引导才是增长最大的杠杆：Airtable 前增长负责人怎么选激活指标

Isford, who worked at Facebook before Airtable, says growth teams can become dependent on experimentation, especially in consumer growth at scale where it feels like second nature. She sees two real reasons to experiment: to understand the precise metric impact of what you ship, and to mitigate the risk of big, dramatic changes. When activation moves from 6% to 7%, she argues that the precision does not help much beyond a line in a performance review.

Experiments are also expensive, since engineers, analysts and PMs spend time interpreting results instead of roadmapping, foundational analysis or shipping. Her advice is to experiment when you need to, treat it mainly as risk mitigation, and let the product development process do more work through customer conversations, mocks and clear problem definition. The aim is to ship with enough conviction that it is fine if every customer sees it tomorrow.

Lenny notes that many growth cultures reward people for metric lifts from experiments. Isford agrees that this is the default as growth teams become organizations, and says the way out is a deliberate culture of doing right by customers and the business. That requires other ways to motivate and reward people, such as qualitative feedback or deals closed, so engineers do not feel they must point to numbers.

Her example is Airtable Forms, where submitters could not get a copy of their submission without creating an account. The team knew from customer analysis that people wanted the feature, so they rolled it out without an A/B test. It moved top-line signups enough to see clearly, and attribution analysis afterward let them learn from what happened. She stresses this cannot be an excuse for rigor, and that the approach works best when leadership sets the tone.

Lenny says his research into startup growth stories shows retention must be cracked and that onboarding is one of the biggest levers on retention. Isford agrees, with the caveat that it depends on the business. If your product has any self-serve element, a sandbox, a premium tier or a free offering, onboarding is the first choke point from which paid conversion, deals closed and team adoption all flow.

Over 12 to 18 months Airtable's activation team rebuilt onboarding in three pieces. The first was an immersive wizard called Guided Onboarding that walks you through setting up a first workflow and reduces the cognitive load of a complicated product. Second came personalization to use case, and third, ongoing education via a bottom-of-screen tip pattern the team nicknamed The Mole. They also retired many tooltips and made incremental optimizations.

Isford says the combined work over about six to eight months drove a 20% lift in activation rate, and she urges celebrating any clear, statistically significant improvement. Credit goes to the team's many hours with customers, and she says understanding what users need to be onboarded well is different from what the business wants them onboarded to. The wizard was the most impactful part.

The wizard asks simple questions, such as what kind of project you are working on and what you want to track, and shows your workflow taking shape on the right half of the screen. Isford is not ready to declare the tooltip dead, since this pattern fits a complicated product with high cognitive load. A second lesson was to meet users where they are: skip advanced features like automations on day one and separate necessary education from ongoing education.

Lenny asks how Airtable actually understood user types. Isford says the team spoke to customers, watched them get started, and looked for behavior patterns and clusters. A technical person comfortable with databases needs something different from someone trying a tool recommended for a one-off project. The first wizard was still one generic flow, which she says solved more than 90% of users' needs, and personalization came afterward.

When they did personalize, grouping by learning style and building style beat classic segments such as marketing, product management or operations. Many kinds of workflows are possible in Airtable, so knowing how someone would go about building was more useful than what they wanted to see at the end. A template might inspire a project-management use case but not teach someone who needs more database familiarity or a more visual experience.

The team combined research from surveys and conversations done outside the product with questions at signup that vary depending on who you appear to be. Isford's example of the next step is a teammate invited to update a sprint, who needs a completely different onboarding from a builder. She says there is a surprising amount of low-hanging fruit in personalizing onboarding in many products if you are precise about who you work with.

Every growth team needs a north star, Isford says, and for Airtable's activation team it was a team-level metric: week four, multi-user, meaning more than one person on a team is active and contributing in the fourth week. Analytics determined it was highly correlated with long-term retention. It sets a high bar, requiring that something substantial was built, people collaborate weekly and contribute meaningfully, which makes the team's job hard.

Her heuristic is that an activation rate in a lower range, for most companies 5 to 15%, beats a higher one. If 40% of customers are active at week four, only a fraction will remain at month 12 or 24. She would rather choose a precise metric that only 5% reach but who stay for the long haul, since moving it from 5% to 7% would have huge downstream effects.

To put the metric to work, the team broke it into components, such as how many teams sign up as a team versus individually, how many reach week four, how many invite or activate more than two people, and what counts as active. They also added an individual retention metric and a roughly defined sophistication score called Build with a capital B. These guarded against treatments that made users more sophisticated without inviting teammates, or that kept users longer without better work.

