Jeff Bullas

Turn every discovery call into a 5-minute scorecard you can trust. One template, one prompt, repeat. That’s how you get faster follow-ups, cleaner forecasts, and fewer “stuck” deals.

Do / Do not (quick checklist)

• Do lock one template for 2–4 weeks before changing anything.
• Do trim small talk and copy only the meaty parts of the notes.
• Do ask the AI for evidence lines and “missing info” questions.
• Do set score thresholds (≥75 propose, 50–74 nurture, <50 disqualify/revisit).
• Do add a one-line human check for extreme scores (≥90 or ≤30).
• Don’t let the AI guess names, budgets, or timelines—return “Unknown” if not stated.
• Don’t keep tweaking weights daily—review weekly.
• Don’t paste entire transcripts—cut to pain, money, timeline, decision-maker, risks.

What you’ll need

• Cleaned call notes or transcript snippet (5–10 key paragraphs or bullet points).
• Any AI chat you already use.
• Your CRM fields or a shared doc with the same field names every time.

Step-by-step (simple and repeatable)

1. Right after the call (within 60 minutes), paste cleaned notes into your AI chat.
2. Run the prompt below. It returns a clear summary, score, evidence lines, and next steps.
3. Scan for 30–60 seconds. If a field looks off, edit once. Add a one-line human confirmation on extremes.
4. Paste the fields into your CRM. Apply your threshold rule and act immediately.
5. End of week: review 8–10 scored calls vs outcomes. Adjust weights once, then lock for two weeks.

Premium trick (insider): score anchoring + evidence lines

• Add 2–3 tiny “example deals” and their scores inside the prompt. This anchors the AI’s scoring to your reality.
• Require 2–3 verbatim lines from your notes that drove the score. This kills hallucination and speeds your review.

Copy‑paste AI prompt (use as-is)

“You are a sales ops assistant. Convert discovery call notes into a structured record and an objective score. Use only what’s in the notes—if info is missing, return ‘Unknown’. Return the following fields: (1) one_sentence_summary, (2) pain_points (3 bullets), (3) impact_signal (what it costs them or delays, if present), (4) budget_estimate (Low/Medium/High/Unknown), (5) decision_timeline (Immediate/1–3 months/3–6 months/6+ months/Unknown), (6) decision_makers (names/roles if stated; else Unknown), (7) competitors_mentioned, (8) risks_or_red_flags, (9) next_steps (2–3 bullets), (10) qualification_score (0–100) with a one-line justification, (11) confidence (High/Medium/Low), (12) evidence_lines (2–3 verbatim lines from the notes), (13) missing_info_questions (3 short questions to close gaps). Scoring weights: pain severity 30%, budget clarity 25%, decision timeline 20%, decision-maker involvement 15%, competition risk 10% (higher risk lowers score). Guardrails: do not infer; if not explicit, return ‘Unknown’. Calibration examples: Example A (Score 88): Strong pain with quantified impact, clear budget range, timeline <90 days, decision-maker present, low competition. Example B (Score 65): Clear pain, budget unclear, 3–6 month timeline, influencer not DM, one competitor. Example C (Score 35): Vague pain, no budget, 6+ months, no DM, active incumbent. Now analyze these notes: [PASTE NOTES HERE]”

Worked example

Sample notes you might paste: “Acme Mfg (250 employees). ERP outages ~8 hrs/month; estimate $15–20k loss per outage. Current vendor: ‘homegrown system’ + manual spreadsheets. Considering Vendor X. Budget ‘approved up to 60–80k if ROI clear’. Decision-makers: CFO (Sara) + Ops Director (Luis). Timeline: target Q2 go-live; want pilot in 8–10 weeks. Needs: reduce downtime, inventory accuracy, light integrations to QuickBooks. Risks: IT bandwidth thin; CFO wants 3 references. Next step: send ROI case study and schedule pilot scope call next Tuesday.”

• Summary: Mid-sized manufacturer with costly ERP downtime seeks pilot in 8–10 weeks; budget likely sufficient if ROI proven.
• Pain points: Downtime losses; manual spreadsheets; inventory inaccuracies.
• Impact signal: ~$15–20k per outage; recurring monthly.
• Budget: High
• Timeline: 1–3 months
• Decision-makers: CFO (Sara), Ops Director (Luis)
• Competitors: Vendor X
• Risks: IT capacity; reference requirement
• Next steps: Send ROI case; book pilot scope; prep two relevant references
• Score: 82/100 — strong pain + budget + short timeline + DMs engaged; moderate competition and IT risk
• Confidence: High
• Evidence lines: “outages ~8 hrs/month…$15–20k loss”; “approved up to 60–80k”; “pilot in 8–10 weeks”
• Missing info questions: Who signs the contract? What integration scope is must-have? What success metric ends the pilot?

