竞品矩阵是产品里最常被滥用的产物。要么变成营销宣传（我们一列全绿对手一列全红），要么 80 行没人看。下面 15 个 Prompt 产出诚实可执行的矩阵——能让团队发现真差距、被惊到、且能直接喂下季度 roadmap。覆盖功能对位、JTBD 覆盖、定价结构，以及更稀缺的”他们做了我们没想到”轴。
适合哪些场景
梳理竞品的 PM、准备融资的创始人、写 battle card 的 GTM lead、做对比页的产品营销。
什么时候不建议这样写 Prompt
你没有亲自用过竞品至少 30 分钟时不要用——只看营销页生成的矩阵不可靠。目标是销售 battle card 时也别用——用专门格式。
Prompt 结构公式
竞品对比 Prompt 一定带这六个要素：
- 角色：让 AI 扮演谁（资深 PM / 独立创始人 / 产品设计师 / 独立开发者 / 增长负责人）。
- 上下文：阶段（想法 / MVP / 增长 / 规模化）、团队规模、流量或 ARR、平台（web / iOS / Android）、受众、限制。
- 目标：一个具体交付物——一段 PRD、一组用户故事、一个实验设计、一篇上线公告。
- 限制：时间线（本 sprint / 本季度）、要砍的范围、不能动的东西（现有流程、计费、合规）。
- 输出格式：表格、清单、可贴 ticket 的 JSON、或带标签的段落，能直接粘到 Linear / Notion / Jira。
- 示例 / 信号：1-2 份你欣赏的参考或竞品、加 1 个想避开的反例。
这套 Prompt 适合用在哪
- 季度竞品景观刷新
- Roadmap 输入——补哪个缺口
- 重新锚定定位
- 营销站对比页
- 投资人 / 董事会的竞品概览
15 个可直接复制的 Prompt 模板
1. 功能对位矩阵
默认型。比大多数团队习惯的更紧、更少。
You are a product strategist. Build a feature comparison matrix between {us} and {3-5 competitors}. Limit to 12-15 rows that actually influence purchase decisions. For each row: feature name, what it means to a buyer in 1 line, status per competitor (yes / partial / no / unknown), severity if missing. Avoid vanity features.
Context: {category, our product, competitors}可替换变量： us, competitors, category
优化建议： 矩阵像营销稿时追加：“For every row marked yes, name the specific way the feature is delivered. If you cannot, mark unknown.”
2. JTBD 覆盖矩阵
Instead of a feature matrix, build a jobs-to-be-done coverage matrix: rows are the 6-8 jobs in {category}, columns are competitors. For each cell: how well is this job covered (full / partial / unsupported), with one-line evidence. Highlight jobs where nobody scores "full".
Products: {paste}3. 定价结构对比
Compare the pricing structure of {us} vs {competitors}. Rows: pricing model, free tier, trial, billing cadence, seat-based vs usage-based, lock-in / contract, enterprise add-ons, hidden costs. Output as a table. End with one observation about where pricing diverges from norms.4. “我们没想到他们做了什么”
Forget our current roadmap. For each competitor in {list}, name 3 things they do that surprised you or that we are NOT considering. Be specific (not "AI features", but the actual flow). Mark each: worth adopting / worth watching / not relevant.
Competitors: {list}5. 诚实自评
Given this draft comparison matrix, rate our column on each row from 1-5 honestly (1 = barely exists, 5 = best in category). Then estimate competitor scores. Flag any row where our score is the same or lower than 2 competitors — those are the gaps to address.
Matrix: {paste}6. 不同客户规模的实际价格
Show the effective monthly cost of {our product} vs {competitors} at 3 customer sizes: solo user, 10-seat team, 100-seat team. Include hidden costs (overages, add-ons, support tier). Use specific numbers from public pricing pages; mark any estimate.7. 数据新鲜度审计
For each row in our existing comparison matrix, check when we last verified the competitor data. Output: row, source (page / version / date), confidence level. Flag rows where the data is more than 6 months old — those need re-checking before quoting externally.
