Estimation

Good estimates are the foundation of fixed-fee pricing and realistic plans. This page describes how Northwind teams estimate. Estimates feed into Pricing guidelines and the SOW process.

Principles

  • Estimate as a team, not alone.
  • Give ranges, not single numbers, when uncertainty is high.
  • Estimate the whole job: design, build, test, documentation, review, deployment and project management.
  • Write down assumptions. Unwritten assumptions turn into disputes.

Methods

Stage Method Accuracy
Lead qualification Rough order of magnitude, from similar past projects plus or minus 50%
Proposal Work breakdown with three-point estimates plus or minus 25%
SOW Detailed estimate by deliverable, reviewed by an L4 or above plus or minus 15%
Delivery Story points and velocity Tracked sprint by sprint

Three-point estimates

For each task, give an optimistic (O), most likely (M) and pessimistic (P) estimate in days. The expected value is:

E = (O + 4M + P) / 6

For example, O = 4, M = 6 and P = 14 days gives E = (4 + 24 + 14) / 6 = 7 days.

Contingency

Project type Contingency to add
Fixed fee, well-understood scope 15%
Fixed fee, new technology or unclear requirements 25%
Time and materials 10%, shown as a range

Contingency is part of the price, not a hidden extra. Do not remove it to win a bid without approval from the delivery manager.

Estimation workshop

  1. Gather the team and the solution architect (Isha Patel for most bids).
  2. Walk through the requirements and list assumptions.
  3. Break work into tasks of no more than 5 days.
  4. Each person estimates privately, then the team compares.
  5. Discuss the biggest differences and re-estimate.
  6. Record the result and assumptions in the proposal workspace.

Reviewing and approving

Estimates for SOWs are reviewed by a delivery manager such as Maya Torres. If the SOW value is above $50,000 it also needs VP Delivery and General Counsel approval under POL-LEG-001.

Learning from actuals

At project close, compare estimates with actuals and record lessons learned. Patterns, such as consistently underestimating testing, feed back into this page. See How we run projects and the definition of done.

Owner: Maya Torres - Last updated: 2026-06-02