Research Tool Bench

Research tools

Last updated 2026-08-16

Cluster · Independent benchmarking and decision tools for evidence-synthesis software

Choosing screening software is a decision with a real cost. Pick a tool that misses relevant records and you damage the review. Pick one your budget cannot renew and you lose the project mid-screening.

The information you would need to decide well is scattered. Vendor pages describe features but not performance. Academic papers report performance, but on datasets and metrics that differ from one another. Prices change without announcement. This cluster holds those three things in one place and keeps them separate rather than blended into a single misleading score.

Start here

Systematic Review Software Selector

Enter your corpus size, team, budget and workflow needs. It applies a published rule set and shows you which tools were eliminated and exactly why.

Pricing & Feature Tracker

Current plans, prices and stated limits, taken only from vendor pages, each with the date it was checked. Includes the trial limits that decide what can actually be tested.

How we test

Our pre-registration, published before we run anything: the three evidence layers, why an unmeasured cell stays empty instead of being estimated, why a single simulation run is never published, and the limits of what we measured.

Benchmark status

Where the screening-performance runs stand. Nothing is published there yet, and the page says so rather than implying otherwise.

Where this cluster currently standsThe screening-performance runs are not published yet. Until they are, this cluster offers the pricing tracker, the selector, and the protocol we committed to in advance. Nothing here estimates a performance number we have not measured.

Why an empty cell is the point

Most "best screening software" pages compare marketing copy and fill every cell. We would rather publish a smaller table with an honest boundary drawn around it. Where a tool's trial limits prevent a full run under our protocol, its performance cell stays empty and saysNOT DIRECTLY TESTED, with the reason. That is not an omission — it is the finding.