{paste matrix + last-verified dates}8. “同名不同物”消歧
Competitors call the same feature different names, and the same name can mean different things. For each of these 10 features, list how each competitor names it and whether the implementation is genuinely equivalent or merely shares a label.
Features + competitors: {paste}9. 对比页（营销安全版）
Convert this internal matrix into a marketing-site comparison page. Rules: never claim a green check we cannot prove, never put a competitor at zero unless verified, end with one differentiator paragraph (not feature list). Output: H1 + 2-line subhead + 8-row table + closing paragraph.
Matrix: {paste}10. 差距 → roadmap
From this matrix, extract the 5 rows where we are weakest. For each: estimated effort to close the gap (S/M/L), estimated impact on win-rate, dependency, and whether it is worth closing or worth ignoring strategically. Output as a 5-row table.
Matrix: {paste}11. 叠到买家旅程
Overlay this competitor matrix onto the buyer journey ({awareness, evaluation, trial, purchase, expansion}). For each stage, list which competitor wins and on what dimension. End with the stage where our product is most differentiated and the stage where we lose most deals.
Matrix: {paste}12. 反功能清单
List 5 things our competitors do that we deliberately do NOT do (no spam emails, no aggressive upsell, no required login, no third-party tracking, etc.). Frame each as a buyer-positive trait. This becomes positioning leverage, not a gap.13. 用户证言支撑
For each row of this comparison matrix, find or write a one-line customer-quote pattern from public review sites (G2, Capterra, App Store) that supports the claim. Mark each as: real quote (paste verbatim) or "AI-paraphrased — needs verification". Do not invent quotes.
{paste matrix}14. 季度变化
Compare last quarter's competitor matrix vs this quarter's. Output: features competitors added, features they removed, pricing changes, positioning shifts. Hypothesize what each move signals about their roadmap.
Q-1: {paste}
Q0: {paste}15. Battle card 提取
From this matrix, generate a 1-page sales battle card per competitor: their strengths (acknowledged), their weaknesses (specific, defensible), our wedge against them, 3 customer-question responses ("Why not them?"). Voice: confident, never disparaging.
Matrix: {paste}容易踩的坑
- 搞 80 行矩阵没人读——限制 12-15 行可影响决策的。
- 对手一列全红、我们一列全绿——内外都失信。
- 只用竞品营销页当唯一来源——他们也在吹。
- 跳过”他们做了我们没想到”——这是 roadmap 的真输入。
- 把矩阵当一次性产物——每季度必须重核。
- 直接从内部矩阵生成对比页——必须先过营销安全轮（模板 9）。
- 只看功能不看定价——功能可以对位，定价模型决定 40% 成交。
优化技巧
- 评分任何竞品前，先自己用过 30 分钟以上。
- 每行标核实日期 + 来源——宣称会迅速过期。
- 对外用之前先跑模板 5（诚实自评）。
- 每个差距都配工作量和影响——没估工的差距是愿望。
- 一定带反功能清单（模板 12）——你不做的就是定位。
- 至少每季度刷新；快变品类每月。
- 维护两版：内部诚实版 + 对外营销安全版——是两个产物。
FAQ
- 应该带几个竞品？: 3-5。多了稀释信号，少了藏模式。按成交频次选，不是按市场热度。
- 怎么保持矩阵诚实？: 每行标日期、每个 yes 要证据、每季度跑模板 5、和销售对一遍。
- AI 能抓竞品数据吗？: AI 能提示查什么，但核实必须人来。AI 直接从公开站点生成的竞品宣称错的多。
- 对比该上营销站吗？: 只在过了模板 9 营销安全轮且每条都可证后才上。
- 和 battle card 有何区别？: 矩阵是结构化对比数据；battle card 是基于它派生的销售产物（模板 15）。