Budget Forecast Engine

P1Cycle Plan

LH-1039 · PM Marcus Lee

Project Pool
2Cycle Plan
3Pre-approved
4Approved
Saved

FP&A rebuilds the annual and quarterly forecast in spreadsheets each cycle. The process is slow, hard to audit and breaks whenever an analyst leaves. This project replaces it with a governed forecasting service that pulls actuals directly from the data platform and lets finance run scenarios without re-keying data.

Saved
  • Forecast cycles take three weeks of manual spreadsheet work.
  • No shared source of truth — every team keeps its own numbers.
  • Scenario changes are error-prone and cannot be traced or reviewed.
Saved
  • Cut forecast turnaround from three weeks to under three days.
  • One governed forecast that every business unit shares.
  • Self-serve scenarios with a full, reviewable change history.
Annual savings
$1.2Manalyst time
New earnings
$0.6Mfaster planning
Outsource budget
$240kcapped

Budget composition

  • Internal mandays520
  • Outsourcing240
  • Infrastructure120
1.8xEst. ROI
Payback period16 months
Break-evenQ2 2027
Net value, 3yr$1.4M
ConfidenceMedium
ROI is most sensitive to adoption in FP&A. If the beta lands on time, payback pulls in to roughly 13 months; a one-month slip pushes it past 18.
Saved

Solution

A batch ETL layer lands actuals from the data platform into a forecast store. A forecasting service exposes a versioned scenario API, and a thin React UI drives scenario editing. Every scenario is immutable and diffable.

Risk points

The data-retention policy for forecast snapshots is unresolved, and the model service depends on the ML Ops pipeline (LH-1067), which is still in cycle plan.

Project epics
EpicLeadStatusProgressEst. md

Data ingestion & ETL

Aiko Tanaka
In progress

24

Forecast model service

Marcus Lee
In progress

30

Scenario planning UI

Ivan Petrov
Not started

18

Integrations & rollout

Priya Shah
Not started

16

4 epics

88 md estimated

Data Platform

8 logged / 30 planned md

Finance eng

4 logged / 34 planned md

Frontend

0 logged / 18 planned md

QA & rollout

0 logged / 6 planned md

14% logged
Mandays logged12 / 88 md

  1. Kickoff

    5 Jun

    Done

  2. Data model sign-off

    28 Jun

    Done

  3. Model MVP

    15 Aug

    In progress

  4. Beta with FP&A

    10 Sep

    Upcoming

  5. Go-live

    30 Sep

    Upcoming

  1. Aiko Tanaka

    Architect

    Held

    15 Jun, 16:20

    Needs a clearer data-retention plan before I can sign off the ingestion design.

    Pre-approvedCycle Plan

  2. Dana Reid

    Sponsor

    Approved

    14 Jun, 09:05

    Business case is solid and the payback holds. Proceed to committee review.

    Cycle PlanPre-approved

  3. Marcus Lee

    PM

    Submitted

    12 Jun, 11:40

    Submitted for cycle planning with the updated ROI model and staffing plan.

    Project PoolCycle Plan

  • BRD-2041.pdf

    1.2 MB · Priya Shah

  • Architecture-overview.md

    18 KB · Aiko Tanaka

  • ROI-model.xlsx

    340 KB · Marcus Lee

  • Data-flow.png

    820 KB · Aiko Tanaka

This project is 12 of 88 mandays in with the Model MVP milestone on the critical path. One review comment is unresolved (data retention) and the model service depends on LH-1067.