For funds
Unit demand for a launch: how simulated consumer surveys estimate it
TL;DR
- Unit demand for a launch is buyers times units per buyer, spread over time. Intent alone is only the first of those, which is why a high intent number so often turns into a missed unit line.
- A Neon Blue study asks the likely buyers four things: would you buy, when, how many, and what would you replace or stop buying. For anything bought more than once, it also asks about the second purchase.
- The output is a set of model inputs by segment: buy-in-launch-window share, units per buyer, repeat intent and the cannibalized product. It is a labeled input, not a stock call.
- Results come as a report. An ongoing data stream and an API are in build. The first study is free.
Neon Blue builds simulated audiences of a company's customers and market, AI models that answer the way a defined group of people would, grounded in Neon Blue's own paid studies and AI-moderated interviews with real people. For funds, a launch study turns those answers into the inputs behind a unit forecast: who buys, how many, how soon, and what they stop buying.
What is unit demand for a launch, and why is intent not enough?
Unit demand is how many of a new product sell in a given period, usually the launch quarter and the two after it. A purchase intent read for a new product tells you how many people like the idea. A unit forecast needs more:
- Timing. Buying in the first month and buying "at some point" are different revenue quarters.
- Units per buyer. One device per household, or one per person. One trial pack, or a case.
- Repeat. For snacks, drinks and beauty, the second purchase decides whether the launch holds after the first shelf reset.
- Replacement. A new product often takes its units from the company's own older line. Units up, revenue flat.
A model that multiplies intent by an addressable population skips all four. The study asks each one directly.
How does a Neon Blue study estimate launch units?
- Define the likely buyers. Current customers of the brand, buyers of the category from rivals, and people who have never bought the category. Each group gets its own cell because their conversion is very different.
- Show the product as it will be sold. The real description, the real price and where it will be sold. Leaving out the price inflates every answer.
- Ask when. In the launch window, within six months, later, or never.
- Ask how many. Units on the first purchase, and for consumables, how often they would buy again if they liked it.
- Ask what it replaces. An older model from the same company, a rival product, or nothing, which is new spend.
- Combine by segment. Multiply launch-window share by units per buyer and by the segment's size, then subtract the units that come from the company's own line.
Stated intent runs high, so the useful parts are the shape of the answer: how much is front-loaded, how much repeats, how much is cannibalized. How simulated audiences work explains how the audience is grounded.
What questions go into a launch unit study?
| Question | Model input | Why it matters |
|---|---|---|
| At this price, would you buy this? | Conversion by segment | The starting point, read as relative size |
| When would you buy it? | Timing curve | Puts units in the right quarter |
| How many would you buy at first? | Units per buyer | Turns buyers into units |
| If you liked it, how often would you buy it again? | Repeat rate | Decides whether a consumable launch lasts |
| What would you stop buying, or replace, to buy this? | Cannibalization | Separates new units from moved units |
| What would stop you buying it in the first month? | Risk to the curve | Price, availability or waiting for reviews |
What does a launch unit study look like?
Hypothetical: a packaged food company has just launched a protein snack line. The analyst's model assumes a strong first two quarters, but nobody knows how much of it repeats or how much comes from the company's existing bars.
- Study question: "This high-protein snack is $3.49 a pack at your usual grocery store. Would you buy it in the next month, in the next six months, later, or not at all? If you tried it and liked it, how often would you buy it again?"
- Audience: current buyers of the company's existing snack bars, buyers of rival protein snacks, and grocery shoppers who don't buy protein snacks today, cut by age and shopping frequency.
- Follow-ups: how many packs on a first trip, and what would come out of the cart to make room.
Example output layout (illustrative, not a real result)
The shape of a Neon Blue launch read. Cells are blank on purpose; a real study fills them.
| Segment | Buy next month | Buy within 6 months | Packs on first trip | Repeat monthly or more | Replaces company's own bar |
|---|---|---|---|---|---|
| Current bar buyers | __% | __% | __ | __% | __% |
| Rival protein snack buyers | __% | __% | __ | __% | __% |
| Non-buyers of the category | __% | __% | __ | __% | n/a |
How the analyst reads it: if current bar buyers lead on trial but say the new pack replaces their usual bar, much of the launch is moved units. If rival buyers show lower trial but higher repeat, the launch may build more slowly and last longer. The analyst splits the unit line into new and moved units, labels both as simulated, and checks the trial curve against what the company reports. Brand switching covers how to read the units taken from rivals.
Want this run on your own launch? Neon Blue's first study is free: start a free first study.
When should a launch study run?
Twice is better than once. Run it as soon as the product, price and launch date are public, to set the first unit line. Run it again a few weeks after launch to see whether people who said "later" have moved forward and whether triers plan to repeat. Both reads land well before the company reports the quarter. Pre-earnings consumer checks lays out the timing, and how to test a key assumption in your model shows how to narrow a wide unit range.
Which launches does this work for?
- Consumer staples: new flavors, formats and lines, where trial and repeat matter most. See consumer research for staples analysts.
- Consumer electronics and devices: where the timing curve and the upgrade from the older model decide the quarter.
- Gaming and subscriptions: where a launch can pull new subscribers or just move existing ones to a higher tier. See subscriber churn for the other side.
- Restaurants: limited-time items, covered in consumer research for restaurant analysts, where the question is whether the item brings visits or just changes what regulars order, which shows up in average ticket and basket size instead of traffic.
Launches with no close category are the hardest, because the grounding covers less of them. Those questions get flagged.
What are the limits of a survey-based unit estimate?
- It assumes people know the product exists. Awareness, distribution and ad spend decide how much intent turns into units, and the study can't see a rollout plan.
- Supply limits are invisible to it. A sold-out launch reads as weak units for reasons no buyer would mention.
- Stated timing is optimistic. Read the split between "next month" and "later" across segments, not as a literal calendar.
- Neon Blue tests simulated answers against real responses on attitudinal and intent questions. Treat the read as one directional input, not a settled number. Can a simulated consumer survey feed an investment model? covers how to label and weight it.
Questions about estimating unit demand for a launch
How is unit demand different from purchase intent?
Purchase intent is the share who say they would buy. Unit demand adds when they buy, how many they buy, whether they buy again and what they stop buying. A forecast needs all five.
How do you estimate repeat before anyone has tried it?
Ask what people would do if they tried it and liked it, then compare that across segments. Run a second study after launch with people who say they tried it.
Can the study measure cannibalization?
Yes. Ask what the new product would replace. Answers that name the company's own older product are moved units, not new ones.
Does this predict a company's unit sales?
No. It gives inputs for a unit line by segment, labeled as simulated. Neon Blue doesn't make stock calls or give price targets.
How do I start a free first study?
Bring the launch, its price, and the unit assumption in your model you trust least. Neon Blue's first study is free. Start a free first study