Common mistakes & quick fixes

• Messy input: Remove pleasantries, jokes, and unrelated side stories. Keep the buyer’s words on pain, money, time, people.
• Score drift: If average scores creep up or down week to week, re-run the same 3 calibration examples in your prompt.
• Invisible risks: Add a field for “risks_or_red_flags” so they don’t get buried in the summary.
• Inconsistent next steps: Ask the AI for 2–3 concrete actions with owners and timing; keep them short and specific.

What to expect

• Time per note: 8–15 minutes at first; down to 3–5 minutes once the template is muscle memory.
• Decision clarity: you’ll triage calls in seconds and stop over-nurturing low-fit deals.
• Quality control: evidence lines make review fast and reduce edits.

One-week action plan

1. Today: Paste your last two calls into the prompt. Save outputs in your CRM under a “Discovery (AI)” section.
2. Tomorrow: Add your 3 calibration examples to the prompt (one high, one mid, one low-fit).
3. Midweek: Run 5 live calls through the flow; apply the 75/50 thresholds.
4. End of week: Compare scores vs outcomes. Adjust one thing only (weights or threshold). Lock for two weeks.

Final nudge: Consistency beats cleverness. One template. One prompt. Five minutes after every call. That’s the system that compounds.

Quick answer: Yes — AI can help you design a logo that reduces trademark risk, but you must use it wisely. Think of AI as a fast sketch tool, then do the human homework to avoid costly conflicts.

Why this matters

Logos can infringe trademarks if they’re confusingly similar to existing marks. AI can produce original-looking designs quickly, but it won’t reliably catch legal conflicts or common-law uses. Combine AI with simple checks and one legal review for a practical, low-cost workflow.

What you’ll need

• A clear brand name or descriptor (even a short list of options)
• An AI image or logo generator (user-friendly tool)
• Access to basic trademark search tools (web search, USPTO TESS, social platforms)
• A place to save versions and dates (folder or cloud drive)
• Budget for a trademark attorney review before launch

Step-by-step workflow

1. Define your brand traits: 3 words (e.g., friendly, premium, local).
2. Use AI to generate multiple, clearly different directions. Ask for simplicity and uniqueness.
3. Run visual and name searches: reverse-image search on the best outputs, search USPTO TESS for similar text marks, and scan major social platforms for names/handles.
4. Refine the favorite designs: tweak shapes, fonts, colors and especially the name element to make it more distinctive.
5. Document dates and versions. Save originals and edits for provenance.
6. Before public use, get a trademark attorney to conduct a clearance opinion.

AI prompt you can copy-paste

Create 6 original logo concepts for a boutique brand named “Morning Bloom.” Do not use or reference existing coffee chain logos, famous marks, or common stock icons. Focus on unique geometric shapes, a custom monogram option, and a warm color palette (terracotta, cream, olive). Provide both a full logo and a simple mark for social profile use, all scalable for vector output. Include a one-sentence distinctiveness note for each concept.

Example

AI produces 6 options. You run a reverse-image search and find one is visually similar to a small local chain. You discard that option, tweak the monogram on two others, then run a USPTO search on the names. Two names are clear; one needs a lawyer check before use.

Common mistakes & fixes

• Assuming “new” = “safe” — Fix: always search for similar visuals and names.
• Using famous style cues (font/shape) — Fix: choose unique fonts and alter key shapes.
• Skipping documentation — Fix: save dated versions to show creative timeline.

Simple action plan (next 7 days)

1. Day 1: Write brand brief (3 words + name options).
2. Days 2–3: Run AI prompt and gather 6–12 concepts.
3. Days 4–5: Do image and name searches; remove risky options.
4. Day 6: Refine top 2 designs and save files with dates.
5. Day 7: Book a short attorney review for clearance.

Closing reminder

Use AI for speed, not as the final legal decision-maker. A little research plus one legal check prevents big headaches and helps you launch with confidence.

Sell smarter, not harder: use simple data + an AI prompt to set resale prices that cover costs, win buyers, and make a profit.

Across eBay, Poshmark and Facebook Marketplace the rules differ — fees, buyer behaviour and shipping expectations change. This guide gives a quick, practical workflow you can run today, even if you’re non-technical.

What you’ll need

• Smartphone with good photos of the item.
• Notes: purchase cost, item condition, and preferred profit margin.
• Quick access to recent sold prices (search “sold listings” on eBay or look at similar posts).
• Calculator or simple spreadsheet (Google Sheets/Excel).
• An AI tool (Chat-style assistant) to suggest price ranges and listing copy.

