Skip to content
Academy
Marketing Academy · Field Work●AI in Marketing
All projects·From: Choosing What to Automate First
MiniHead-to-Head· 20 minutes

Two Candidates, One Build Slot: Formula vs Judgment

RateGain Travel Technologies

Objective: Given two competing automation candidates with equal-looking urgency, apply the scoring formula and the 'keep human' filter side by side to decide which one actually gets built this sprint.

You're on the lifecycle marketing team at RateGain Travel Technologies, the travel-tech SaaS company, and your manager wants one automation shipped this sprint, not two half-built ones.

Compare Candidate A (a stable, repetitive reporting task) against Candidate B (a still-changing onboarding sequence) head-to-head on score and stability, then recommend which one to build now and which to hold.

Before you start

What you'll need

Free path (everything below is enough to finish)

Google Sheets
FreeCompare candidates on score and stability side by side

No account friction, a simple two-row table is enough

The process

1 step

Step 01 of 01

Filtering high-scoring tasks that still require human judgment or process stability

Step 3 warns that a high score alone isn't enough: don't automate brand-judgment calls, high-stakes customer tone, or a process that hasn't stabilized for a few months.

Candidate A: copying weekly hotel-partner API-uptime stats into a shared report, 3 hrs/wk, 4 setup hrs, score 39.0, unchanged for 8 months. Candidate B: the new-customer onboarding email sequence, 2 hrs/wk, 3 setup hrs, score 34.7, rewritten twice in the last six weeks. Which one ships this sprint, and why does the higher combined score not automatically decide it?

Google Sheets— A two-row comparison sheet with score, stability (months unchanged), and a judgment flag column.

Procedure

  1. Score both candidates with the Step 2 formula
  2. Add a 'months unchanged' column for each
  3. Apply the Step 3 filter: exclude any candidate whose process hasn't held steady for a few months
  4. Recommend the surviving candidate, and log the excluded one as 'wait, re-score next quarter'
Sample output
RateGain candidate comparison

              Score  Months unchanged  Verdict
Candidate A    39.0   8                 BUILD (stable, high score)
Candidate B    34.7   1.5               HOLD (still changing, would need a rebuild in weeks)

Healthy

Candidate A ships despite a smaller score gap, because Candidate B's process is still in flux and would need rework almost immediately.

Unhealthy

Building Candidate B because 34.7 is 'close enough' to 39.0, then rebuilding the automation three weeks later when the sequence changes again.

What this means

When two scores are close, stability is the tiebreaker, not raw score; automating a moving target erases the time saved.

So what do I do about it?

SymptomActionEffort
Two candidates score within a few points of each otherUse months-unchanged as the tiebreaker before defaulting to the higher number5 min
YouYou can do this yourself, no engineering access required.

Final deliverable

A one-page head-to-head verdict (build vs hold) for the two competing candidates, with the reasoning documented.

See a reference example
Sample output
Ustraa candidate comparison (excerpt)

                     Score  Months unchanged  Verdict
Inventory-alert sync   45.0   6                 BUILD
Influencer outreach DM 41.2   1                 HOLD (script still being tested)

Success criteria

You're done when you can:

  • Scores both candidates correctly
  • Does not pick the higher score by default when stability disqualifies it
  • States a clear build-vs-hold verdict with reasoning