Jason Schemmel

Four marketing projects competing for your next hour? Write: what I observed, what I don't know, and one check I can finish. For an active or near-launch offer, the free Marketing Bottleneck Scorecard can help you choose a starting point: bottleneck.gsdlab.co

Before reusing an older post, check the offer, links, dates, prices, screenshots, instructions and permissions. Try the tutorial steps again. Keep the useful explanation and save a dated revised master so an outdated version does not return to the queue.

Turn one training note into work: “Using [lesson], I will finish [small output] for [current task]. I will check it by [observable review].” Set a short work block. Keep the output or record the exact obstacle. That gives the next support conversation a purpose.

Giving AI an old post as a style example? Separate STYLE EXAMPLE from CURRENT FACTS. Tell it to borrow tone and structure, not old dates, prices, stories or results. Check the new draft against the current facts, then read it aloud.

Before sending “just checking in,” ask what would help the customer decide. Recall their request, add one accurate answer or useful detail, and give the actual next step. Use it in an appropriate existing conversation and honor contact preferences.

Give each inquiry-form field a job: what decision does the answer support, and is it needed before the first response? Check dependencies before moving it. Then test a demonstration inquiry. The goal is enough useful information for the next step.

Check the start of a video: “This fixes it” may mean nothing to a new viewer. Name the subject: “Your quote form asks for the same details twice.” Then show the example. The first sentence and first visual should help someone follow the same task.

Show the judgment behind your work: situation → options → choice → reason → tradeoff → when you would reconsider. Use a real decision, remove private details and say what remains untested. Expertise becomes useful when someone can follow your reasoning.

Help a buyer choose: This is useful when ___. To get value, you need ___. If you need ___ instead, ask about ___. Use your actual service boundaries. Clear scope is more helpful than saying your offer is for everyone.

Before saving an AI prompt as a template, test two cases: complete facts, then one missing fact. Does it flag the gap or invent an answer? Add a rule against made-up prices, dates and promises. Retest with demonstration details.

Match the next action to the lesson: try the demonstrated step, clarify a missing detail or find a starting point. Unsure which marketing task comes first? The free Scorecard can help structure the question: bottleneck.gsdlab.co

Keep two inquiry questions separate: How did you first hear about us? What prompted contact this time? Record the answer, any source data you can observe and what is unknown. A form submission is a point in the path, not proof of which channel caused a sale.

Before automating a reply, define the exception: approved facts, requests needing a new decision, the person responsible and what the customer should expect next. Test one routine case and one exception. A prepared answer should not invent a new promise.

After a project, ask about the process: what was unclear, what helped and what should change? Ask customers with mixed experiences too. Use specific feedback to improve delivery. Get separate permission before sharing their words publicly.

Build a question log before asking AI for content topics: actual question + source reference + explanation needed. Use anonymized excerpts. Ask AI to extract only questions present, then check every entry against the notes. Repetition alone is not purchase demand.

A useful screen video can show one checklist with your narration. Script: the customer question, the actual details to gather and the next step. Use demonstration data. Watch on your phone for readable text, clear sound and a complete explanation.

GHL basics: contact = person; opportunity = potential deal; pipeline/stage = sales process and current step; workflow = actions after a triggering event. Describe these for one process before building it. Use demonstration data to test your setup.

Give AI four inputs: who the post should help, the question it should answer, the facts it may use and the output you need. Ask it to flag missing information. Then check accuracy and whether it sounds like you before using the draft.

Busy client week? Choose one real customer question, prepare one useful answer and give it a publishing slot. Protect a separate inquiry-review slot without replacing your existing response commitments. Review what you finished and what still needs attention.

Define one weekly marketing check: suspected constraint, evidence, one action, owner and time, and what you will review. Record what changed and what remains unknown. Seven days is a review window, not a sales-result promise.