Step-by-step (do this now)

1. Gather 5–10 sold comps for the same or similar item and note low/high/median prices.
2. Estimate costs: purchase cost + shipping estimate + marketplace fees (eBay ~10–12%, Poshmark ~20%, FB local usually 0–10% if shipping).
3. Calculate a baseline price: target_price = (cost + shipping) / (1 – fee_rate – desired_margin). Use this as a starting point.
4. Ask the AI for a recommended price range, suggested title, and 3 listing bullets tailored to each platform.
5. List at the recommended price. Monitor views and offers for 48–72 hours. If slow, try a modest price drop or featured upgrade.

Example

Vintage leather jacket: purchase $20, shipping $8, target margin 40%, platform fee 12%.
Baseline: (20+8)/(1-0.12-0.40)=28/0.48≈$58. Start listing at $75 (room to negotiate), expect fees ~$9, net profit ≈$38.

Common mistakes & fixes

• Mistake: Ignoring fees. Fix: Always subtract marketplace fees before deciding profit.
• Mistake: One price fits all. Fix: Price for each platform’s audience and fees.
• Mistake: Poor photos and vague titles. Fix: Use 6 clear photos and include brand, size, condition in the title.

Copy-paste AI prompt (use this)

You are an ecommerce pricing assistant. I have a [item name], condition: [new/good/fair], purchase cost $[X], shipping estimate $[Y]. Recent sold listings show low $[A] high $[B]. Marketplaces: eBay (fees 12%), Poshmark (fees 20%), Facebook Marketplace (local seller, fees 0–5%). Suggest a recommended listing price range for each marketplace, a 1-line title and 3 bullet points highlighting value, plus two discount strategies (percent or time-based). Explain your reasoning briefly.

Action plan (next 48 hours)

1. Collect 5 sold comps and take 6 photos.
2. Run the AI prompt above with your numbers.
3. List one item on each platform using the AI title/bullets.
4. Monitor for 72 hours, adjust price by 10% if necessary.
5. Log results and repeat — treat it like a small experiment.

Small experiments win. Use data, confirm with real sales, and let AI speed the thinking — but always validate with real-world results.

Make it always-on without being always on-call: lock in a calm, self-improving monitor that catches the real fires, bundles the rest into a neat daily brief, and learns what you actually care about.

What you’ll need (simple stack):

• Keywords list (10 core + 5–10 exclusions).
• Three sources to start (one social, one news/RSS, one forum/review).
• An automation connector (any tool that moves data into a sheet or dashboard).
• An AI model for classification and summaries.
• One shared log (sheet) and clear owners for Support, PR, Sales, and Legal/Safety.

Build the resilient loop (8 steps):

1. Collect cleanly: run your queries with exclusions; capture timestamp, source, snippet, URL, language, author_followers, engagement_count, keyword_matched.
2. Deduplicate: create a unique_id (URL+timestamp hash) and ignore near-duplicates (same text ±10 characters or same URL).
3. Classify with topic-first logic: ask AI to assign a topic, then sentiment, then urgency with a reason. This reduces sarcasm mistakes.
4. Score priority: Priority = (Urgency×0.5) + (Influence×0.3) + (Engagement×0.2). Multiply by 1.5 if topic is Legal/Safety (cap at 100).
5. Route smartly: Priority ≥70 or Legal/Safety → instant alert to the right owner; everything else → daily digest to reviewer.
6. Quiet hours: 10pm–7am only alert if Priority ≥85 or a spike is detected (≥5 negative mentions in 15 minutes).
7. Summarize threads: when 5+ mentions share the same URL/headline, create one summary card and mute duplicates.
8. Learn weekly: drop noisy keywords, add temporary exclusions, adjust thresholds ±10 based on misses and over-alerts.

Copy-paste prompts (battle-tested):

• Classifier (v2, JSON-only):

“You are a brand monitoring assistant. Input is one mention with fields: text, source, language, author_followers, engagement_count, url, timestamp. Output a single JSON object with: sentiment (positive|neutral|negative), topic (Product issue|PR/News|Influencer mention|Customer support|Legal/Safety|Sales lead|Off-topic), urgency (0–100) and reason (≤20 words), influence_tier (Nano <10k|Mid 10–100k|Major >100k), legal_safety_flag (true/false), off_topic (true/false), short_summary (≤25 words), recommended_action (reply|escalate|archive) with one-sentence note, confidence (0–1), suggested_priority (0–100) computed as: Priority = (Urgency×0.5) + (Influence score×0.3) + (Engagement band×0.2), where Influence score = 30/60/100 for Nano/Mid/Major and Engagement band = 0 (<5), 40 (5–49), 80 (50+). If topic is Legal/Safety, multiply Priority by 1.5 and cap at 100. Return valid JSON only.”

