Quantum computing is not automatically the right tool for every difficult problem. This feasibility checklist helps you compare a proposed quantum approach with strong classical and hybrid alternatives by scoring problem structure, data movement, algorithm maturity, hardware requirements, validation cost, and the evidence needed to justify a pilot.
Overview
A useful quantum computing use case is more than a problem that is large, complex, or expensive. It must have a structure that a quantum algorithm can represent, a measurable outcome, and a credible path from an abstract circuit to a useful workflow. The right question is not simply, “Can a quantum computer run this?” It is, “Can a quantum workflow produce a valuable result at an acceptable total cost compared with the best available classical method?”
This distinction matters because quantum experiments often include costs that are easy to overlook: converting business data into a circuit, transferring data to a device, repeating measurements, managing noise, interpreting probabilistic output, and integrating the result into an existing application. A small circuit can still be a poor business solution if these surrounding steps dominate the workflow.
Use the assessment below as a repeatable first-pass filter. It is suitable for optimization, simulation, machine learning, cryptography-related analysis, and research prototypes. It does not prove quantum advantage. Instead, it identifies whether a proposal deserves simulation, benchmarking, a hardware trial, or a pause while the classical baseline is improved.
How to estimate
Start by writing the use case as a measurable decision rather than a technology statement. For example, “select a portfolio under constraints while minimizing risk” is more useful than “apply a quantum optimizer.” Record the current method, the input size, the output required, and the business or engineering metric that determines success.
Then score the proposal from zero to four across eight dimensions:
- Problem structure: Does the problem map naturally to a known quantum formulation, such as a Hamiltonian, amplitude-estimation task, or constrained binary objective?
- Input cost: Can the required information be encoded without spending more time or resources than the proposed speedup could recover?
- Output cost: Is the answer compact and useful, or would extracting a complete solution require many measurements?
- Objective clarity: Can success be expressed as a precise function, constraint, energy, probability, or error metric?
- Algorithm maturity: Is there a documented algorithmic route, or would the project require inventing and validating most of the method?
- Hardware fit: Are the expected qubit count, connectivity, circuit depth, precision, and error sensitivity compatible with the target execution environment?
- Classical baseline: Has the proposal been compared with a competitive solver, heuristic, tensor method, sampling method, or specialized hardware approach?
- Hybrid workflow: Can classical and quantum components exchange data efficiently, with a clear division of responsibilities?
Give each category a score from 0, meaning no credible path has been shown, to 4, meaning the requirement is well defined and supported by a testable plan. Add the scores for a maximum of 32 points, but also apply a gate rule: a low score in input cost, output cost, classical baseline, or hardware fit should prevent a high total score from being interpreted as readiness.
A practical interpretation is:
- 0–10: Keep the idea at the research or education stage. Clarify the problem and improve the baseline before requesting hardware access.
- 11–19: Build a small simulator experiment. The goal is to test representation, scaling assumptions, and measurement requirements.
- 20–26: Run a controlled benchmark against classical alternatives and, where appropriate, a hardware trial.
- 27–32: Consider a focused pilot, provided no gate category is weak and the value of the result is independently defined.
These ranges are decision aids, not industry standards. Keep the individual scores and explanations so another reviewer can challenge the assumptions.
Inputs and assumptions
The quality of the assessment depends on the inputs recorded before implementation. Capture the following in a one-page worksheet.
1. Problem and success metric
Describe the decision, not just the dataset. Specify whether the desired output is one candidate solution, a ranked list, an estimated quantity, a probability distribution, or a physical property. Define the acceptable error and the minimum improvement that would justify adoption.
2. Data loading and movement
Document where the data starts, how it is encoded, how often it changes, and whether the quantum routine must receive the full dataset or only a compressed representation. A theoretical algorithmic improvement may have little practical value if state preparation, preprocessing, or repeated transfer dominates execution.
3. Algorithm and circuit assumptions
Name the proposed method and its role. Examples include a variational workflow for estimating an energy, a QAOA-style formulation for a combinatorial objective, or a quantum simulation routine for a structured physical model. Record the number of logical qubits, circuit layers, repetitions, parameter updates, and classical optimization steps. If these values are unknown, mark them as estimates rather than presenting them as requirements.
For background, readers can review the VQE tutorial, QAOA Max-Cut tutorial, and quantum gates reference before completing this section.
4. Baseline and measurement plan
Choose a classical baseline that is strong enough to be credible. Include its runtime, memory use, solution quality, approximation gap where available, and operational constraints. Decide in advance how many runs or samples are needed, how uncertainty will be reported, and which result will count as a win. A single favorable run is not a reliable benchmark.
5. Total cost model
Estimate total effort as:
Total quantum workflow cost = modeling + data preparation + compilation + execution + repetitions + error handling + post-processing + integration.
Use the same cost units for the classical comparison where possible: elapsed time, compute resources, engineering hours, energy, or cost per decision. If some values are unknown, use low, central, and high scenarios instead of one falsely precise estimate.
Worked examples
Example A: Constrained scheduling
Suppose a team wants to assign jobs to machines while respecting capacity and timing constraints. The problem may map naturally to a binary objective, so problem structure and objective clarity could score highly. However, input and output costs may be weaker if the schedule changes frequently or if the complete assignment must be measured and reconstructed repeatedly.
The appropriate next step is not an immediate hardware claim. Build a small instance, encode the constraints, and compare the quantum formulation with an established classical solver. Test whether solution quality remains meaningful as the number of variables and constraints grows. A QAOA experiment can be educational and technically valid even if it does not yet beat the baseline.
Example B: Molecular energy estimation
A molecular simulation proposal may have a clearer connection to quantum simulation than a generic machine-learning task. Still, the assessment must include basis selection, active-space reduction, circuit depth, measurement overhead, optimizer behavior, and the accuracy required by the downstream decision. A variational method may be a reasonable prototype route, but the result should be validated against a trusted classical calculation for the same reduced instance.
In both examples, the strongest deliverable is a reproducible comparison rather than a claim based on the use of quantum terminology. Store the problem instance, software versions, circuit description, baseline configuration, random seeds where relevant, and raw results. The reproducible experiments workflow provides a practical structure for doing this.
When to recalculate
Revisit the assessment whenever a major input changes. Recalculate after a new classical algorithm or baseline becomes available, when hardware access conditions or execution costs change, when an algorithm is redesigned, or when the problem size and data cadence move beyond the original assumptions. Also review the worksheet if error-mitigation overhead, compilation behavior, or measurement requirements materially change the circuit.
For an active project, schedule a review at each milestone: problem formulation, simulator prototype, noisy simulation, hardware experiment, and production-readiness decision. At every review, preserve the previous score and explain what changed. This prevents optimistic assumptions from quietly becoming project requirements.
Your next action should be simple: choose one candidate use case, complete the eight scores, write down the classical baseline, and identify the weakest gate category. If the weakest category is data loading, output extraction, or baseline quality, improve that analysis before increasing circuit size. If the proposal survives those checks, run the smallest experiment that can falsify its central assumption. Quantum computing becomes a more useful engineering option when it is evaluated as part of a complete workflow, not as a substitute for one.