B2B SaaS Growth — PLG / SLG Playbook

Podcast-derived stage evidence and claim boundaries: references/podcast-evidence.md.

🌍 Language / 语言: English | 中文

"People like it but revenue isn't growing" is rarely one problem. It's usually a mis-matched motion: a PLG product run with an SLG playbook, or vice versa. Start by diagnosing the motion, then walk the staged path.

---

English Version

Pick your motion (the flywheel)

        PLG (product-driven)
              ↑
   existing traffic + short decision
   ←─────────────────────────────────→
   needs trust + long cycle
              ↓
        SLG (sales-driven)

Signal	Motion	Reference
ACV < $1k/yr and decision < 1 week	PLG	Vercel, Supabase
ACV > $10k/yr and decision > 1 month	SLG	Snowflake, Databricks
In between	PLG-to-SLG hybrid	HeyGen, Deel

The 5-stage path

Stage	ARR	Key move
Pre-PMF	$0–$100k	10–30 user interviews; confirm the must-have before anything else
Early PMF	$100k–$1M	Invest in 1–2 channels deeply — do not spray
Growth	$1M–$5M	Add affiliate + integration partners; PLG gets sales-assist
Scale	$5M–$10M	Enterprise team, compliance (SOC 2), channel acceleration
Ecosystem	$10M+	Open platform, investor / agency distribution

Conversion-first: grind signup-to-paid before you scale

Counter-intuitive but field-proven: do not scale traffic until signup-to-paid conversion clears the bar. Below it, paid traffic is negative-ROI ("suicide marketing").

Benchmark	Threshold
ToC AI signup-to-paid	>= 1.5% before scaling (good products: 6-8%; best seen ~40%)
ToB AI signup-to-paid	> 10-15% before scaling
Teams that actually executed this	4-8%+ signup-to-paid, and smoother fundraising

The SOP (200+ teams heard it; fewer than 10 executed):

1. Freeze paid ads, influencer pushes, and launch campaigns until the benchmark is hit.
2. Find the persona with the highest willingness to pay - talk to one user 5 times if needed.
3. Resist FOMO from competitors' PR; hold the line for months.
4. Proof case: one coached ToC product stayed quiet for ~6 months, reached 6-8% signup-to-paid - ~$5,000 MRR on ~10,000 registered users - then scaled, and the growth curve (and fundraising) took off.

Unit economics, both ends:

• Interview-driven path: Gamma, HeyGen, Lovable each ran ~500-1,000 user interviews within roughly half a year before scaling (self-reported industry accounts). HeyGen's 5-minute signup-to-pay flow was distilled from ~1,800 interviews - 1 free credit, paywall exactly at the aha moment.
• Burn path (anonymized counterexample): tens of dollars per signup, hundreds of dollars per paying user.

Day-0 pricing discipline

• Charge from Day 0. A pricing-test window is fine, but collect money - even $1 plants the "this is not a free product" perception. Paying users vote with their wallet; free users just have opinions.
• Don't wait for the paid tier to exist: ship a (deliberately vague) pricing page with a contact form. Field result: dozens of price inquiries within two weeks; inquiry mix ~70% community-free / 20% personal / 10% team - and the question shifts from "is this free?" to "how much?"
• Users genuinely pay for at most ~3 points in any product. Instrument usage, find them, gate there.

PLG-to-SLG watershed and lean-team benchmarks

• Below ~$1M ARR: PLG + organic + light paid marketing is enough. Above ~$1M ARR (and most ToB): you need SLG / local key-account channels to keep scaling.
• Lean-team bar (podcast-recalled figures, treat as approximate): Notion ~4-8 people at million-user scale; ClickUp ~24 people at $1M ARR; several 8-10 person teams in the same space at $1M revenue. Operator rule: over 25 people without $1M revenue = the org is the problem, not the market.
• Inbound-heavy SLG pattern (founder's own account, self-reported): Mattermost's community edition drives search ranking, self-serve trials, then inbound deals - growing from ~$100k contracts to ~$1M contracts without a large sales team. Dify sells a standardized on-prem edition at ~$30-50k per deal with a single deploy engineer, routing custom work to partner vendors.