• Daily digest summarizer:

“You are an analyst. Given today’s classified mentions (JSON array), produce: 1) Top 5 items by suggested_priority with one-line summaries and owner (PR/Support/Sales/Legal), 2) Three bullet insights (trends, risks, opportunities), 3) Suggested changes: new exclusions, keywords to add, threshold tweak (±10), 4) Draft 2 reply templates for the most actionable item. Keep the whole brief under 250 words.”

• Spike detector (calm nights):

“You detect anomalies. Given mentions from the last 60 minutes and a baseline average per hour for the past 7 days, decide if there’s a spike. Criteria: volume ≥2× baseline OR ≥5 negative mentions OR any Legal/Safety. Output JSON: spike (true/false), reason (≤15 words), examples (up to 3 URLs), recommended_action (notify now|wait for digest).”

• Reply template generator:

“You craft public replies. Input: one mention JSON + brand voice (friendly, concise, solution-first) + constraints (no promises, no personal data). Output three variants: 1) Public reply (≤40 words), 2) DM opener (≤35 words), 3) Escalation note for internal team (≤40 words).”

Insider upgrades that compound:

• Two-tier AI: use a low-cost model for first-pass topic/off-topic. Send only Priority ≥40 to a higher-accuracy model. This cuts cost and noise.
• Dynamic exclusions: every false positive adds a 14-day exclusion; auto-expire unless it blocked a real mention.
• Keyword expansion: once a week, ask AI to propose 5 co-mentions (competitor names, product nicknames) from your digest; test them in the digest only for 7 days.
• Language gate: detect language pre-classification; translate only if market-relevant and confidence ≥0.9.

Mini example (from alert to action):

• Post: “Acme’s new update breaks logins. Paying customers locked out.” Author 88k, 63 comments.
• Classifier → topic Customer support, sentiment negative, urgency 82 (login failure + high engagement), influence Major, priority 87, action escalate.
• Routing → instant alert to Support; PR is cc’d if three similar mentions appear within 30 minutes.
• Reply template → short public apology + DM link; internal note auto-fills ticket with URL, device, version if present.

Common mistakes and quick fixes:

• Too many instant alerts → raise threshold to 75 and require influence ≥Mid at night.
• Missing quiet crises on forums → add one niche forum feed and include it in spike detection even if it’s not in instant alerts.
• Classifier overconfidence → if confidence <0.7 and sentiment negative, require human review before public reply.
• Data hoarding → store only public snippets, links, counts, and derived labels; avoid personal data.

7-day upgrade plan:

1. Day 1: Finalize 10 core keywords + 8 exclusions; map topics → owners.
2. Day 2: Wire 3 sources; normalize fields; enable dedupe.
3. Day 3: Add classifier prompt; test on 50 past mentions; adjust taxonomy if needed.
4. Day 4: Implement priority scoring, routing, and quiet-hours + spike detector.
5. Day 5: Go live; send instant alerts only for Priority ≥70; reviewer tunes exclusions.
6. Day 6: Add digest summarizer; ship two reply templates for top issues.
7. Day 7: Review metrics (false positives, response time, useful escalations); tweak thresholds ±10; enable two-tier AI if costs/noise are high.

Expectation check: Week 1 will be busy. By Week 2, you should see false positives under ~30%, urgent response inside an hour, and a one-screen digest that actually drives action.

Bottom line: keep the layers, let the AI triage, cap alerts at night, and teach the system weekly. That’s how you get continuous brand monitoring that stays quiet until it truly matters.

Quick win: You’re one small tweak away from fast, trustworthy retrieval. Let’s lock it in with a simple, do-first checklist and a worked example you can copy.

Do / Don’t (read this first)

• Do split long docs into 200–500 word chunks with 20–30% overlap.
• Do store ID, source, date, category, and embedding version with every vector.
• Do normalize vectors before cosine search; keep k small (5–10) and re-rank the final list.
• Do use metadata filters (date, category) to shrink the search space.
• Do cache hot queries (top 100–500) and precompute their results.
• Do track latency (p50, p95) and Recall@k on a small test set.
• Don’t rebuild the entire index for small updates; batch them.
• Don’t mix similarity types (cosine vs dot) without adjusting scores.
• Don’t skip evaluation; set thresholds so low-quality matches are filtered out.
• Don’t upgrade to a managed vector DB until your metrics (latency/recall) tell you to.

