Comparative analysis prompt builder
Asking an LLM to “compare X and Y” usually yields a lopsided, hand-wavy answer that quietly favors whatever it has seen most. This tool builds a prompt that forces a rigorous, even-handed comparison: explicit dimensions, optional weighting, bias-avoidance rules, and a structured output you can act on. It is assembled locally in your browser.
How it works
You list the items to compare and, optionally, the dimensions to evaluate them on — or let the model pick the decision-relevant ones and justify them. You choose equal weighting or importance weighting, and an output format: comparison table, pros and cons, weighted scorecard, or narrative. The builder produces a prompt that tells the model to apply one consistent standard across all items, mark unknowns honestly, separate fact from opinion, and finish with a context-specific “Best for…” line per item.
Why a plain “compare X and Y” prompt falls short
A bare comparison request leaves the model free to:
- Emphasize dimensions that happen to suit the most-covered option
- Invent or estimate figures it does not actually know
- Apply different standards to different items (“X’s weakness is minor, Y’s is major”)
- Declare a winner without accounting for the reader’s context
The generated prompt closes each of these gaps explicitly. It instructs the model to justify each dimension it uses, apply the same scoring standard to every item, mark missing information as “unknown” rather than filling the gap, separate what is documented fact from what is opinion, and produce a “Best for…” recommendation that acknowledges different readers have different priorities.
Output format options
Depending on the nature of your comparison, one format will serve better than another:
- Comparison table — best when items share a clear set of attributes and you need a scannable grid for a team review or a report.
- Pros and cons — useful for two or three options when the audience is less technical and wants a narrative feel.
- Weighted scorecard — best for high-stakes decisions where factors genuinely differ in importance. The prompt asks the model to assign weights explicitly and show its arithmetic.
- Narrative — best when the comparison is nuanced or context-dependent and a flat table would misrepresent the trade-offs.
An illustrative use case
Suppose you want to compare three cloud database services for a startup project. You enter the three service names, specify dimensions such as managed-service overhead, pricing model, scaling behavior, vendor lock-in, and community support, and choose importance weighting with a scorecard output. The generated prompt tells the model to: (1) score each dimension 1–5 for every service, (2) weight each dimension by importance and show the totals, (3) flag any cells where it lacks current knowledge, and (4) close with a “Best for…” line for each service — for example “Best for teams that prioritize zero operational overhead” versus “Best for teams with strict cost budgets.”
You paste the prompt into your LLM, get a structured analysis back, and then verify the time-sensitive cells (current pricing, version-specific features) against official documentation before making a decision.
Tips and notes
Choose dimensions that actually drive your decision; a comparison on irrelevant axes looks thorough but does not help you choose. Importance-weighting is worth it when the factors clearly differ in significance — for example, price might matter far more than a minor feature. Because model knowledge can be out of date, always verify time-sensitive cells like current pricing or version numbers against primary sources before relying on the verdict. The per-item “Best for…” output is especially useful when you are sharing the analysis with others who may weigh the trade-offs differently than you do.