Most people who run a systematic review end up with the same setup by accident. A reference manager holds the search results. A separate screening tool tracks include and exclude decisions. Extraction lives in a spreadsheet that grows a new tab every week. Then the numbers move into whatever statistics package you learned in grad school, and the figures get rebuilt by hand for the manuscript.
Every one of those handoffs is a place for something to go wrong. A study drops out between screening and extraction and nobody notices until a reviewer asks. The effect sizes in the spreadsheet don't match the ones in the forest plot because someone updated one and not the other. The PRISMA counts stop adding up. None of this is exotic. It happens on careful, well-run reviews, because the work is spread across tools that were never meant to talk to each other.
I built Verflux to keep the whole thing in one place, from the first import to the figures you drop into the paper.
What the workflow actually looks like
You bring in your search results however you already have them. Verflux reads RIS exports from the databases, and it pulls folders straight from Zotero, so you're not re-exporting and re-formatting between rounds. Duplicates get flagged on the way in.
Screening is built for the volume you actually deal with. Title and abstract decisions move quickly, and when a search comes back bloated with obviously irrelevant records, you can clear them in bulk instead of clicking through a thousand at a time. Every decision stays logged, which is what feeds an honest PRISMA diagram later rather than a reconstructed guess.
Risk of bias is where the tool-awareness matters. Verflux understands the structure of the common appraisal tools, so importing your assessments lines up with the right domains instead of forcing everything into a generic table. When you get to reporting, the GRADE summary and the PRISMA flow diagram export as figures you can use, not screenshots you have to apologize for.
The part I care about most is the math
This is where I have opinions. A meta-analysis platform is only worth using if you can trust its numbers and check its work, and a surprising number of tools ask you to take the output on faith.
Verflux uses the standard, published estimators, and I can tell you exactly which ones. Standardized mean differences use Hedges' g with the small-sample correction from Hedges (1981), not the uncorrected d that quietly inflates effects in small studies. Random-effects pooling uses DerSimonian and Laird (1986). Prediction intervals follow Borenstein and the Higgins-Thompson-Spiegelhalter approach, so they describe the spread of true effects across settings rather than pretending your point estimate is the whole story.
I audited these formulas against the primary sources rather than against other software, because copying another tool's output only propagates whatever that tool got wrong. If you want to recompute a pooled estimate by hand to satisfy yourself, you can, and the pieces will line up.
That matters for a reason beyond correctness. A systematic review is supposed to be reproducible. If a reader can't tell which estimator produced your forest plot, they can't reproduce it, and the review is weaker for it. Being explicit about the methods isn't a compliance checkbox. It's the point of the whole exercise.
Who this is for
Verflux fits the person doing the review, not a large team with a methodologist on staff and a budget for enterprise software. If you're a PhD student running your first meta-analysis, a postdoc synthesizing a literature that nobody has pulled together yet, or a clinician doing a rapid review between shifts, the tool assumes you know your field and don't want to fight your software.
It won't design your search strategy or decide your inclusion criteria. That judgment is yours, and it should be. What it does is remove the friction between the parts of the work you actually trained for, and it keeps the record clean enough that when a reviewer asks how you got a number, you have an answer.
Try it on a review you're already doing
The honest test of a tool like this is whether it survives contact with real data, so use it on a project that's already underway. Import your current search, screen a batch, and pull a forest plot. If the numbers match what you'd compute yourself and the figures are ready for the manuscript, you'll know within an afternoon whether it belongs in your workflow.
That's the whole pitch. Fewer handoffs, statistics you can verify, and figures you don't have to rebuild by hand.
Start a review at verflux.com
Start your systematic review today
All analytical features included in the free trial. No credit card, no installation, no R or Python.
Create free account
Trial · 1 project · Full meta-analysis engine · Demo data included