What you’ll need

• A small corpus (start with 500–2,000 chunks) with stable IDs and basic metadata.
• An embedding model or service; a simple batch script to compute vectors.
• One index choice to start: Annoy/FAISS locally, Postgres+pgvector, or a managed vector DB.

Fast setup path (step-by-step with expectations)

1. Prepare content: Chunk 200–500 words, 20–30% overlap. Expect 2–5x more chunks than original docs.
2. Compute embeddings: Batch in 100–1,000 chunks. Save vector + ID + metadata + model name + vector dim. Expect a few minutes for a few thousand chunks.
3. Normalize + optional compress: Normalize for cosine. If needed, try PCA to 128–256 dims for speed/storage; expect a small nuance trade-off.
4. Build the index:
   • Annoy/FAISS HNSW (0–50k vectors): sub-100ms queries on a laptop; increase trees/efSearch for better recall.
   • pgvector (up to ~100k rows): simple SQL + joins; a few hundred ms typical; great when you already use Postgres.
   • Managed DB (millions): minimal ops, higher cost; scale when p95 latency or recall slips.
5. Query flow (keep it tight): Compute query embedding → apply metadata filters → retrieve top‑k (k=5–10) → re‑rank those candidates by exact cosine → optionally apply a keyword boost (hybrid) → return IDs, scores, and a short reason.
6. Thresholds & fallback: If top score < 0.25, return “no strong match” or fall back to keyword search.

Worked example: pgvector mini-build (beginner-friendly)

1. Table: text, doc_id, chunk_id, title, date, category, embed_version, vector (dim=768).
2. Load: upsert rows in batches of 500–1,000; keep an index on (category, date).
3. Search: filter by category/date first, then ORDER BY vector <-> query LIMIT 10.
4. Re‑rank: take those 10, recompute exact cosine in memory, sort, and return top 5 with scores and snippets.
5. Expectations: For ~50k chunks, p50 ~100–250ms, p95 a few hundred ms, depending on hardware and filters.

Insider tips that save hours

• Two-stage always wins: fast approximate search for candidates, then exact cosine re‑rank. Clean, cheap accuracy boost.
• Hybrid helps when queries have names/numbers. Combine keyword (BM25) with vectors using a simple weighted score or reciprocal-rank fusion.
• Score calibration: Sample 50 “no match” queries, record top scores, and set your “no-answer” threshold just above their 95th percentile.
• Version vectors: add embed_version so you can re-embed later without breaking searches.
• Cache the obvious: store the last 500 query results and precompute answers for FAQs or dashboards.

Common mistakes & quick fixes

• Unnormalized vectors → normalize all vectors when using cosine.
• Rebuilding on every edit → batch updates hourly/daily; rebuild only after big changes.
• No metadata filters → index and filter by date/category to cut latency and noise.
• Too many dimensions → try 128–256 with PCA if speed/storage matters.
• One-size-fits-all k → start at k=8 and adjust based on Recall@k and latency.

Copy-paste prompts (ready to use)

• Retrieval + thresholding: “You are a retrieval assistant. Given a user query, a function to compute a query embedding, and a list of documents with embeddings, IDs, and metadata, do the following: (1) Compute the query embedding. (2) Filter candidates to those matching any provided metadata (category, date range). (3) Return the top 8 by cosine similarity with scores. (4) Re‑rank those 8 with exact cosine and return the top 5 with a one‑line reason and the matching snippet. (5) If the best score is below 0.25, respond: ‘No strong match found’ and include the top 3 keywords for follow‑up.”
• Build a tiny evaluation set: “Act as a data annotator. Given the following document titles and summaries, generate 20 realistic user questions and list the most likely doc_id for each. Keep questions short and varied. Output a JSON list of {question, doc_id} for offline testing.”

Scale triggers (move up when these happen)

• p95 latency consistently > 400ms on pgvector even after filtering and caching.
• Recall@5 < target after you increased k and tuned chunk size.
• Index updates block normal queries or take many minutes during business hours.

7‑day action plan (tight and achievable)

1. Day 1: Choose index (Annoy/FAISS for prototypes; pgvector if you already use Postgres). Define k, threshold, and metrics.
2. Day 2: Chunk content and compute embeddings for 1,000–2,000 chunks. Normalize and store with metadata + embed_version.
3. Day 3: Build the index, wire metadata filters, implement two‑stage search, add caching.
4. Day 4: Create a 20‑question eval set. Measure p50/p95 latency and Recall@5. Set your threshold.
5. Day 5: Tune: trees/efSearch (Annoy/HNSW) or indexes (pgvector). Try PCA 256 if slow or large.
6. Day 6: Add hybrid scoring for named entities and numeric queries. Validate no‑answer behavior.
7. Day 7: Review metrics. If p95 and Recall@5 meet targets, ship. If not, increase k a little, improve filters, or plan a move to managed vectors.