Pre-product B2B outreach gate

• Do not wait for prospects to prepare perfect workflow documents. Build a rough demo from public evidence, ask for whatever artifacts already exist during the call, and refine after observing the real workflow.
• Segment each request into standard product / guided setup / paid customization. If setup quality affects the product experience, publish minimum standards and an evaluation checklist instead of leaving the buyer to improvise.
• Start with reachable operators and smaller accounts; test roughly ten named enterprise accounts separately because procurement and internal politics make them a different motion.
• Before outreach, remove or replace every broken CTA with a waitlist or explicit beta path. Outreach may precede a finished product; a dead first step may not.
• Measure the funnel as contacted -> replied -> meeting -> qualified test -> activated -> paid, and reuse winning messages across owners.

Enterprise deal execution (private deployment & custom 2B)

Winning the contract is half the job — getting paid in full is the other half. Distilled from a Multi-Agent open-source commercialization project retro.

Payment structure & final-installment defense

Rule	Detail
Staged payments	Private deployment splits 3:4:3 or 4:4:2 — first features live / full deployment / final payment after 3–6 months of ops
Collect on signature	30% due the day the contract is signed
Late penalty	Written into the contract: 0.5% per day overdue
Final-payment leverage	Keep 1–2 features deployed on your own servers until the last payment clears — say it before signing, write it into the terms
Who to chase	Chase the tail payment with the highest-ranking contact; if your champion stalls, escalate to the CEO once or twice
Long-term direction	Every new deal: prepayment up, tail payment down

Scope control: SOR / SOW / CCB

• Keep SOR (customer's raw asks) separate from SOW (acceptance criteria + delivery milestones + change terms); sign the SOW together with the contract.
• All changes go through a CCB: customer initiates in writing → both PMs assess impact (cost / schedule) → both sides sign a change confirmation.
• Leave a paper trail for everything; name the acceptance contact in the contract.
• Customer gets excited on a call and promises the moon? Only paperwork counts.

Commercial push-pull

• Customer asks for 50 features and you could do 40 → "we can commit to 10, and because we want this to succeed we'll stretch to 15." Turn concessions into favors.
• Requests you don't want → price them up (pad the person-days) instead of refusing outright.
• Private deployment has only two legitimate drivers: data privacy + internal permission-system integration. Everything else belongs on SaaS.
• SaaS customization → sell a License. SSO and similar integrations → open API + standard docs, let the customer build it themselves.

Revenue & channel math

Path to $1M ARR	Breakdown
Enterprise route	9 × $100k 2B customers
Subscription route	~~5,000 × $200/yr subscribers (≈500 per $100k) per $100k (≈5,000 for $1M)

Channel tier	Terms
Reseller	10–20% referral fee
Silver (>$100k/yr sales)	15%
Gold (>$300k/yr sales)	20%

Reseller due diligence — three questions: what did you resell before? where do your customers come from? reseller or service partner?

2B content formula: 80% how you helped customers succeed + 20% about the customer. Deck narrative: pain first, then the fix.

Deep dive: references/deal-execution.md. The full operator layer (contract-clause quick cards, the 12-question pre-signing checklist, negotiation scripts) ships with the Gingiris paid field manual → https://gingiris.tools/playbook/2b-deals/

B2B content is different from B2C

• Must have a case study (ROI / time saved / headcount saved).
• Must have verifiable numbers ("customer X saved Y hours/month").
• Must have a CFO-friendly pricing page (ROI calculator, annual discount, compliance docs).
• ❌ No B2C hype words ("amazing", "magical").

Anti-patterns

• ❌ Paid acquisition before PMF — you burn cash 10× faster than you learn.
• ❌ Forcing PLG → SLG at $1M just because. Vercel held PLG to $100M.
• ❌ Watching ARR but ignoring NRR. NRR < 100% = a leaking bucket.
• ❌ One contract template for everyone — split startup / mid-market / enterprise.

What to do next

• Pre-PMF → run formal user interviews to find the must-have.
• Post-PMF, need content → move to the blog/SEO workflow (B2B voice: data > story).
• Going global → localize the listing and pricing.
• Benchmarking a rival → run a competitor teardown.

