checklist

• Do clean and summarize the transcript before asking AI to write.
• Do define target keywords, audience and article purpose up front.
• Do add original insights, facts and CTA — don’t publish verbatim.
• Do‑not paste a raw transcript and expect perfect SEO copy without prompts or edits.
• Do‑not rely solely on AI for facts — verify sources.

What you’ll need

• A readable transcript (text file or pasted content).
• A primary keyword and 2–3 related keywords.
• Target audience and desired tone (e.g., friendly expert).
• An AI writing tool (chat model) and 15–30 minutes for human editing.

Step‑by‑step

1. Skim and edit the transcript: remove stutters, filler, and off-topic tangents.
2. Create a short summary (1–3 sentences) capturing the episode’s main idea.
3. Ask AI to produce an outline with H2/H3 headings optimized for your keyword.
4. Generate the draft section by section — keep prompts focused (one heading at a time).
5. Add a compelling intro, meta title and meta description that include the keyword.
6. Human edit: tighten prose, verify facts, add links, and insert images with alt text.
7. Publish and track performance; iterate based on real user metrics.

Copy-paste AI prompt (use as-is)

Convert the following cleaned transcript into a 1,200–1,400 word SEO article for small business owners over 40. Use a friendly expert tone, include the primary keyword “content repurposing for small business” in the title and twice in headings, and use related keywords: “repurpose content”, “podcast to blog”. Produce an outline with H2 and H3 headings, a 20–25 word meta description, a 10–12 word SEO title, and a 3-question FAQ at the end. Keep sentences short and actionable. Transcript: [paste cleaned transcript here]

Worked example (mini)

Transcript excerpt: “We talked about turning podcasts into blog posts — start with a summary, then build an outline…”

AI result (snippet): “How to Turn a Podcast into a Long-Form Blog Post (content repurposing for small business). Start with a 2-sentence episode summary, then create H2s for ‘Key Takeaways’, ‘Step-by-step Process’ and ‘Tools’. Expand each takeaway into 2–3 paragraphs, add examples and a closing CTA.”

Mistakes & fixes

• Problem: Article is just a cleaned transcript. Fix: Ask AI to rewrite into original paragraphs and add examples.
• Problem: Keyword stuffing. Fix: Use keyword naturally in title, 1–2 H2s and meta only.
• Problem: Factual errors. Fix: Verify and add citations or links during editing.

Practical action plan (next 30–60 minutes)

1. Pick one transcript (5–20 minutes to clean).
2. Run the copy-paste prompt above (5 minutes).
3. Edit the draft for clarity and facts (10–20 minutes).
4. Publish and monitor search and engagement over 2–4 weeks.

AI speeds the process, but your edits make it authoritative and human. Start small, measure, improve — then scale.

Quick win (5 minutes): Before you launch, paste your ad script, headline, and a short description of your thumbnail/opening frame into an AI and ask it to score fatigue risk and give three fixes for the first 3 seconds. Use the prompt below. You’ll have actionable tweaks in minutes.

Can AI predict ad fatigue? It can’t guarantee outcomes, but it can reliably estimate risk and spot early fatigue triggers: weak hooks, sameness to your past winners, high cognitive load, unclear CTA, or creative that will “wear out” fast at common frequencies. Think of AI as a pre-flight checklist + risk radar, not a crystal ball.

What you’ll need

• 3–10 past ads with basic metrics: CTR (day 1), conversion rate or key action rate, average frequency, reach, spend, negative feedback (hides, reports), and when performance dropped.
• Your new ad: script/captions, visual description or storyboard, headline, CTA, target audience, and platform (Meta, YouTube, TikTok, etc.).
• An AI assistant (any mainstream chat model) and a simple spreadsheet or notes doc.

Step-by-step: a pre-launch fatigue check

1. Snapshot your baseline. Note typical day-1 CTR and CVR for your account. This becomes the yardstick.
2. Define “fatigue” for you. A practical rule: fatigue starts when CTR drops 20–30% from day-1 and frequency passes 2–3 (feed) or 5–7 (in-stream), or when hides/negative feedback rise noticeably.
3. Teach the AI your context. Share your baseline metrics, audience, and platform. AI guidance is only as good as the context you give it.
4. Run five checks on the creative:
   • Hook clarity (0–10): Does the first 3 seconds state a problem or promise?
   • Novelty vs history (Low/Med/High): How similar is it to your past winners? High sameness = faster fatigue.
   • Cognitive load (Low/Med/High): Too much text, split attention, or clutter kills attention.
   • CTA clarity (0–10): One action, one benefit, visible and spoken.
   • Platform fit: Aspect ratio, pace, captions, UGC feel, safe zones.
5. Ask for a shelf-life estimate. Request a banded forecast: Short (3–5 days), Medium (6–14 days), Long (15+ days) under your typical frequency and spend. It’s directional, not a guarantee.
6. Variant stress test. Have AI generate 3–5 alternate hooks and first 3 seconds. Keep the offer; change the opening, visual pattern, and tone.
7. Rotation plan. Decide how you’ll swap hooks when CTR dips 20% from day 1 or negative feedback climbs.

