Core engagement
UA Decision Logic Sprint
2 weeks
Remote · async-first
One decision workflow per sprint
From messy UA manager logic to validated recommendation rules and production-ready specs — built around one real recurring decision your team makes every week.
Who it's for
Good fit if your team is
This isn't for everyone. It works best when there's already a decision being made — just not formalized.
Building UA automation or recommendation systems
You're designing the system, but the decision logic behind it lives in people's heads — not in specs. The sprint extracts and documents that logic.
Connecting AI agents to ad accounts
Before an agent can safely act on a budget signal, you need to know: which signals are ready, which need trust checks, and what's forbidden without human approval.
Scaling a UA team where logic is in one person's head
One senior UA manager holds all the decision logic. The sprint turns that into a documented, transferable, and testable system.
Experiencing unexplained budget efficiency drops
ROAS looks fine, but LTV is drifting. Trials are growing, but renewals aren't. The sprint maps which signals you're actually acting on — and whether you should be.
How it works
Two weeks, one decision workflow
We pick one recurring UA decision — scale, hold, cut, or reallocate — and build the full decision logic layer around it.
Days
1–2
Decision Inventory & Kickoff
Map all recurring UA decisions. Select the one with the most automation potential or the most inconsistency. Identify inputs, owners, and current failure modes.
Days
3–5
Signal Rights & Source-of-Truth Mapping
Classify every signal feeding this decision: budget-ready, diagnostic-only, or immature. Identify the source-of-truth gaps and trust check requirements.
Days
6–9
Decision Tree & Recommendation Rules
Build the decision tree. Define the logic: if signal X is budget-ready and trust checks pass → allowed action Y within guardrail Z. Validate against 3–5 historical decisions.
Days
10–14
Guardrails, Approval Rules & Automation Roadmap
Define budget change limits, human approval thresholds, rollback rules, and the learning loop. Deliver the automation readiness roadmap: what can be automated now vs. what needs more validation.
Deliverables
What you get at the end
All outputs are in a format ready for engineering, data, or product teams to act on.
01
Decision Inventory
All recurring UA decisions mapped: type, trigger, frequency, owner, current inputs.
02
Signal Rights Map
Every key signal classified: budget-ready / diagnostic-only / immature. Allowed actions per class.
03
Source-of-Truth Gap Report
Where attribution, MMP, finance, and product data conflict — and how to resolve it for decision-making.
04
Decision Tree
Documented, testable decision logic for the selected workflow. Validated against historical cases.
05
Recommendation Rules
Structured rules ready for a recommendation system or agent workflow. Includes trust checks and data requirements.
06
Guardrails & Approval Logic
Budget change limits, human approval thresholds, rollback rules, and the learning loop spec.
07
Automation Roadmap
What can be automated now, what needs more validation, and what requires permanent human oversight.
08
Data Requirements Spec
Exactly what data the decision tree needs, at what granularity, and with what freshness — for engineering handoff.
Fit check
Not a good fit if
You don't yet have a repeating UA decision process — the sprint needs at least one real, recurring decision to work with.
You're looking for someone to set up ad accounts, run campaigns, or manage day-to-day UA — this is decision logic work, not campaign management.
You want a dashboard or BI tool — the output is decision specs and logic, not a visualization layer.
You need a fully automated system delivered end-to-end — the sprint delivers the decision logic layer; engineering implementation is your team's work.
Start with the scorecard
Before we talk, take 10 minutes to score your current UA decision logic across 9 blocks. Send me your total score and your 2 weakest areas — I'll tell you whether the sprint is the right fit and where we'd start.
Or just reach out directly — describe the UA decision workflow you want to formalize and we'll take it from there.