Isford says this group of metrics set the team free, because activation is more complicated than a single number and they could celebrate improvements on several dimensions. Lenny sums it up as treating the headline metric as an output and the others as things you can move directly. She agrees, saying it also helped the team grapple with tensions about what onboarding is supposed to achieve.

On the debate between a workspace-level activation metric and a user-level one, Isford says it depends on the product. For Airtable, with seat-based pricing and horizontal adoption, a workspace measure makes sense, since you want to know how many people in an organization find value. Tools with different pricing or growth mechanisms can differ.

Her spicy take is that growth teams must be agile, so north stars will change, and working on the same metric forever suggests inefficiency. Stability helps with momentum and expertise, but she says teams overfocus on picking the perfect metric and by the time they find it, it may be time to move on. If team activation looks healthy, perhaps user retention or conversion should be next.

At the organizational level, Airtable's growth org changed its north star a couple of times, notably from revenue to user and customer growth. The team took a decades-long view, felt a revenue focus led to short-term decisions, and noted that in product-led sales it can be right to delay revenue so customers hand off to sales later. Isford says they gained strategic clarity and moved faster.

She considers keeping a north star stable for at least six months table stakes. When changing, start from the mechanics of the business, such as drawing the funnel on a whiteboard and judging where the biggest opportunity lies, and then build a metric that reflects the output of that strategy. The metric should measure whether your strategy delivers results, not drive all decisions from the start.

Isford defines a free trial as limited-time use followed by paying, and freemium as an option to use a version for free indefinitely. She prefers what she calls a reverse trial, which offers both. A trial gives you a window of, say, seven, 14 or 30 days to show everything your product can do, including premium features, and onboarding helps users reach aha moments.

When deciding, she suggests stepping back from competitors' pricing pages and asking about your philosophy on monetization. Offering a free plan signals that you value letting millions of people try the product even if they never pay, prioritizing user growth, brand awareness and exposure. She says you only get a few moments to ask for a credit card or an enterprise conversation before losing trust, so helping people discover value first matters more.

For products that need sales and handholding, she shares Elena Verna's hope that self-serve will eventually be possible for everyone, but notes it is expensive to build. The key is something users can do alone, such as a sandbox or demo, which need not be the full product. Alternatives include custom landing pages that showcase value, signup with learning resources, or using top-of-funnel signups to expand inside existing enterprise customers from the same email domain.

Isford's main caution is that employees build onboarding around what they think customers want. One pattern is naming features: a tooltip saying "This is automations" is an announcement, not help, because users do not yet know if it is relevant. Better is to explain relevance, or enable a one-touch setup that does part of the work and shows the value. Lenny adds smart defaults, and she adds that segmentation should decide who needs to learn it at all.

The second trap is mapping onboarding to pricing and packaging. If a premium feature like automations correlates with paying customers, it is tempting to push it to new users, but that is not in the user's best interest. She says onboarding should be rooted in the customer's need and in helping them reach maximum value.

To operationalize this, she pairs north star metrics with guardrail metrics. An onboarding team that builds a great flow but causes a 10% revenue drop needs to know, and should have the rigor to find out why. She values strong analytics partners and says everyone on the team should think about guardrails, and describe their work in terms of what is best for the business and customer, not their narrow swim lane.

Isford describes a framework she uses for any business with product-led growth: join, evaluate, upgrade and expand. People join by finding the product or being invited. Evaluate is chosen deliberately instead of onboard or activate, because it is about what users must see to believe they will get value and build a habit. Upgrade follows once they see premium value, and expand happens when more people in an organization adopt, driving net dollar retention, renewals and referrals back to join.

She says the framework is too abstract to dictate team structure or four KPIs, but it grounds the organization in how the business machine works and makes it easier to say where an investment fits, such as landing pages for join or onboarding for evaluate. At Airtable the teams were acquisition, activation and monetization, which mapped to join, evaluate and upgrade. A dedicated expansion team emerged later as larger companies adopted the product, which she cites as an example of staying agile about org structure.

Isford sees B2B growth as an emerging space where the playbook has not been written. In consumer products you can experiment at high volumes, but in B2B a small number of customers may represent a large share of revenue and must be treated with care. That makes customer conversations, beta testing, live prototypes and demos more important than in B2C growth. She would like to see more people from consumer growth try applying it to B2B.