Copy-paste AI prompt: Ad Fatigue Rater + Variant Generator

Paste this into your AI, then insert your details where shown.

“You are an Ad Fatigue Rater for [PLATFORM]. Use my historical baseline and creative to estimate fatigue risk and give me fixes I can implement before launch.

My baseline (typical): Day-1 CTR: [X%]. CVR or key action rate: [Y%]. Fatigue usually begins when CTR drops [20–30%] and frequency > [2–3 feed / 5–7 in-stream]. Audience: [describe]. Offer: [describe].

Past ads (summary): [List 3–10 ads with format, hook, CTR day-1, when performance dipped, common negatives].

New creative: Script/Captions: [paste]. Opening visual/frame: [describe]. Headline: [paste]. CTA: [paste].”

“Output in this format:

• Hook clarity (0–10) and why.
• Novelty vs my history (Low/Med/High) with a cosine-similarity style ‘sameness’ estimate (0–1) based on language and theme. Flag >0.8 as high sameness.
• Cognitive load (Low/Med/High) with specific elements causing overload.
• CTA clarity (0–10) and one-line rewrite.
• Platform fit checklist: pass/fail with fixes.
• Predicted shelf-life band: Short / Medium / Long. State assumptions: expected frequency, day-1 CTR range, and the trigger I should watch to rotate.
• Top 5 pre-launch fixes (first 3 seconds, visuals, captions) with exact wording and timestamps.
• Generate 5 alternate hooks (7–12 words each) and the matching first 3 seconds of visual direction.
• Risk note: specific factors most likely to cause fatigue in week 1.

Keep it concise and actionable.”

Insider trick: measure “novelty distance.” Ask the AI to compare your new script and opening frame to your last 5 top-spend creatives. If the language, promise, or visual motif is too similar, shelf-life shrinks. Aim for medium similarity: recognizably on-brand, but with a new pattern interrupt (new opener, angle, color scheme, or setting).

Example

Brand: Skincare DTC. Baseline day-1 CTR: 1.2%. Fatigue at frequency ≈2.2.

• AI flags high sameness (0.86) to a recent winner: same “derm says this” hook and bathroom setting.
• Cognitive load: Medium (on-screen text plus fast cuts).
• Shelf-life: Short-to-Medium. Assumes day-1 CTR 1.0–1.3%, rotate when CTR <0.9% or hides increase.
• Fixes: open with split-face visual, reduce text to 6–8 words, swap bathroom for outdoor light, CTA moved to 5s with on-screen button.
• Variants: “Your glow in 7 days? Watch this.” “Redness gone, makeup optional.” etc.

Mistakes to avoid (and quick fixes)

• Guessing without context. Fix: always feed baseline metrics and audience specifics.
• Over-indexing on generic best practices. Fix: compare to your own winners; sameness matters more than generic tips.
• Text-heavy openers. Fix: first 3 seconds = one bold visual + one-line hook, max 6–10 words.
• No rotation trigger. Fix: set automatic swaps when CTR drops 20–30% from day 1 or hides climb.
• One creative for every platform. Fix: adjust aspect ratio, pacing, captions, and tone to the channel.

What to expect

• AI will give you directional scores and concrete edits, not guarantees.
• Expect to improve first-3-seconds clarity, reduce clutter, and get 3–5 solid variants to rotate before fatigue hits.
• Biggest gains usually come from a new opening visual pattern and a crisper hook, not from changing the offer.

One-week action plan

1. Today (15 minutes): Run the prompt above on your next ad. Apply the top 3 fixes to the first 3 seconds and CTA.
2. Tomorrow: Feed 5 past ads into the AI, get your novelty distance notes, and build 3 alternate hooks.
3. Midweek: Prepare a rotation plan: when CTR dips 20–30% from day 1 or hides grow, swap to the next hook.
4. Before launch: Final pass for platform fit (safe zones, captions, aspect ratio). Save your scores and assumptions so you can learn next time.

