0 of 25 answered
0%
AI Initiatives · Maneuver Marketing

Help us build the
right thing first.

We're mapping AI opportunities across growth, creative, and operations. Your honest answers — real numbers, actual workflows, genuine failures — directly shape what we build first.

~20 minutes
📋
7 sections · 25 questions
🔒
Goes to Ruki & Vishakha only
Why this matters

We're deciding what to build first across AI tooling, workflow automation, and data infrastructure. The more specific your answers, the better we can prioritise. Vague answers get vague solutions.

Meta launched an official Claude connector (MCP) on 29 Apr — it lets Claude read and act on ad accounts directly. Jason flagged it. Read-only pilot starts 14 May. A few questions below ask for your take.
👤
Q1

Your name, role, and team

Full name
Role / title
Team / department
Q2

Which products and markets are you directly responsible for?

Be specific — this tells us which ad accounts, data pipelines, and tools to connect first

Q3

Volume snapshot — roughly how many right now?

Ballpark figures are fine

Active ads in test
New ads / week
Daily spend (~USD)
Q4

Who else touches your workflow — from brief to live ad?

Include everyone in the chain — who gives you briefs, who you hand off to, who reviews before launch

Q5

How long do these tasks take today vs your ideal?

Include data collection, formatting, re-running prompts, chasing files — not just the AI part

TaskToday (hrs)Ideal (hrs)
Weekly performance report
Creating one image brief
Generating + reviewing one static ad
Compiling research for a new angle
Pulling + prepping data before prompting
Assembling + uploading ads to Meta
Q6

Which two tasks would you most want to get time back on?

Pick your top two — these become our highest-leverage targets

🧪
Q7

Describe the most useful AI workflow you've built or used — even if scrappy.

What does it do, what inputs does it need, what does it output, how often do you use it?

Q8

What have you tried that didn't work — and why do you think it failed?

Be honest. "It gave generic outputs" is useful. "It hallucinated data" is even more useful.

Q9

How do you set Claude up before starting a task?

🔗
Q10

Which data sources do you currently pull from manually? Tick all that apply.

Q11

If three sources were auto-connected to Claude — which three would change your week most?

Think about what you export most often, or what takes longest to collect before you can start analysing

#1 highest impact
#2
#3
Q12

How do you define a winning ad? What specific metrics or thresholds matter to you?

This tells us what any automated reporting needs to measure. Be specific — "good performance" isn't enough.

🎨
Q13

Walk through exactly how a new ad gets made — from first idea to live in Ads Manager.

Who does what, which tools are touched, where things slow down, how long each handoff takes

Q14

When a test fails, how do you figure out why? Walk through your actual process.

Is there a system, or gut feel? Where is the data and how long does working it out take?

Q15

Is there a record anywhere of what angles, hooks, or formats you've tested and what happened?

Q15b

Which tools are you currently using to manage briefs, tasks, and creative tracking?

Tick everything you actively use — even if it's just a personal spreadsheet

Q15c

If we built one central place for briefs, test results, and creative tracking — which platform would you actually use?

The best platform is the one people will actually open every day — be honest

Q16

Rate how painful each gap is right now.

1 = barely notice it  ·  5 = costs me real time or performance every single week

1 · No impact5 · Major weekly cost
Gap12345
No record of past test results
Manual data collection before any analysis
Brief creation takes too long
Image generation is manual and repetitive
No visibility into what others are testing
AI outputs too generic — missing brand context
Q17

In the read-only pilot, what specific reports would be most useful for Claude to pull automatically?

Think: what do you check every Monday morning, or what data takes longest to pull manually before you can start work

Q18

What would need to be true for you to be comfortable with Claude eventually pausing a fatigued ad set — notification only, no manual approval?

This shapes the guardrails before we ever touch write access

Q19

If Claude generated a weekly briefing from your Meta account — what would you always want it to include?

🎯
Q20

Rank these six initiatives — assign each a unique number from 1 (most urgent) to 6 (least urgent).

Each number can only be used once. If it's not relevant to your role, rank it 5 or 6.

InitiativePriority 1–6
Business memory system
Fireflies + Slack + BigQuery → weekly AI brief · GCP-hosted
Closed-loop creative engine
Research → brief → image → performance → back to research
Static ad production at scale
Brief → image API → human review → Meta upload
Unified workflow tracker (Notion)
Briefs, tests, results, team management in one place
Company-wide Claude skills library
Shared report, brief, and creative strategy skill files
Meta Ads MCP pilot
Read-only Claude connection to ad accounts · 14 May start

Duplicate number — each initiative needs a unique rank 1–6.

Q21

For your #1 initiative — what does success look like in 8 weeks?

Not "it works" — what specific metric, workflow change, time saving, or decision would be different?

Q22

What's the one thing that would make you stop using an AI tool after trying it?

This helps us build quality checks into the right places from day one

Q23

Are you available the week of 14 May for a 45-min alignment session?

Ready to submit?

Your responses go directly to Ruki and Vishakha. We'll share a consolidated summary of priorities across the team before the 14 May session.

Something went wrong — please try again in a moment.

Thank you — this really helps.

Your responses have been recorded and are with Ruki and Vishakha.

1
We'll review all responses and map the priorities across the team
2
You'll receive a summary showing how your answers compare with others before 14 May
3
We'll align on Phase 1 priorities together in the 14 May session — your input directly shapes what gets built first