About
Starting Semester: Fall 2026Assigned: No
Location: Atlanta
Accenture
Client Profile
Accenture houses their brand new Design-to-Value lab where we take innovative approaches to re-engineer and deliver the best value we can for our clients.Project Description
Problem / NeedToday, DtV models physical products. It does not model services.
Every parametric model the practice maintains rests on the same primitive: physical inputs multiplied by process rates. Material mass × price per pound. Line speed × machine hours × burdened rate. That structure holds for anything manufactured, and it is why the practice can walk into a negotiation with a defensible number.
Labor-based services have no equivalent primitive. There is no bill of materials for cleaning a building or installing a countertop. To model that cost bottoms-up, an analyst would need three things the practice does not have:
1. A task taxonomy – a structured decomposition of the service into discrete, observable, repeatable work elements.
2. Time standards – how long each element takes, and how much that varies by crew, site, and conditions.
3. A capture method – a repeatable way to generate the first two for a category nobody has modeled before.
Because none of these exist, DtV scopes almost exclusively to direct spend – the physical goods that go into a client's product. Indirect and services spend is left unaddressed, not because it lacks opportunity, but because the practice has no method to attack it. Janitorial, facilities maintenance, landscaping, security, installation, and field service are negotiated on price history and market benchmarking rather than cost structure. These are precisely the categories where a should-cost model would create the most leverage, because incumbent pricing is least disciplined.
The gap compounds: every engagement that touches services spend either skips those categories or estimates them anecdotally, and no reusable asset is left behind. The practice starts from zero each time.
Objective
Design and validate a generalized methodology and supporting tool that allows an analyst to build a defensible bottoms-up labor cost model for a services category that has not been modeled before.
The system must:
• Decompose an unfamiliar labor service into a structured, repeatable task taxonomy
• Capture task-level time data through a defined instrument, with statistically defensible sample sizes
• Convert that data into a cost model producing cost per unit of output, with explicit sensitivity to workforce composition, wage and burden structure, service frequency, site characteristics, and productivity drivers
• Be operable by a new analyst on a new category without rebuilding the framework
Validation strategy: calibrate and validate on janitorial services as the primary category, then run a transfer test on installation services to demonstrate the framework generalizes rather than being purpose-built for one type of work.
Scope: Four Components
1. Task Taxonomy Schema
A generalized structure for decomposing any labor service into observable, time-able elements – including a granularity rule so that two analysts independently decomposing the same service arrive at substantially the same task list. This is the core intellectual contribution and the most reusable asset.
2. Time Capture Instrument
The mechanism for timestamping work: direct time study, video-based coding, work sampling, or a hybrid. The team should determine which method fits which service type, establish required sample sizes for target confidence intervals, and test inter-observer reliability.
Critical elements of the process to be included, but not limited to, in the time study output to support cycle efficiency optimization:
• Takt Time vs. Cycle Time
• % Value Added
• Lead Time vs. Process Time
• Utilization
• Bottleneck Step
• Labor Content (hours/unit)
• Quality Losses (rework/scrap)
3. Parametric Calculation Engine
Task times × frequency × crew composition × fully burdened wage rates × allowance factors (personal, fatigue, delay) → cost per unit of output. Must expose driver sensitivity, not just produce a point estimate.
4. Category Onboarding Playbook
The procedural wrapper: how an analyst goes from zero knowledge of a category to a calibrated model, and how long that should take.
Build vs. Buy – a required early task, with a bounded decision date
Published standards exist and the team should validate against them rather than reinvent them: ISSA cleaning time standards, RSMeans facilities maintenance data, predetermined motion time systems (MOST, MTM), BLS occupational wage data by metro area, and commercial video time-study tools.
The distinction to test: these are static reference tables for known tasks. What DtV needs is a parametric model tied to cost drivers, plus a repeatable method for onboarding categories where no published standard exists. The team must reach a documented build/buy decision by [Date TBC] and proceed.
Skills
Industrial Engineering Methods RequiredTime and motion study with statistical sample size determination · work sampling · predetermined motion time systems (MOST/MTM) · process mapping and value stream analysis · design of experiments across site and crew factors · regression modeling of cost drivers · Monte Carlo or discrete-event simulation for variability · inter-rater reliability testing · Lean Six Sigma waste classification · activity-based costing and cost engineering. Also knowledge of AI use will help create the general dashboard.