Closing thought: AI won’t replace live testing, but it will stop preventable fatigue before you spend. Use it to sharpen the first 3 seconds, increase novelty, and plan rotations. That’s how you get more days of strong performance from every creative.

You’re asking the right question: using AI to estimate tax implications before you sell digital products globally is a smart, low-risk way to avoid costly surprises.

Here’s the short answer: AI can’t replace a tax professional or your filing software, but it can give you fast, practical estimates, a clear checklist, and the exact logic your checkout needs so you collect the right taxes in the right places.

Do / Don’t (quick wins)

• Do use AI to build a country-by-country tax matrix (rates, thresholds, B2B/B2C rules, filings).
• Do ask AI to produce if/then checkout logic and the data you must collect (e.g., customer country, VAT number).
• Do run revenue scenarios and get estimated monthly tax liabilities by jurisdiction.
• Do have AI draft registration and filing task lists with due dates and frequency.
• Don’t treat AI’s output as final tax advice—use it to brief a tax tool or professional.
• Don’t assume marketplaces or payment processors always handle tax for you—rules vary.
• Don’t ignore thresholds (EU OSS, UK VAT, US economic nexus by state, AU/NZ/CA/IN GST/HST).
• Don’t mix B2B and B2C without logic (reverse charge vs. consumer VAT/GST).

What you’ll need (gather this once)

• Your business location and legal entity type.
• Where you sell (website, app store, marketplace) and whether the platform collects tax.
• Product types (download, streaming, SaaS, course, subscription, bundle).
• Customer mix by country/state and B2B vs. B2C split.
• Average price points and 12-month sales per jurisdiction (actual or forecast).
• Existing tax registrations (VAT/GST/sales tax IDs) and filing frequencies.

Step-by-step: use AI to build your tax blueprint

1. Describe your business. Give AI the bullets above in plain English.
2. Get a tax matrix. Ask for a table per jurisdiction: tax type, threshold, digital-product rules, B2B/B2C treatment, estimated rate band, registration required, filing cadence, due dates, data to collect.
3. Add checkout logic. Have AI convert the matrix into if/then rules (e.g., “If EU B2C and product = e-service, charge VAT at customer country rate; collect two location evidences; show VAT on invoice.”)
4. Estimate liabilities. Provide your forecast by country/state. Ask AI to compute monthly/quarterly tax owed by jurisdiction and total.
5. Decide your tooling. Use AI to compare options like Stripe Tax, Paddle, Quaderno, TaxJar, or native marketplace collection vs. self-collection.
6. Create a playbook. Ask AI for a registration checklist, filing calendar, invoice requirements, and a monthly reconciliation process.
7. Review and confirm. Sanity-check with your accountant or the tax tool’s guidance before you go live.

Copy-paste prompt (robust, start here)

Paste this into your AI tool, replace the brackets, and run:

“Act as a tax research assistant for digital products. As of today’s date, build a country/state matrix for my business. Business: [country of incorporation], sells [digital product types] via [channels: own site/marketplace/app store], customers in [list countries/states], B2B share = [x%]. Last 12 months (or forecast) sales by jurisdiction: [list]. I need, per jurisdiction: 1) tax type (VAT/GST/sales tax), 2) whether my digital products are taxable and at typical rates (show a range if needed), 3) registration threshold and whether I’m over it, 4) B2B vs. B2C rules (e.g., reverse charge), 5) who collects tax (me vs. platform), 6) filing frequency and due dates, 7) required evidence and invoice notes, 8) risks/edge cases. Then produce: A) if/then checkout logic, B) a monthly tax liability estimate from my figures, C) a registration and filing task list. Clearly state ‘as of’ dates for the rules and flag any uncertainty.”

