The rule

Elements of a process of experimentation

  1. Identify the uncertainty State the capability, method or design question the team could not answer at the outset.
  2. Identify alternatives Formulate hypotheses or candidate designs intended to eliminate the uncertainty.
  3. Evaluate alternatives Test, model, simulate, prototype or analyse the alternatives systematically.
  4. Refine or discard Use results to refine hypotheses or select a design; iterate as needed.
  5. Record conclusions Document what was learned, including failures.

Methods that can constitute experimentation

MethodWhat makes it experimentalWhat makes it routine
Systematic evaluation of alternativesComparing candidate designs against defined criteria where the best choice is unknownSelecting from a catalogue by known specifications
TestingTests designed to answer an open technical questionRoutine quality-control or acceptance testing
PrototypesBuilding to learn whether a design worksBuilding a production unit from a finished design
Modeling and simulationFEA, CFD or performance models used to evaluate alternativesRunning a standard calculation to document compliance
Design iterationsRevisions driven by test or analysis resultsCosmetic or customer-preference revisions
Engineering analysisAnalysis to resolve which approach will meet requirementsCode-compliance checks with known outcomes
Trial and errorStructured, recorded trials with hypothesesUnstructured tinkering without evaluation
ValidationValidation that tests whether uncertain design goals are metValidation after the design is final and known to work (may be after commercial production)

Examples by industry

Example · Software

Distributed system architecture

A team designing a multi-region replication system faced uncertainty about achieving consistency and failover targets under network partitions. They modeled three consensus approaches and ran fault-injection tests.

Business component: Replication subsystem

Potentially qualifying factors
  • Modeling and simulation of alternatives
  • Fault-injection experiments with recorded outcomes
  • Design iterations driven by test results
Potential exclusion issues
  • Deploying the chosen design to production
  • Operations and on-call work
  • Configuration of managed cloud services without experimentation
Documentation required
  • Design documents comparing alternatives
  • Fault-injection test logs
  • Version-control history
  • Contemporaneous records of the uncertainty at the outset
  • Records of alternatives evaluated and results
  • Time allocation for personnel involved

Illustrative example. Qualification depends on the specific facts of each business component and is not automatic.

Example · Manufacturing

Tolerance refinement

A precision component required tolerances tighter than the shop had achieved. Machinists and engineers trialed tooling paths, fixtures and thermal controls.

Business component: Precision shaft (product) and machining method (process)

Potentially qualifying factors
  • Uncertainty about method to achieve tolerance
  • Iterative trials with measurement data
Potential exclusion issues
  • Routine first-article inspection
  • Production runs after the method was established
Documentation required
  • CMM measurement data by trial
  • Fixture design iterations
  • Contemporaneous records of the uncertainty at the outset
  • Records of alternatives evaluated and results
  • Time allocation for personnel involved

Illustrative example. Qualification depends on the specific facts of each business component and is not automatic.

Example · Engineering

Structural analysis

An engineering firm designing an unusual long-span roof could not determine from codes alone whether the proposed system would meet deflection limits. It modeled alternatives and iterated the design.

Business component: Roof structural system design

Potentially qualifying factors
  • Uncertainty beyond code-prescribed solutions
  • Iterative modeling of alternatives
Potential exclusion issues
  • Routine code-compliance calculations
  • Funded research if the client pays regardless of outcome and owns the results
Documentation required
  • Model iterations
  • Design memos
  • Contract terms on risk and rights
  • Contemporaneous records of the uncertainty at the outset
  • Records of alternatives evaluated and results
  • Time allocation for personnel involved

Illustrative example. Qualification depends on the specific facts of each business component and is not automatic.

Example · Food & Beverage

Formulation testing

A food manufacturer sought to reduce sodium by 30% while maintaining texture and microbial stability through processing. Food scientists trialed alternative salt replacers and processing parameters.

Business component: Reduced-sodium product formulation

Potentially qualifying factors
  • Functional objectives (stability, texture under processing)
  • Structured bench and pilot trials
Potential exclusion issues
  • Taste-preference panels alone
  • Seasonal flavor variants
  • Packaging graphics
Documentation required
  • Formulation trial records
  • Lab analyses
  • Contemporaneous records of the uncertainty at the outset
  • Records of alternatives evaluated and results
  • Time allocation for personnel involved

Illustrative example. Qualification depends on the specific facts of each business component and is not automatic.

Example · Agriculture

Growing-condition experimentation

A grower tested irrigation schedules and substrate mixes to determine whether a crop variety could be grown profitably in a hotter climate zone.

Business component: Growing process for a crop variety

Potentially qualifying factors
  • Controlled plots with recorded variables
  • Reliance on agronomy and plant science
Potential exclusion issues
  • Routine farming practices
  • Weather-driven adjustments without experimental design
Documentation required
  • Plot designs and yield data
  • Contemporaneous records of the uncertainty at the outset
  • Records of alternatives evaluated and results
  • Time allocation for personnel involved

Illustrative example. Qualification depends on the specific facts of each business component and is not automatic.

Example · Medical Devices

Materials

A company evaluated polymer blends for a flexible housing that needed to withstand repeated sterilization cycles.

Business component: Device housing material

Potentially qualifying factors
  • Material trials with sterilization-cycle testing
  • Supplies consumed in trials
Potential exclusion issues
  • Selecting a material with known certified properties
Documentation required
  • Material trial matrices
  • Sterilization test data
  • Contemporaneous records of the uncertainty at the outset
  • Records of alternatives evaluated and results
  • Time allocation for personnel involved

Illustrative example. Qualification depends on the specific facts of each business component and is not automatic.

Evidence of experimentation

  • Test plans, protocols and results.
  • Design review minutes showing alternatives considered.
  • Version-control history, pull requests and benchmark results for software.
  • Simulation input files and outputs.
  • Prototype build records and failure analyses.
  • Lab notebooks and electronic lab records.

Practical examples of experimentation activities

  • Prototype iteration
  • Evaluating engineering alternatives
  • Simulation
  • Testing
  • Modeling
  • Design iterations
  • A/B evaluation of technical alternatives, where appropriate
  • Laboratory trials
  • Manufacturing trials
  • Evaluating software architecture alternatives

Frequently asked questions

Does the process of experimentation require a formal scientific method?

No formal lab setting is required, but the process must be designed to evaluate alternatives to resolve technical uncertainty and rely on scientific, engineering or computer-science principles. Structured, recorded trial and error can qualify.

What if only part of a project involves experimentation?

The 'substantially all' requirement is applied per business component. If less than 80% of activities are experimental, the shrinking-back rule may identify a subcomponent that qualifies.

Official sources

Explore State R&D Tax Credits

A neutral selection of state research, listed alphabetically. A federal credit does not determine a state credit: some state programs incorporate federal concepts, while others use state-specific definitions and calculations.

All state R&D tax credits · Compare state rules · State changes by year

General research and educational information, not legal or tax advice. Rules may vary by tax year and taxpayer-specific facts may affect results. See the disclaimer.