Final nudge: Keep it simple, measure weekly, and change one variable at a time. Fast retrieval isn’t magic — it’s clean chunks, normalized vectors, small k, smart filters, and a tiny re‑rank.

5-minute quick win: Grab 10 real customer quotes from one competitor (reviews or comments), paste them into an AI chat, and ask it to cluster complaints and suggest one tiny paid offer that fixes the biggest pain in under 14 days. You’ll immediately see where money is likely hiding.

Why this works: Competitor pages sell. Reviews and forum comments complain. AI turns those raw, messy quotes into clear themes and a ranked action list so you don’t guess.

What you’ll need

• A browser and a notes app or spreadsheet
• Access to public reviews or comments (app stores, marketplace reviews, discussion threads)
• A basic AI assistant
• Optional: a simple landing-page tool or a marketplace listing

Step-by-step: from noise to a testable gap

1. Pick 3 competitors. For each, collect 10 short quotes that mention frustrations, delays, refunds, or “I wish it had…”
2. Paste all 30 quotes into AI and run the prompt below to cluster and score them.
3. Choose the top gap (highest score + fastest to fix). Translate it into one tiny offer you can deliver this week.
4. Launch a one-page test with a clear headline, simple benefit bullets, price (or free with waitlist), and email capture.
5. Drive 100–300 visits from 2–3 relevant communities or a small $20 ad. Measure interest, not perfection.

Copy-paste AI prompt (premium scoring + offer template)

“I’ll paste customer quotes from 3 competitors. Cluster into themes. For each theme, return: Frequency (# of mentions), Pain (0–5), Willingness to Pay Signal (0–5, based on urgency/ROI cues), and 2 example quotes. Compute GAP SCORE = Frequency x Pain x Pay. Rank themes high to low. For the top 3 themes, propose: 1) a minimum viable offer deliverable in 7–14 days, 2) a one-sentence value prop, 3) a 5-bullet feature list, 4) a simple pricing suggestion (low/mid), 5) 3 KPIs for a 7-day test, 6) likely objections and short answers.”

Insider trick: Ask AI to exclude generic complaints like “bad support” unless they appear 5+ times across multiple competitors. This removes noise and surfaces patterns people will actually pay to fix.

How to turn a gap into a micro-offer (the Offer-on-a-Page template)

• Headline formula: Stop [big pain] without [common objection] in [short time].
• Subhead: One clear outcome, one specific audience.
• Bullets (3–5): Features that directly map to the top complaints.
• CTA: “Get early access” or “Buy now” with price and delivery timeline.
• Risk reducer: “7-day no-questions refund” or “free preview sample.”

Worked example (so you can copy the pattern)

Niche: Resume help for mid-career job switchers.

• Top gaps from quotes: Generic templates, missing ATS keywords, slow turnaround.
• Micro-offer: 48-hour ATS Tune-Up: targeted keywords, modern format, and a 10-minute video critique.
• Price test: $49 (low) vs $79 (mid) on two separate pages.
• Headline: Stop getting filtered out by resume bots without rewriting your whole resume — in 48 hours.
• KPIs (7 days): 150 visits, 10–20% leads, 2–4% paid at the tested price.

Advanced prompt to create your landing page copy

“Using the top gap you identified, write a one-page offer. Include: headline (using ‘Stop [pain] without [objection] in [time]’), subhead for [specific audience], 5 benefit bullets mapped to complaints, 2 price options (low/mid) with positioning, a short guarantee, and an FAQ answering 5 likely objections. Keep language simple and direct.”

Prioritization cheat sheet (GAP score made practical)

• Green light: GAP score tops the list, appears across 2–3 competitors, and you can ship a fix in under 14 days.
• Yellow: High pain but tough delivery or unclear buyer. Save for later.
• Red: One-off complaints or vague “nice to have.” Ignore for now.

Common mistakes & fixes

• Mistake: Asking AI for gaps without feeding real quotes. Fix: Always paste actual customer language first.
• Mistake: Chasing a big idea you can’t deliver fast. Fix: Prefer small, high-pain fixes you can ship this week.
• Mistake: Vague audiences. Fix: Name them in the headline (e.g., “freelance designers,” “Etsy sellers”).
• Mistake: One price only. Fix: Test low vs mid price on separate pages for clearer signals.
• Mistake: Stopping at clicks. Fix: Collect emails and ask one question: “What made you click today?”