Worked example (how this looks in practice)

• Scenario: US creator sells a $29 downloadable course from her own site. Monthly sales: 300 units. Mix: 40% US, 30% EU, 10% UK, 10% AU, 10% CA. Mostly B2C.
• AI matrix output (abridged): Indicates EU B2C e-services VAT by destination with OSS option; UK VAT on digital services; AU GST registration likely if annualized revenue crosses threshold; CA GST/HST by province; US sales tax in states where economic nexus is met.
• Quick estimate method: AI multiplies revenue by each region’s typical digital tax rate where applicable, applies “no tax” where below threshold or B2B reverse charge, and sums a monthly total. Example math (illustrative only): EU revenue ≈ 0.30 × 300 × $29 ≈ $2,610; if average effective VAT across your EU mix were ~21%, estimated VAT ≈ $548. Repeat for UK, AU, CA, and qualifying US states to get a monthly total. AI will show its assumptions and an ‘as-of’ date.
• Checkout logic (sample): “If customer in EU and B2C, collect VAT at customer country rate; capture two location evidences; show VAT on invoice. If EU B2B and valid VAT number, apply reverse charge and display note. If UK B2C, collect UK VAT. If AU B2C and registered, collect GST. If US, charge sales tax only in nexus states.”

Insider tips that save time

• Ask AI to output three artifacts together: the matrix, the checkout logic, and a filing calendar—so your developer, bookkeeper, and you stay aligned.
• Have AI produce a “known unknowns” list (e.g., “Do you sell via marketplace X?”) to close gaps quickly.
• Build a pricing “tax buffer” (e.g., +1–3%) for regions with higher rates so your margin survives.
• Use AI to draft invoice notes and email copy that explains tax on digital services in plain language.

Common mistakes and quick fixes

• Mistake: Ignoring thresholds and registering everywhere. Fix: Have AI tag “register now” vs. “monitor” jurisdictions.
• Mistake: Treating B2B like B2C. Fix: Validate VAT/GST numbers and apply reverse charge where allowed.
• Mistake: Assuming marketplaces always collect. Fix: Map each channel’s responsibility in your matrix.
• Mistake: Using a single “average rate.” Fix: Use country/state-specific rates and show an ‘as-of’ date.
• Mistake: Missing evidence and invoice rules. Fix: Ask AI for a data checklist (two location proofs, invoice wording, retention).
• Mistake: No review cadence. Fix: Monthly 15-minute check: rates, thresholds, filings coming due.

Action plan (90 minutes to clarity)

• 15 min: Gather your inputs (products, prices, sales mix, channels).
• 30 min: Run the prompt above; ask AI for the matrix, logic, and estimates.
• 20 min: Iterate—add missing details, request examples, flag uncertainties.
• 15 min: Choose your collection method (processor’s tax tool vs. marketplace vs. manual).
• 10 min: Create a filing calendar and a monthly reconciliation checklist.

Bottom line: AI is excellent for scoping your obligations, estimating tax by region, and generating the exact rules and checklists you need. Use it to do the heavy lifting fast—then confirm the plan with your tax tool or advisor before you flip the switch.

Quick win: In the next 5 minutes, pick one persona and run this prompt to get 6 subject lines and 3 short cold-email templates you can test today.

Why this works: cold emails succeed when they speak to a specific person, not a crowd. AI helps you turn a clear persona + value into concise, human-first messages.

What you’ll need

• A clear persona: role, industry, key pain, decision power, and a line about their day.
• Your value proposition: one sentence that explains benefit (not feature).
• An AI chat tool (ChatGPT, Claude, etc.) or any AI writer you’re comfortable with.
• A spreadsheet or email tool to store variables (first name, company, metric).

Step-by-step

1. Define your persona in one paragraph. Example: “VP of Marketing at mid-size SaaS, wants predictable lead gen, time-poor, metrics-driven.”
2. Write a short value sentence: what you do and the measurable outcome. Example: “We help SaaS companies cut lead costs 30% in 90 days.”
3. Use the AI prompt below. Ask for subject lines, 3 short templates (cold, follow-up 1, follow-up 2), token placeholders ({{first_name}}, {{company}}), and a 1-line test metric suggestion.
4. Pick 2-3 variants, personalize first 2 lines manually using LinkedIn or the company site, then send a small A/B batch (20–50 emails each).
5. Measure opens, replies, and meetings. Iterate weekly—change subject, CTA, or personalization angle.

AI prompt (copy-paste):

“You are a professional copywriter experienced in B2B cold email. Persona: VP of Marketing at a mid-size SaaS company, pain: unpredictable lead flow and high CPL, decision driver: revenue growth and predictable pipeline. Our value: we reduce lead cost by 30% and increase qualified demos in 90 days. Produce: 6 subject lines, 3 short cold-email templates (cold, follow-up 1, follow-up 2), each 3–5 sentences, include personalization tokens {{first_name}}, {{company}}. Use a friendly, direct tone, and a single clear CTA (30-minute call or reply with interest). Also suggest a simple A/B test to run and one metric to track.”

