返回提示词目录第 1439 页

新产品需求预测与上市规划

综合类比、市场漏斗和早期信号,输出分情景需求区间、上市曲线及复盘规则。

完整提示词

<context>
New-product forecasts are mostly wrong, and the job is to be wrong by less and find out fast. Experienced planners triangulate rather than trust one method: analogue forecasting (scale the launch curve of similar past products by differences in price, distribution and marketing), a market funnel (people reachable times awareness times trial times repeat), and early signals (pre-orders and waitlists, which overstate demand because sign-ups cost nothing and need a conversion discount). Trial and repeat behave differently: launch spikes come from trial, the long run depends on repeat. Distribution and supply limit sales before demand does, so a forecast must separate demand from what can be shipped. New products also cannibalise existing ones. Stated purchase intent from surveys overstates real buying, often heavily. The useful output is a range, a launch curve shape, and rules for updating as soon as real sales arrive.

Product: {{product}}
Horizon: {{horizon}}
{{#analogues}}

<analogues>
{{analogues}}
</analogues>
{{/analogues}}
{{#signals}}

<signals>
{{signals}}
</signals>
{{/signals}}
</context>

<task>
1. List what is known (price, channels, launch date, reach) and the assumptions you must make, each marked as an assumption.
2. Build three independent estimates for the first full period of the horizon:
   a. Analogue: pick the closest analogues, say why each fits or does not, and scale them with stated adjustments.
   b. Funnel: reachable people or outlets times the conversion rates at each stage, with the source of each rate or a clearly labelled assumption.
   c. Signals: convert pre-orders, waitlists or intent into expected purchases with an explicit discount and why.
   If a method has no inputs, say so and do not fake it.
3. Triangulate: compare the estimates, explain why they differ, and give a low, central and high forecast for the whole horizon. The range should be wide; say what it would take to land at each end.
4. Shape the launch curve over the horizon: initial trial spike or slow build, the repeat rate that sets the steady level, seasonality, and the effect of planned marketing bursts. Separate cannibalised sales from net new sales if the product sits next to existing ones.
5. Check supply and distribution: flag any period where planned stock or outlets cap sales below the demand forecast.
6. Set re-forecast rules: which early metrics to read (week 1 to 4 sell-through, repeat rate, rate of sale per outlet), how they map onto the low, central or high path, and when to reorder or cut.
7. List the assumptions to test before launch, most consequential first.
</task>

<constraints>
- Never present a single number without the range. Never use survey intent at face value.
- Do not invent analogue sales figures or market sizes. If none are given, use clearly labelled placeholders and say what data would replace them.
- Keep units consistent (units versus revenue, sell-in to retailers versus sell-through to customers) and say which one each number is.
- Say plainly when the evidence is too thin for anything better than an order-of-magnitude estimate.
{{> output/uncertainty}}
</constraints>

<output_format>
## Summary
The low, central and high forecast for the horizon, the main driver of the range, and the first metric to watch.

## What we know and assume
Two bullet lists.

## Three estimates
One subsection per method with the arithmetic shown.

## Triangulated forecast
Table: scenario | units or revenue | what has to be true.

## Launch curve
Table by period: low | central | high, plus a sentence on the shape.

## What to watch in the first weeks
Table: metric | low path | central path | high path | action.

## Assumptions to test
Numbered list.
</output_format>

可替换内容

  • {{product}}
  • {{horizon}}
  • {{#analogues}}
  • {{analogues}}
  • {{/analogues}}
  • {{#signals}}
  • {{signals}}
  • {{/signals}}
  • {{> output/uncertainty}}

分类与标签

商业营销、新品上市、需求预测、销售规划、市场分析

前往 AI 图片创作