48-hour action plan

1. Hour 1–2: Pick 3 competitors. Copy 30 quotes (10 each) mentioning frustrations.
2. Hour 3: Run the clustering prompt and pick the top gap.
3. Hour 4–6: Draft your one-page offer using the landing-page prompt. Set two price points.
4. Day 2: Share in 2–3 relevant communities and send to your network. Aim for 150–300 visits. Track visits, leads, paid.
5. Decision: If leads ≥ 10% and paid ≥ 2%, iterate copy and raise price. If not, try the next gap.

Expectation reset: Your first win is a signal, not a salary. One validated micro-offer is the seed. Stack two or three of these and you have a steady side income.

Tell me your niche and the 3 competitors you’ll pull quotes from. I’ll help you craft the exact landing page headline and two price points to test this week.

Fast win (5 minutes): Take one SKU and use this 4-part snap template — Benefit → Mechanism → Proof → Reassurance. One headline, two short sentences. You’ll have a human-sounding description you can publish today.

Why this works: Generic AI fails because it skips specifics. Short copy that names a clear outcome, shows how it happens, and backs it with one fact is what gets clicks, adds, and sales.

What you’ll need

• Product facts: 1–2 key features, material, a number (size, hours, rating).
• One customer benefit: time saved, comfort, confidence, convenience.
• One proof point: rating, guarantee, standard, lab/material, trial, or a short customer quote.
• Your voice notes: straightforward, friendly, or premium-simple.

The 4-part micro-template

• Benefit: what the customer gets, in plain words.
• Mechanism: the single feature that makes it real.
• Proof: one concrete fact (number, material, rating, standard).
• Reassurance: remove risk (trial, guarantee, easy returns).

Step-by-step (run this once per product)

1. Collect 60 seconds of specifics: one number, one material, one risk reducer.
2. Pick your template flavor:
   • Snap Buy (everyday item): Benefit first, fast proof, light reassurance.
   • Considered Buy (sleep, health, appliances): Benefit, mechanism, stronger proof, clear risk reducer.
   • Spec-Led (tech/DIY): Benefit, mechanism with a key spec, proof tied to a standard, reassurance.
3. Draft one headline + two sentences using the micro-template.
4. Run the AI for 3 angles (practical, emotional, aspirational). Keep the one that fits your brand.
5. Apply the 3R edit: Replace one vague word with a number; Remove any extra feature; Reassure with trial/warranty/compatibility.
6. Ship and test: A/B the headline; track CTR and add-to-cart for a week.

Copy-paste AI prompt (premium, structured)

You are a conversion-focused ecommerce copywriter. Write three short product descriptions for [PRODUCT]. Each description must include 1 benefit-led headline + 2 short sentences. Use the format: Benefit → Mechanism → Proof → Reassurance. Deliver three angles: practical, emotional, aspirational. Keep total words per description to 25–35. Use simple language (Grade 6–8). Include exactly one concrete proof point (a number, material, rating, guarantee, or standard). If a claim lacks a number, ask me to supply it. Avoid generic words: innovative, premium, world-class, amazing, ultimate, revolutionary, top-notch. Emphasize outcomes. Inputs: Product facts: [LIST 2–3 FACTS]. Benefit: [BENEFIT]. Proof available: [NUMBER/MATERIAL/RATING/TRIAL]. Voice: [STRAIGHTFORWARD | FRIENDLY | PREMIUM-SIMPLE].

Worked example — Cordless Handheld Vacuum (Snap Buy)

• Headline: Quick clean-ups, no cord — crumbs gone in seconds.
• Sentence 1 (Mechanism): High-suction motor and a narrow crevice tool lift grit from seats and corners.
• Sentence 2 (Proof + Reassurance): 20-minute runtime and a washable filter keep costs low, backed by a 2-year warranty.

Another example — Blood Pressure Monitor (Considered Buy)

• Headline: Reliable readings at home — confidence in under a minute.
• Sentence 1 (Mechanism): The cuff auto-adjusts to your arm for consistent placement and fewer errors.
• Sentence 2 (Proof + Reassurance): Clinically validated to ISO standard with a 90-day trial, so you can return it if it doesn’t fit your routine.

Insider trick: build a “Specifics Bank” once, fuel every description

• Numbers: hours, capacity, weight, decibels, cycles, lifespan, ratings.
• Materials/standards: stainless steel, FSC paper, OEKO-TEX fabric, IP67, BPA-free.
• Risk reducers: 30–90 day trial, 1–3 year warranty, free returns, compatibility lists.
• Voice-of-customer snippets: 6–10 word phrases from reviews (“stopped my 3 a.m. wake-ups”).