Example output

Subject: “Cut SaaS lead costs by 30%—quick question”

Email: “Hi {{first_name}}, I noticed {{company}} is scaling its product marketing. We help mid-size SaaS cut lead costs 30% and double qualified demos in 90 days. Quick 15-minute call to share two examples from your space? If not, reply and I’ll send a short case study.”

Mistakes & fixes

• Too generic -> add one concrete pain or metric in first line.
• Too long -> keep cold email under 80–120 words.
• No CTA -> always ask for a small, specific next step.
• No personalization -> add one line referencing company news or role.

7-day action plan

1. Day 1: Define 1 persona and run the AI prompt.
2. Day 2: Personalize 2 variants and send 40 emails total.
3. Day 3–7: Track results, tweak subject/first line, send follow-ups.

Start small, measure, and iterate. AI speeds creation—your judgment and personalization win the replies.

Short answer: Yes—AI can flag creative elements likely to cause early fatigue and give a risk score before you launch. It won’t be perfect, but it can give fast, actionable guidance so you can tweak creatives and run smarter tests.

Why this matters: Ad fatigue wastes budget fast. If we can predict which ads will decay quickly, we can refresh creative earlier, allocate budget better, and improve ROI.

What you’ll need

• Historical ad performance (CTR, conversion, frequency, CPM) — even a few months is useful.
• Creative assets or screenshots and copy for each ad.
• Simple spreadsheet or basic analytics tool.
• An AI tool or service that can analyze text and images (many no-code platforms exist).

Do / Don’t checklist

• Do start small: test a handful of ads before scaling.
• Do combine AI insights with a short live test (5–7 days).
• Don’t trust a single AI score—use it to prioritize experiments.
• Don’t ignore audience signals like frequency and CTR trends.

Step-by-step: Quick method you can follow today

1. Gather 20–100 past ads in a spreadsheet: creative type, headline, CTR, conversion, frequency, lifespan.
2. Ask an AI to rate each creative on novelty, emotional intensity, clarity of CTA, and repetitiveness.
3. Use simple rules: low novelty + high frequency history = high fatigue risk.
4. Score new creatives with the same rubric to produce a “fatigue risk” (0–100).
5. Run prioritized small-scale A/B tests for high-risk ads to confirm and refine.

Practical AI prompt (copy-paste)

“You are an ad strategist. Rate the following ad creative on a 0–100 fatigue risk where 0 is unlikely to fatigue quickly and 100 is very likely. Consider novelty, clarity of message, emotional intensity, color/visual complexity, and likely repeat-viewer irritation. Explain 3 short reasons for the score and one quick suggestion to reduce fatigue.”

Worked example

Take a Facebook carousel ad with the same image repeated, headline: “Save 20% Today”. AI assesses: novelty low, emotional intensity low, CTA repetitive → fatigue risk 78. Fix: swap images, add user testimonial, and test headline variation. Run 7-day test; watch CTR and frequency. If CTR drops by 20% and frequency >3, refresh creative.

Mistakes & fixes

• Mistake: Relying only on AI scores. Fix: always validate with a short live test.
• Mistake: Ignoring audience segments. Fix: score per audience—what fatigues one group may not another.
• Mistake: Overcomplicating features. Fix: start with novelty, clarity, emotion, repetition.

Action plan (next 48 hours)

• Collect 20 past ads into a sheet.
• Run the copy-paste AI prompt on three new creatives.
• Pick one high-risk and one low-risk creative to A/B test for 7 days.

AI gives you a head start. Use it to spot risks, not to replace fast experiments. Small tests + quick creative tweaks = real, measurable wins.

Cheers, Jeff

Nice question — you’re right to ask whether AI can help estimate tax when selling digital products internationally. It’s a practical, high-impact problem and AI can give quick, useful estimates — but not legal certainty.

Quick context: tax on digital goods depends on where you’re based, where your buyer is, platform rules, and local VAT/GST/service tax rules. AI can crunch scenarios fast and show likely obligations, but you must verify with official guidance or an accountant before filing.

What you’ll need

• Your business country and tax registration details
• Buyer country (or countries) and volumes
• Product type (ebook, software, subscription, course)
• Price, currency, and platform fees
• Any marketplace handling payments (they may collect taxes)

Step-by-step: how to use AI to estimate tax implications

1. Gather the facts above for a typical sale and for monthly totals.
2. Pick an AI tool (chatbot or LLM API). Use a secure environment for sensitive info.
3. Run a clear prompt (copy-paste prompt below). Ask for assumptions and step calculations.
4. Review output: look for tax type (VAT/GST/sales tax), rate, place of supply rule, collection responsibility, and filing pathway (e.g., OSS/one-stop-shop).
5. Cross-check with one official tax site or a short accountant consult for each major market.
6. Automate: save the prompt and template results so monthly estimates are quick.

