Excerpt
Here are 12 of the 23 questions — a selection from all five blocks. In the PDF, every question also comes with: example answers (good/bad), typical follow-up questions, and what the answer means for the setup.
🎯 Block A · Goal & Context
Question 01
If the agent is working perfectly in 6 months — what has concretely changed?
This question replaces the usual "what do you want to achieve?" — it forces concrete pictures instead of abstract goals. Answers like "more efficiency" are warning signs. Answers like "I spend 20 minutes a day on email instead of 3 hours" are gold — that's 13 hours back per week.
Question 02
What is the one metric that would show the setup is working?
Anyone who can't name a single metric, but names five instead, hasn't thought the project through. We insist on one main metric plus at most two supporting ones. More is a distraction, fewer is avoidance.
Question 03
What's your current bottleneck — time, leads, or capacity?
These three are the only relevant bottlenecks in 95% of SMBs. Which one applies directly determines the first agent. Time bottleneck → inbox/research. Lead bottleneck → outreach. Capacity bottleneck → recruiting. Anyone who names all three doesn't have one.
⚙️ Block B · The Process Today
Question 06
What steps does a person who runs this process today take — step by step?
Anyone who can't describe the process in 7–12 numbered steps doesn't have a process — they have situational improvisation. That's not automatable. This question is the first reality check before every setup.
Question 07
How often does the process occur, and what variations exist?
Frequency × standardization = ROI potential. 100x a week with 80% identical steps? Top candidate. 5x a week with something different every time? Poor candidate — even if it hurts. Variations are the biggest hidden complexity.
Question 08
When does the process go wrong today — and who notices?
Escalation logic comes from this question. Anyone who can't describe the failure scenarios doesn't know what they're automating. Bonus: the answer also reveals whether the person really knows the process — or is just reciting theory.
💾 Block C · Data & Systems
Question 11
Where does the data the agent needs to process live — system, format, access?
Excel on a drive? Cloud CRM with an API? Notion database? Every answer has different consequences for setup effort and ongoing cost. "In my employees' heads" is a real answer we hear — and a stop signal.
Question 12
Which tools do you actively use today in the relevant workflow?
Your existing tool stack determines the integration paths. Anyone using Slack, Hubspot, and Notion can integrate via API. Anyone doing everything through Outlook and Excel attachments has a different path. Both work — but at very different effort levels.
⚖️ Block D · Risk & Compliance
Question 16
What happens if the agent makes a wrong decision?
"Nobody notices" → no escalation effort needed. "Customer upset" → escalation logic matters. "Legal consequences" → human review is mandatory. This question decides how autonomously the agent is allowed to work.
Question 17
Does the agent process personal data — if so, which category?
GDPR relevance is decided right here. "Standard contact data" is one world. "Health data" or "children's data" is a completely different one. Answering vaguely here risks nasty surprises during implementation. Better to clarify for 30 minutes than lose 30 days later.
👥 Block E · Team & Ownership
Question 20
Who is the person responsible for this agent on a daily basis?
If the answer is "all of us" or "we still need to figure that out," we stop the project. 91% of successful setups have a clearly named owner. Anyone without one fails within the first 60 days — we've seen it often enough.
Question 21
How much time can this person realistically invest in the first 30 days?
Realistic: 4–8 hours a week during setup, 1–2 hours after. Anyone who says "1 hour a week" is underestimating the effort. Anyone who says "I can't promise that" reveals that the owner and the person actually responsible aren't the same. Both are warning signs.
Question 23
Who on the team is against this project — and why?
This question surprises people — and it's the most important one in the block. Anyone who can't name any opponents hasn't communicated the project openly. Setups that run in secret fail on internal resistance. Surfacing concerns early is mandatory — not optional.
How to use this questionnaire
Go through it alone — before talking to anyone else. Write your answers as bullet points. Wherever you struggle to form a sentence — that's where clarification is needed. Those gaps are the most important insight the questionnaire produces. Only after you've answered all 23 questions for yourself should you take it into the team discussion or to an outside consultant.