Put WhimsyLabs Through the Checklist
Bring your science team, run the practicals you know best, and see how the results compare.
Book a Demo
How to Choose Virtual Lab Software: A Buyer’s Guide for Schools
The right virtual lab depends on what you want students to learn. If the goal is genuine practical skill, the two questions that separate platforms are: does the software simulate real physics or play scripted animations, and does it assess the process students follow or just their final answers? This guide gives you a working checklist, the questions to put to vendors, and an honest look at where WhimsyLabs fits.
What Should Schools Look for in Virtual Lab Software?
Look first at simulation depth, assessment method, accessibility, data protection, and total cost — in that order. Curriculum coverage and polish matter, but they can’t compensate for a platform where students click through predetermined steps without ever practising a technique.
The pressure is real. The Royal Society’s Science Education Tracker 2023 found that the share of GCSE pupils doing practical work at least fortnightly fell from 44% in 2016 to 26% in 2023, and EngineeringUK’s 2024 research reports that 27% of UK science teachers say their schools cannot afford the equipment needed for practical lessons. Virtual labs can close that gap — but only if the one you pick builds skills rather than replaying content.
Simulation Depth: Animation or Physics?
Ask what happens when a student does the experiment wrong. That single question sorts the market into its main categories:
- Video-based tools show recordings of real experiments. Useful for introducing concepts, but watching a titration is not performing one.
- Guided step-by-step simulations walk students through predetermined sequences. Clicking the “wrong” option typically triggers an error message and returns them to the correct path, so the cause-and-effect link between action and outcome is lost.
- Physics-engine sandboxes calculate outcomes from physical laws. Add acid too fast and the pH overshoots; contaminate a sample and the analysis shows artefacts. Mistakes produce realistic consequences students must diagnose.
Research supports the deeper approach: Finkelstein et al. (2010) found physics-based simulations significantly improve conceptual understanding compared with simplified animations, and Kapur (2015) reported that productive-failure approaches can roughly double the effect of direct instruction alone. If a vendor cannot explain their physics engine, assume there isn’t one.
How Should the AI Assessment Work?
It should assess what students do, not parse what they write. AI systems that grade written answers must interpret ambiguous language, and they make mistakes doing it. Assessment built on physical actions has nothing to misinterpret: the student either pipetted 2.5 mL or they didn’t, either rinsed the burette or skipped it.
Process-based assessment also resists AI-assisted shortcuts. When follow-up questions are tied to a student’s own experimental data — “why does your graph show an unexpected peak?” — a chatbot can’t answer, because it doesn’t know that the student forgot to rinse the burette three steps earlier.
Finally, check the AI safety design. Does the tutor expose a free-text chat window to pupils, or does it work purely from their actions in the lab? A tutor with no student chat surface can’t deliver generated text to children or be prompt-injected by them — a much easier system to safeguard.
The Virtual Lab Evaluation Checklist
Put every shortlisted platform through these twelve criteria. A trial with your own teachers and students will reveal more than any demo.
- Physics simulation, not canned animations — wrong technique should produce wrong (but realistic) results, not an error message.
- Emergent, noisy data — results generated from the student’s actions, so they practise handling outliers and error, not perfect pre-canned graphs.
- Process-based assessment — grading of technique, sequence, and safety compliance, not just final answers or multiple-choice questions.
- AI safety by design — no open student chat window; guidance inferred from actions, with teachers able to review and override AI judgements.
- SEND accessibility — control remapping, text-to-speech, self-paced modes, and documented accessibility support.
- Low-bandwidth mode — keeps lessons running on unreliable school internet connections.
- Device coverage — works on the hardware you already own (Chromebooks, PCs, Macs), with VR as an option rather than a requirement.
- Clear data protection answers — where data is stored, whether it trains AI models, and what deletion rights you have. Vague assurances are a red flag.
- Pricing transparency — a stated pricing model, no long lock-in contracts, and a real trial with your own students before you commit.
- Curriculum fit — mapping to your exam board’s required practicals, not a generic library built for a different education system.
- Teacher dashboards and LMS export — per-student technique data you can act on, exportable reports, and integration with your existing systems.
- Teacher customisation — the ability to adjust experiments, scaffolding, and assessment focus rather than being locked to fixed content.
What Does Virtual Lab Software Cost?
Pricing models vary — per-student, per-seat, or site licences — so compare total cost of ownership, not headline price. Factor in training, integration, any required hardware, and what happens to your data if you don’t renew.
For context on the physical side: CLEAPSS (2024) puts a traditional school chemistry lab setup at £25,000–£70,000 initially, with £5,000–£15,000 a year in consumables, maintenance, and compliance. Virtual labs don’t replace physical labs, but they let students repeat experiments without consuming materials — which is where the budget case usually lands.
If cost is the blocker, funding routes exist. UK schools have used programmes such as Royal Society Partnership Grants to fund virtual lab subscriptions — see our guide to grants and funding for what’s available and how to apply.
What Should You Ask Vendors About Data Protection?
Ask where student data is stored, whether it is encrypted, whether it trains AI models, and how deletion works when you leave. EdTech platforms hold sensitive information about children, and high-profile breaches have made scrutiny here non-negotiable. Request the vendor’s data processing agreement up front — if they don’t have one ready, walk away.
You can see how WhimsyLabs answers these questions on our data security page: student data is never used to train AI models, and the platform is built for GDPR compliance.
How Does WhimsyLabs Score Against This Checklist?
WhimsyLabs is one option in the physics-engine sandbox category, and we built it around the criteria above: a real-time physics engine with emergent, noisy data; process-based assessment of technique and safety; an AI tutor (WhimsyCat) with no student chat window; SEND accessibility features and a low-bandwidth mode; and delivery on Chromebooks, desktops, and VR headsets alike.
We won’t claim it’s the right choice for every school. If you mainly need short concept demonstrations, a lighter guided-simulation tool may serve you fine. If you want students to build practical technique and to assess how they work, we think WhimsyLabs compares well — and the fair test is to run your own experiments on a trial and judge the results.
Go Deeper: Buyer’s Guides and Research
This page summarises our longer guides. For the full detail:
- Virtual Lab Software for Secondary Schools: A Buyer’s Guide — vendor questions and red flags for GCSE and A-Level.
- Virtual Labs in Higher Education: A 2026 Buyer’s Guide — solution categories and making the case to leadership.
- Why Other Virtual Labs Fail: The Physics Engine Solution — the animation fallacy and emergent data, in depth.
- Why AI Cannot Misgrade a Pipetting Technique — process-based assessment versus text-parsing AI.
- Virtual vs Physical Labs: A Cost-Benefit Analysis — the full numbers behind the budget case.