“How Much Will They Buy?” Why Adoption Studies Aren’t Volumetric Forecasts

When teams talk about “demand forecasting,” the conversation often jumps straight to adoption: How many customers will try it? How fast will the market pick it up? What share will we capture? Those are important questions—especially for new products—but they’re only part of the story. The forecasting mistake that shows up again and again is treating an adoption study as if it were a volumetric forecast.

Adoption research can tell you a lot about who is likely to try, why they might switch, and what could accelerate or slow uptake. What it usually cannot do on its own is tell you the thing operations and finance ultimately need: How much will they buy? Volume is driven by a different set of mechanics—some overlapping with adoption, many not.

The key is to recognize what adoption studies measure well, what they don’t, and how to bridge the gap without pretending the data is more precise than it is.

Adoption answers “will they try?” Volume answers “how does usage accumulate?”

Adoption is about the probability of starting. Volume is about the quantity over time—how consumption or purchasing accumulates once a product is in market. You can have strong adoption with weak volume if trial doesn’t convert to repeat, if usage is occasional, or if the product is purchased infrequently. You can also have modest adoption with strong volume if the buyers who do adopt are heavy users, if reorder cycles are short, or if the product becomes embedded in workflow.

That’s why “adoption percentage” is a risky proxy for “demand.” It’s a component of demand, but it isn’t demand.

The hidden variables adoption studies often miss

Adoption research tends to focus on perceived value, willingness to try, barriers, and what it would take to switch. All of that is useful. But volume is heavily influenced by variables that are easy to overlook if you’re staying at the “intent and preference” level.

For example, volume depends on purchase frequency. Are customers buying weekly, monthly, quarterly, or once a year? It depends on usage rate. Is this a product used daily, occasionally, or only in specific situations? It depends on pack size and replenishment behavior. Do buyers stock up? Do they trial a small size first? Do they buy in bulk once they commit?

Then there are structural multipliers that have nothing to do with your product’s concept score: distribution and availability, channel constraints, contracting and procurement, budget cycles, training and onboarding time, and switching costs that slow down “real” adoption even when stated interest is high. And of course, competitive response can reshape volume faster than it reshapes stated preference.

If those drivers aren’t modeled explicitly, a forecast based on adoption alone will almost always feel too clean—and often too optimistic.

The “trial-to-repeat” cliff is where volume forecasts break

One of the most common volume surprises is the gap between trial and repeat. Early adoption studies can capture curiosity and initial interest, but they’re less reliable at predicting what happens after first use—when friction, habit, and competing priorities show up.

This is where forecasting teams can unintentionally bake in a fatal assumption: that trial equals ongoing usage. In reality, many launches produce an early spike followed by a drop-off as customers revert to old behaviors or discover that the product isn’t as easy to integrate as it looked in concept. If you don’t explicitly model repeat rate, reorder intervals, and retention, you can end up with a forecast that “works” on paper but collapses in the field.

Bridging the gap: how to turn adoption research into volumetric inputs

The fix isn’t to abandon adoption research. It’s to use it for what it does best, then add the missing pieces deliberately.

Start by treating adoption outputs as probability inputs, not volume outputs. Use them to estimate who enters the funnel, not how much the funnel produces. Then layer in a simple usage model that forces the real drivers onto the table. You don’t need a complicated architecture to start—just a clear decomposition of demand into components you can sanity-check:

• How many buyers are eligible?

• What share becomes aware and has access?

• What share trials?

• What share converts to repeat?

• What is the average quantity per purchase?

• How often do they repurchase?

• How long do they stay active?

Once you frame demand this way, you can pressure-test the assumptions. Some can be informed by research (workflow fit, switching barriers, likely use cases, category norms). Others can be anchored with analogs (similar products, comparable launches, historical reorder patterns). The key is to make the assumptions explicit so you can refine them quickly as early sales data arrives.

Ship ranges, not certainty

Because volume relies on multiple uncertain inputs, single-number volume forecasts can be especially misleading. A better approach is to publish a range tied to the assumptions that matter most. Sensitivity analysis can show you where uncertainty actually lives—maybe it’s repeat rate, not trial. Maybe it’s distribution, not interest. And if you run simulations across plausible values, you’ll produce ranges that match reality better than any point estimate.

Adoption studies are incredibly valuable, but they’re not volumetric forecasts. If you want a forecast that operations and finance can trust, treat adoption as the starting line—not the finish. The real work is translating “will they try?” into “how does demand accumulate?”—and being honest, with ranges, about what you can know before the market teaches you the rest.