Steal Our Ideas · no email gate, no watermark

The ideas are free. The judgement is the practice.

Most consultancies guard their thinking like it’s the product. It isn’t. Anyone can hold an idea. Knowing which one fits your business, in what order, at what cost, is what you actually pay for. So here: take the ideas.

The insights

Articles, given away whole.

Longer-form thinking: each one is a hypothesis, an argument, and a mechanism you can run on Monday. No gate, no summary-only tease. Every article has its own page; send one to whoever needs it.

The prompts

Prompts we actually use.

Working prompts from live engagements. Copy one, feed it your real material, and judge the answer like an operator.

Prompt · Positioning

The one-sentence position test

Put your copy, a competitor's copy and three customer quotes in front of a model and let it write the sentence a busy buyer would use for each of you.

Here is our homepage copy, our top competitor's homepage copy, and three customer quotes. Write the one sentence a busy buyer in our industry would use to describe each of us to a colleague. If ours is generic or wrong, propose three sharper positions we could defensibly own, with the proof each would require.
Prompt · Positioning

Let the losses write your positioning

Losses are cleaner positioning data than wins: nobody flatters you on the way out.

Here are our last five lost deals: what was said, who won, and at what price. Name the position each winner occupied that we didn't. Cluster the patterns, then propose the single position we should claim to pre-empt the most common loss, with the proof it would require.
Prompt · Proof

The proof-mining prompt

Turn a raw dump of results into ranked, one-line proof statements. No adjectives allowed; numbers do the talking.

Here is a raw dump of our results, projects and client outcomes. Extract every claim that has a number, a name, or a date attached. Rank them by how surprising they would be to a skeptical operator in our industry, and rewrite the top five as one-line proof statements. No adjectives allowed, numbers do the talking.
Prompt · Proof

The skeptic pass

Run every case study past a simulated burned-by-agencies operator before it ships. What survives is your actual proof.

You are a skeptical operator who has been burned by agencies. Here is our case study. Mark every sentence you don't believe and say why. Then rewrite the case study keeping only what survived, and list the evidence we would need to make the cut sentences believable.
Prompt · Operations

The heroics audit

Find every step in the business that depends on one person's memory, and classify the fix.

I will describe how work actually gets done in our business: who does what, in which tools, with which handoffs. Identify every step that depends on one specific person's memory or manual effort. For each, state what breaks when they are on leave, and whether the fix is a process, an integration, or an automation.
Prompt · Operations

The senior-hours ledger

Price the process work your senior people are quietly doing, then rank the automations by hours returned.

Here is a week of my calendar and my team's recurring tasks. Classify every block as judgement work (only a senior human can do it) or process work (a system could do it with oversight). Total the process hours, multiply by our loaded hourly cost, and list the top three automations by hours returned.
Prompt · GTM

The funnel confession

The refusals are the insight: most engines leak before the stage everyone is optimising.

Here are our funnel numbers, stage by stage. Where would you put the next $10k, and what would you refuse to spend on until something else is fixed? Explain each refusal, and name the single joint whose repair would raise the value of every upstream channel.
Prompt · Strategy

The pre-mortem prompt

Run the failure post-mortem before the kickoff deck, not after the burn.

It is 18 months from now and this initiative failed. Write the honest internal post-mortem: what we ignored, who warned us, which assumption broke first, and what the early warning signs were. Then list the three cheapest tests we could run this month to check those assumptions.

The stack

Tool stack, tiered.

Seven levels of the design and build stack the practice actually runs, from reference to data. Steal the whole ladder.

Level 02InspirationdribbbleMobbin
Level 04Componentsshadcn/ui21st.devMagic UI
Level 05MotionGSAP
Level 06GenerationOpenArt
Level 07Datafirecrawl.dev

Stolen enough?

The next step is knowing which idea your business should run first. That’s the Audit: a named diagnosis, from $5,000, inside two weeks.