Feed 1–2 from each line into the prompt. Specifics beat adjectives every time.

Anti-generic guardrails (use these rules in your prompt or edit)

• One outcome per description. No feature piles.
• One number, one material, one risk reducer.
• Sentences under 14–16 words; verbs over adjectives.
• Ban list: innovative, premium, world-class, amazing, ultimate, cutting-edge.
• Swap vague words with a measurable or sensory detail (“soft, cool-to-touch bamboo”).

Mistakes & fixes

• Reads like a spec sheet? Lead with an outcome, hide the rest in one mechanism phrase.
• No proof? Add a single number or recognized standard; remove two adjectives.
• Sounds robotic? Insert a short customer phrase or a tiny sensory cue.
• Too long? Cut the weakest clause. Keep headline + two sentences, max.
• Claims risk? If you lack data, frame as benefit-led intent (“designed to…”) and lean on trial/warranty.

7-day action plan

1. Day 1: Build a 10-item Specifics Bank from your top 10 SKUs.
2. Day 2: Rewrite 5 SKUs using the 4-part template; create 3 angles each.
3. Day 3: Publish A/B headline tests on 3 SKUs; log baseline CTR and add-to-cart.
4. Day 4–5: Replace underperforming angles; keep the winner per SKU.
5. Day 6: Roll the winners to email or ads for cross-channel consistency.
6. Day 7: Review metrics; lock 2–3 reusable templates (Snap, Considered, Spec-Led).

Expectation check

• AI gives you speed and variation; your edit adds credibility and voice.
• Two rounds usually get you from generic to conversion-ready.
• Small lifts compound: +10–20% headline CTR often nudges add-to-cart and revenue.

Keep it short, specific, and proof-led. Use the template to draft fast; let your human touch supply the detail that converts.

Quick win: In under 5 minutes you can turn a messy thread into a one-page task list. Clean the thread (remove duplicated quoted replies), paste the cleaned text into an AI, and ask: “List clear action items with owners and suggested due dates.”

One small correction: don’t delete every header or signature. Keep one line with each message’s sender and timestamp for context. And instead of auto-assigning a “default owner,” assign a tentative owner and ask them to confirm or reassign in the follow-up.

What you’ll need:

• A cleaned copy of the thread (remove full duplicate quoted text, keep sender & timestamp lines).
• A simple list of participants and roles (helps AI map responsibilities).
• An AI assistant (email client built-in, cloud tool, or a local/private model if confidentiality matters).

Step-by-step (what to do and what to expect)

1. Scan & trim (2–5 mins): remove repeated replies but keep one-line sender/timestamp and unique messages.
2. Highlight asks & decisions (3–5 mins): mark sentences like “Can you…”, “Please provide…”, or “We agreed to…”.
3. Run the AI (1–2 mins): paste cleaned text and use the prompt below to extract action items, owners, and suggested deadlines.
4. Verify & edit (2–5 mins): check owners and dates, fix any misassignments or ambiguous items.
5. Send a short follow-up email (1–3 mins): list items, owners, deadlines, and ask for quick confirmations or reassignments.

Example output (from a hypothetical marketing thread)

• Alice — Send final Q3 budget spreadsheet to finance by Fri Nov 29 (Owner: Alice).
• Raj — Schedule kickoff meeting and send calendar invite for Dec 2 (Owner: Raj).
• Marketing team — Provide creative brief draft by Mon Dec 6 (Owner: Marketing Lead; confirm who).

Copy-paste AI prompt (use this directly)

Here is a cleaned email thread. Please extract a concise list of action items. For each item: write a single clear sentence with who is responsible, the task, and a suggested due date (mark as tentative if not explicit). If ownership is unclear, list the most likely owner based on roles and flag it as “tentative.” Also provide a 2-line follow-up email that lists the actions and asks for confirmations or reassignments. Finally, flag any ambiguous points that need clarification.

Common mistakes & fixes

• AI misses nuance — fix by adding short context lines (e.g., “Budget impacts deadline”).
• Wrong owner — assign tentative owner and request confirmation in the follow-up.
• Sensitive data risk — use your in-house assistant or a local model, or redact sensitive details before pasting.

Simple 3-step action plan (do now)

• Pick one thread and trim duplicates (5 mins).
• Run the prompt above in your chosen AI tool (2 mins).
• Send the short follow-up asking for confirmations (2 mins).

Reminder: clear action items save time and reduce follow-ups. Start small, verify once, and you’ll cut future email churn dramatically.