---

中文版

选你的增长引擎（飞轮）

        PLG（产品驱动）
              ↑
   已有流量 + 短决策
   ←─────────────────────────────────→
   需要信任 + 长周期
              ↓
        SLG（销售驱动）

信号	引擎	对标
ACV < $1k/年 且 决策 < 1 周	PLG	Vercel、Supabase
ACV > $10k/年 且 决策 > 1 个月	SLG	Snowflake、Databricks
中间	PLG-to-SLG 混合	HeyGen、Deel

5 阶段路径

阶段	ARR	关键动作
Pre-PMF	$0–$100k	10–30 次用户访谈，先确认 must-have
Early PMF	$100k–$1M	投 1–2 个渠道做深，不要广撒网
Growth	$1M–$5M	引入 affiliate + 集成伙伴，PLG 加 sales-assist
Scale	$5M–$10M	Enterprise 团队、合规（SOC 2）、渠道加码
Ecosystem	$10M+	开放平台、investor / agency 分销

转化优先：先磨注册→付费，再放量

反常识但实战验证：注册→付费转化率没到基准线之前，不要放量——投放 ROI 打不正（"自杀式营销"）。

Benchmark	基准线
ToC AI 注册→付费	≥1.5% 才可放量（好产品 6–8%，最好见过 ~40%）
ToB AI 注册→付费	>10–15% 才可放量
真正执行者的结果	注册→付费 4–8%+，融资也更顺

SOP（200+ 团队听过，真正执行完的不到 10 个）：

1. 达标前冻结投放、红人、launch campaign。
2. 找出当前阶段付费意愿最高的画像——必要时"和一个用户聊 5 次"。
3. 抵抗竞品 PR 带来的 FOMO，顶住几个月。
4. 实证：一个被辅导的 ToC 产品沉寂约半年，注册→付费做到 6–8%——约 1 万注册用户做到 ~$5,000 MRR——之后放量曲线起飞、融资顺利。

单位经济学两端对照：

• 古典产品主义：Gamma、HeyGen、Lovable 都在约半年内做了 ~500–1,000 场用户访谈再放量（口述行业参照）。HeyGen 的"注册到付费 5 分钟"链路提炼自 ~1,800 场访谈——只给 1 个 free credit，付费墙卡在 aha moment 最高点。
• 烧钱投放反例（匿名）：单个注册成本几十美金、单个付费用户成本几百美金。

Day-0 收费纪律

• 从 day 0 开始收钱。 允许 pricing test 调整期，但要勇敢收费——哪怕收 $1，也种下"这不是免费产品"的心智。付费用户用脚投票，免费用户只是发表意见。
• 不要等付费版做出来：先挂一个（刻意写模糊的）pricing 页 + contact 表单。实测：两周收到几十个问价；比例约 70% community free / 20% 个人 / 10% team——用户口径从"这免费吗"变成"大概多少钱"。
• 一个产品用户真正付费的点不超过 3 个。做埋点找到它们，把卡点设在那里。

PLG→SLG 分水岭与精简人效基准

• ~$1M ARR 以下：PLG + organic + 常规投放足够；~$1M ARR 以上（及多数 ToB）：必须配 SLG / 本地 KA 渠道才能持续规模化。
• 人效基准（播客口述回忆，按约数看）：Notion 约 4–8 人时已达百万用户量级；ClickUp 约 24 人做到 $1M ARR；同赛道多个 8–10 人团队做到百万美金营收。操盘铁律：超过 25 人还没到 $1M 营收 = 组织有问题，不是市场有问题。
• Inbound 型 SLG 范式（founder 面授口述）：Mattermost 社区版带搜索排名→用户自助试用→主动 inbound，从 ~$10 万一单做到 $100 万一单，无需养大型销售团队。Dify 只卖标准化私有部署版（$3–5 万美金/单，派一个工程师部署），定制需求全部转给合作 vendor。

产品未完全就绪时的 B2B 外联门槛

• 不等客户把工作流资料准备齐：先用公开信息做粗版 demo，在会上直接索取已有材料，再基于真实流程迭代。
• 把需求拆成 标准产品 / 引导式配置 / 付费定制；凡是会影响产品体验的配置，都提供最低质量标准与验收清单，不能完全交给客户自行摸索。
• 优先从更容易触达的小型客户和一线操作者开始；大型企业单独抽样约 10 家验证，避免把采购流程和内部协同难度误判成产品需求。
• 外联前逐个检查 CTA：未完成的登录或试用入口改成 waitlist / Beta 申请，不能让潜客第一步就卡死。
• 漏斗统一记录为 触达 -> 回复 -> 会议 -> 合格测试 -> 激活 -> 付费，跨负责人复用已验证的文案和异议处理。