Copy-paste AI prompt (use and adapt)

“I sell digital product(s) from [Your Country]. I want an estimate of tax obligations for selling to buyers in [List of Countries]. For each country, state: 1) likely tax type (VAT/GST/sales tax), 2) expected rate or range, 3) who must collect (seller, marketplace), 4) registration thresholds if any, 5) filing mechanism (e.g., OSS) and 6) assumptions you used. Use conservative rounding and flag areas needing human review.”

Short example

Sell an online course from the UK to a customer in Germany for €50. AI might tell you: VAT applies, German rate ~19%, place of supply = customer’s location, you may need to collect VAT and report via VAT OSS if thresholds exceeded. Then verify with HMRC and a tax advisor.

Common mistakes & fixes

• Relying solely on AI: fix by validating with official docs/accountant.
• Missing marketplace rules: check payment provider’s tax handling.
• Currency and rounding errors: normalize to one currency and show totals.

Action plan (next 48 hours)

1. Run the prompt for your top 5 buyer countries.
2. Create a 1-page checklist for each market (what to collect, where to report).
3. Book a short call with an accountant to review one month’s sample.

Reminder: AI gives fast, practical estimates and helps prioritize work — but don’t skip human verification for filing and compliance.

Want cold emails that get replies — not deletes? Use AI to craft tight, persona-driven messages that feel human. Start small, test fast, and iterate.

Context: AI is a copy tool, not a magic bullet. Your job is to give it a clear brief about the persona and the outcome you want. Then refine and test.

What you’ll need

• A one-paragraph persona: role, company size, industry, common pain.
• Your value proposition in one sentence.
• A clear, low-friction CTA (calendar link, 15-min call, quick question).
• An email sending tool (Mailchimp, Gmail, Outreach, etc.) and basic tracking (open/reply).

Step-by-step

1. Define the persona. Write one paragraph: who they are, what frustrates them, and what success looks like.
2. Create a short brief for the AI (see prompt below).
3. Ask AI for 3 subject lines and 3 email variants (short: 40–80 words).
4. Pick the best two, personalize with a trigger (recent news, product usage, mutual connection).
5. Send an A/B test to small batches (50–100 each). Measure opens and replies over 3–5 days.
6. Iterate: keep what works, tweak what doesn’t, and scale slowly.

Copy-paste AI prompt (use as-is)

“Act as a professional cold-email writer. Target persona: [ROLE] at [COMPANY TYPE], size [EMPLOYEES], industry [INDUSTRY]. Their main pain: [PAIN]. Our offer: [ONE-SENTENCE VALUE PROPOSITION]. Tone: friendly, concise, and professional. Output: 3 subject lines (5–7 words each) and 3 email variants of 50–80 words. Each email should: 1) open with a personalized line, 2) state the benefit in one sentence, 3) include a low-friction CTA (15-min call or quick question). Also provide a 1-sentence preheader.”

Example

Persona: Head of Marketing, 50–200 employee B2B SaaS. Pain: low MQL→SQL conversion. Offer: a one-week audit that improves conversion by identifying 3 quick wins. Subject: “Quick audit — 3 conversion wins” Email:

Hi Jane — noticed your recent product launch. I help B2B SaaS teams find 3 quick conversion wins in one week without new tech. If you’re open, I can run a focused audit and share two immediate changes in a 15‑minute call. Does Thursday or Friday work?

Mistakes & fixes

• Too long: keep bodies under 80 words. Fix: remove company history and industry jargon.
• Generic claims: add a specific trigger or micro proof. Fix: reference a recent event or a short result (“drove 18% more SQLs”).
• Weak CTA: don’t ask for a big commitment. Fix: offer a 15-min call or a single question reply.

Action plan (next 7 days)

1. Day 1: Write 1–2 persona briefs.
2. Day 2: Generate email variants with the prompt above.
3. Day 3: Personalize 2 winners and prepare A/B lists.
4. Day 4: Send and track for 5 days.
5. Day 7: Review results and repeat the loop.

Closing reminder: Start with one persona, measure replies (not vanity opens), and tweak. AI speeds writing — your judgment makes it persuasive.