Most assessment is a proxy. Learners build understanding one way — investigating, arguing, solving — and then get measured another way, recalling facts on a test that looks nothing like the work itself. We’ve all made peace with that gap, but it’s worth naming how strange it is: we ask learners to show understanding in a format deliberately stripped of the context that made the understanding meaningful.
Our games close that gap. When you build an experience around a real problem and ask learners to actually solve it, the things they produce to move forward are the evidence. The act of learning and the act of demonstrating it become the same act.
So the real question is never whether assessment is happening. It always is. The question is what that evidence is for.
The constant: the work is the evidence
The clearest way to explain our model is to look at what learners actually do — and notice that every one of these is already evidence of understanding.
They model data to figure out what’s going on. A population is crashing; something in the environment shifted. Learners pull up the data, build or interpret a model, and identify the pattern that explains it. The model is a direct, observable artifact of their reasoning.
They build arguments from evidence. Learners assemble a claim, line it up against the evidence they’ve gathered, and commit to it — explaining a phenomenon or arguing for a path forward. Argumentation from evidence is famously hard to capture with a multiple-choice item, and trivially easy to see when someone is doing the actual arguing.
They make the diagnostic call. Given a water sample and a sick population, learners examine the candidates and select the cell most likely causing the illness. That single decision sits on top of a whole chain of reasoning — what they noticed, what they ruled out, what they trusted.
None of these are quiz questions in a game skin. They’re the genuine work of solving the problem — which is exactly why the evidence they throw off is richer and more honest than a detached test could capture. And this holds no matter where a game sits in your sequence.
The variable: what the evidence is for
What changes across our four use cases is what the evidence is for — and that depends on where a learner is in their learning. Are we watching someone reason their way toward an idea they don’t have yet? Or checking whether a skill they’ve already been taught holds up on its own? The same authentic work means something different depending on whether we expect a learner to know it yet — whether it’s a glimpse of an idea still forming, or a verdict on one that should be solid.
Introducing a concept (Discovery). Here learners meet an idea for the first time by figuring it out, not by being told. They model the data before they have the vocabulary; they reason toward an explanation no one has handed them. Evidence is everywhere — but it’s formative. You’d never grade it, because the whole point is productive struggle. A learner fumbling toward a model is doing exactly what this phase asks of them. Scoring it would punish the learning.
Exploration & exposure. Same low-stakes spirit, different goal: building interest and intuition, widening what a learner has encountered. The evidence tells you what’s landing and what’s sparking curiosity. Still formative, still ungraded.
Applied practice. Now the skill has been taught, and learners are using it where it matters — handling a real situation instead of an abstract exercise. The evidence here is gradeable. A learner constructing an evidence-based argument in applied practice is demonstrating a skill they’re expected to have, in context. That’s legitimate, defensible performance data you can stand behind.
Assessment. This is the dedicated case, and we think of it as a transfer task. The context is built specifically to host authentic assessment — whether it reuses the scenario from the learning unit or drops learners somewhere new. It answers the question that matters most: can a learner take what they learned and apply it somewhere they haven’t seen before? It’s summative by design. The meaningful context isn’t there to teach; it’s there to give the assessment somewhere real to live.
What changes, and what doesn’t
Two different things move across those use cases, and it’s worth being precise about both.
The experience itself is designed differently each time. An inquiry-driven discovery quest is built on a fundamentally different structure than an applied-practice scenario or a dedicated assessment task — we’re not dropping the same activity into four slots and swapping the stakes.
What stays constant is how we capture understanding. It’s always authentic — real context, a real work product, never a disconnected question pulled out of the situation that gave it meaning. We don’t get more "test-like" as the stakes rise. Where assessment usually swings from a quick informal check at the low end to a formal, decontextualized exam at the high end, ours doesn’t. High stakes just means a transfer task that’s still authentic and still in context, designed to stand on its own. The learner is always doing real work; what shifts is the experience we build around them and what we do with the evidence it produces.
Why a game, and not a test?
If the goal is just good assessment, it’s fair to ask why go to the trouble of building a game at all. The answer is that some of what a game offers here isn’t a nicer version of a test — it’s something a test can’t do.
You see the process, not just the answer. A traditional test records where a learner landed. A game records how they got there — the data they pulled, the paths they ruled out, where they backtracked and revised. So you can tell the difference between someone who guessed their way to a right answer and someone who reasoned carefully to a wrong one. The reasoning becomes visible, and the reasoning is usually the thing you actually care about.
You can see whether learning transfers. A test can tell you a learner reproduced what they were taught. A game can drop them into a genuinely new situation and show you whether they carry the skill across — generalizing and applying it somewhere they haven’t seen before. That’s the line between someone who memorized something and someone who understands it well enough to use it.
You measure the skill directly, not a proxy for it. When you want to know whether a learner can build an argument from evidence, you have them build one. The task is the skill, not a written stand-in for it. That closes the gap between what you’re trying to measure and what you’re actually observing — a cleaner signal, and one plausibly less distorted by the anxiety and reading load that come with a decontextualized exam.
You get authentic performance at scale. Rich performance assessment has always forced a tradeoff: a lab practical or a portfolio captures real capability, but it’s slow, costly, and inconsistent to score. A game delivers that same authentic, performance-based task to every learner under identical conditions, with evidence that comes out structured and consistent.
You assess knowledge in use — which is what standards increasingly ask for. Across subjects, standards have moved away from recall and toward application: learners using practices like modeling, analyzing data, building arguments, and solving problems to make sense of real situations. The examples earlier aren’t just engaging activities — they are those practices, performed and captured. Assessing this way isn’t a workaround; it’s a direct match to where the standards are already headed.
From work product to instructor insight
Authentic evidence is only useful if an instructor can use it to make decisions. Everything a learner produces exists at the work-product level — the model they built, the argument they assembled, the call they made. We can help convert that into a clear, standards-aligned readout an instructor can scan, with the underlying work product available when they want to dig into an individual learner’s thinking.
This matters most in applied practice and assessment, where the evidence is carrying real weight. But it means an instructor always has both the quick signal and the full picture.
Why this matters for what you build with us
When assessment is the same act as learning, the evidence comes built into the experience — there’s no parallel test bank to write, maintain, and keep aligned as the work evolves.
And because that evidence is authentic wherever it shows up, you get to decide its role. Design an inquiry quest and let the evidence inform you quietly while learners are still figuring things out. Design an applied-practice scenario and grade the skill they’re now expected to have. Design a transfer task to certify that the learning stuck. Different experiences, different stakes — matched to where your learners actually are.
That’s the whole idea. Set up a real problem, let learners do the real work of solving it, and the evidence is already there — authentic, in context, and ready to do whatever job the moment calls for.