2B 打单与交付（私有化 & 定制）

签下合同只是一半，全额收回款才是另一半。提炼自某 Multi-Agent 开源商业化项目复盘。

付款结构与尾款防御

规则	细节
分段付款	私有化部署 3:4:3 或 4:4:2 —— 第一批功能上线 / 全部部署完 / 运维 3–6 个月后付尾款
签约即收款	签合同当天收 30%
违约金	直接写进合同：拖一天多付 0.5%
尾款筹码	交付时留 1–2 个功能部署在自己服务器，尾款结清才迁 —— 签合同前说清楚、条款里写清楚
找谁要	尾款直接找对接的最高的人；对接人推诿可问 1–2 次 CEO
长期方向	每一单：预付款越来越高、尾款越来越少

需求边界管理：SOR / SOW / CCB

• SOR（客户原始输入）与 SOW（验收标准 + 交付节点 + 变更条款）分开；合同和 SOW 一起签。
• 变更一律走 CCB：客户先书面发起 → 双方项目经理评估影响（金额 / 工期）→ 双方签变更确认文件。
• 一切留痕（会议 / 对话 / 验收记录）；合同指明"项目验收对接人"。
• 客户聊嗨了拍胸脯？只有 paperwork 有效。

商务推拉

• 客户提 50 个需求、内部判断 40 能做 →"先答应 10 个，因为想帮你所以做到 15 个"——把让步做成人情。
• 不想做的需求抬价过滤（人/天多标几天），不用直接拒绝。
• 私有化需求只有两个真点：数据隐私 + 接内部权限系统，其他都该买 SaaS。
• SaaS 二开 → 直接卖 License；SSO 等集成 → 开放 API + 标准手册，让客户自己做。

营收与渠道账

$1M ARR 拆解	组成
2B 路线	9 个 $10 万企业客户
订阅路线	每 $10 万 ≈ 500 个 $200/年订阅用户（$1M ≈ 5,000）

代理分层	条件
Reseller	10–20% 推荐费
银牌（年销 >$10 万）	15%
金牌（年销 >$30 万）	20%

代理尽调三问：之前代理过什么产品？客户从哪来？Reseller 还是 Service Partner？

2B 内容公式：80% 讲自己如何帮客户成功 + 20% 讲客户；deck 叙事"先讲怎么痛，再说怎么好"。

B2B 内容特殊要求

• 必须有 case study（ROI / 省时 / 省人力）。
• 必须有可验证数据（"customer X saved Y hours/month"）。
• 必须有 CFO 友好的 pricing 页（ROI 计算器、年付折扣、合规文档）。
• ❌ 不用 ToC 话术（"amazing"、"magical"）。

反模式

PLG 信号门（2026 会议纪要更新）

• 免费 DAU 不等于 PLG 成立。先定义一个与价值直接相关的激活事件，再看 注册 → 激活 → 重复使用 → 付费/合格扩张。

• 剔除内部、测试和活动赠送用户，并按注册月份做 cohort；只有登录、没有完成核心任务的用户不算激活。

• 产品未达到稳定激活与留存前，不扩付费投放、不盲目加销售团队；回到访谈、onboarding 和定价实验。

• 当团队使用、席位增长、安全/采购需求或明确扩张意图出现时再引入 sales-assist，而不是见到免费用户就交给销售。

• 所有转化阈值注明日期、样本范围与数据来源；历史案例用于形成假设，不作为跨产品的永久标准。

• ❌ 没 PMF 就投流（烧钱比学习快 10x）。

• ❌ 到 $1M 就盲目 PLG → SLG（Vercel 坚持 PLG 到 $100M）。

• ❌ 只看 ARR 不看 NRR（NRR < 100% = 漏水桶）。

• ❌ 只给一个合同模板（分 startup / mid-market / enterprise 三版）。

---

By Iris (生姜 Iris) · ex-COO @ AFFiNE, 150+ AI startups advised · Install: npx skills add Gingiris-1031/gingiris-b2b-growth