Peach Pilot
Analysis Plan - Internal

Reducing Chargebacks at SafeLife

A step by step plan that takes us from validated data to a working solution. Every figure traces back to the source.
Where we stand today
~32%
Of the book charges back
~26%
Of commission dollars
$1.24M
Clawed back to date
5%
Target rate

At a glance

Six phases, run in order. Each one produces a deliverable your team can review before we move on. The last phase is where the solution starts.

1
Data

Consolidate the Data

Agree on exactly what data we can use and consolidate it into a single source of truth for future analysis.

Steps

  • Retrieve raw and formatted data from SafeLife and Fintary.
  • Consolidate the data into a single location.
  • Validate the completion and truth of the data across sources.
  • Define a chargeback precisely: a labeled, net negative clawback.
  • Note issues in the source data separately*

Source data issues flagged

Issue in the source dataSolution
Same policy in two number formatsMerge to one record per policy
Duplicate policy rowsCollapse to one per policy
Agency and house accounts mixed inExclude from chargebacks
Repeated chargeback linesRemove the duplicates
Chargeback dated before the saleFlag for review

Each issue is flagged in an audit column, and large issues are recorded separately.

Deliverable: Single Source Data Sheet, condensed and validated. Easily understandable by all stakeholders.
2
The Now

Chargeback Foundation

See where we've been, where we are, and where we are going. Gathering baseline statistics.

Steps

  • Calculate the current chargeback rate by policy count and by commission dollars.
  • Plot important trends across and within the past 12 months
  • Analyze the dollar impact and highlight where it lands (agents vs SafeLife).
  • Calculate recovered dollar amount at target chargeback level
  • Begin identifying notable patterns such as sale-to-chargeback timing and chargeback amounts.

Baseline today

ViewRate
Full book, by count~32%
2026 sales, by count~23%
By commission dollars~26%
Total clawed back$1.24M
Deliverable: Baseline Scorecard, show how the problem lives and exist within SafeLife.
3
The where

Identify Hot Spots

Dig deep into all potential causes of chargebacks and identify areas of focus.

Steps

  • Break down chargeback rates with respect to measurements in source data.
  • Isolate specific segments that exhibit unusually high corelation to chargeback rates.
  • Rank segments relative to their chargeback rates
  • Speak with agents on the call floor to hear their stories.
  • Prepare hypotheses for each hot spot, to be tested in the next phase
Deliverable: Ranked Hot Spots List, stress the areas of highest impact and actionablility.
4
The why

Hypothesis Testing

Rigourously test hypotheses to separate real drivers from coincidence. Look for the story in the details.

Steps

  • Form a clear hypothesis for each hot spot from the previous phase.
  • Test each hypothesis for statistical significance against the source data.
  • Control for overlapping factors so one cause is not mistaken for another.
  • Separate real drivers from coincidence and random noise to determine the path moving forward.

Validation step

  • Compute over claim.
  • Run every candidate driver through an adversarial review.
  • All claims must be supported with cited evidence.
  • Any debunked theories go on a discard list, with the reason.
Deliverable: a short list of validated drivers, each with evidence, effect size, and confidence, plus the discard pile.
5
The so what

Prioritize the Levers

Begin turning validated drivers into a ranked action shortlist for later solution development.

Steps

  • Score each driver by impact: the chargeback dollars and policies it explains.
  • Investigate how existing solutions can realistically change it.
  • Brainstorm potential ways to address each lever, note overlap.
  • Rank the levers into a shortlist of where to act first.

Scoring shape

LeverImpactCan we change it
Driver AHighHigh
Driver BHighMedium
Driver CMediumLow

Illustrative only. Real levers from Phase 4.

Deliverable: Lever Board, ranked, with the expected upside of each.
6
Production start

Production Start

Begin building, aimed at the top one or two validated levers with highest impact and coverage. Handoff from analysis into production.

Steps

  • Turn the top levers into a concrete pilot product.
  • Set the success metric up front: the chargeback rate change we expect.
  • Consider integration with pre-existing Peach Pilot solutions
  • Set the pilot guardrails and the review cadence.

Transfer to Production

  • A pilot charter ready to build.
  • A measurable target tied to the 5% goal.
  • A system to monitor the pilot's progress.
Deliverable: Initial Pilot Plan, the beginning of the solution, ready to build.

Questions/Comments?