早点收费、把价格定高一点、别设完就忘：Naomi Ionita 谈 SaaS 变现

Naomi Ionita is a partner at Menlo Ventures focused on seed-to-Series B SaaS. Her operating career began with engineering and consulting, then moved through a stint at the Stanford Design Institute to Evernote in 2011, where she helped start the growth product function as the product went from about 10 to 100 million users. She then built product, data, growth engineering and design teams at the bootstrapped mobile company Invoice2go.

Asked by host Lenny Rachitsky why Evernote never became Notion, Ionita points to the jump from single-player to multiplayer to team to enterprise. Evernote was "philosophically antisocial," built as a personal second brain, and she says you can't retrofit collaboration; it has to be collaboration-first from the start.

She explains why that bridge matters for every metric. Acquisition rises through referrals and shared workflows, retention rises because shared workflows are sticky, and monetization rises because revenue scales with usage, as with Figma for design or Jira for ticketing. Evernote remained a prosumer favorite but never became a larger, high-contract-value enterprise product.

Ionita names three common missteps: waiting too long to monetize, underpricing (including offering only one plan for every segment), and setting pricing once and never touching it again. Early on, she treats free beta feedback as an R&D cost, but she says too many companies take far too long to shift from building a product to building a business.

Not charging has hidden costs. People assign a low or zero dollar value to what you built, you miss the feedback loop about what customers will actually pay for, and you set yourself up for backlash when you finally introduce fees. For her, the true signal of product-market fit is people paying, which should also steer the roadmap toward what customers value.

She stresses this is not an argument against freemium. Evernote was the darling of freemium, and the real questions are where to place the paywall and how to package the paid version. Whatever gets users to the aha moment and habit formation should stay free, as should anything that drives virality, because free users who bring in others help hold down acquisition costs.

At Invoice2go, Ionita's team rebalanced pricing and packaging around a simple distinction between day-one and day-100 features. Premium features that deliver value the first time you use the product belong in the paid plan, while advanced functionality that depends on accumulated data should be an upsell. The result was twice as many upgrades from starter to pro while the pro price rose by something like 30%.

She recommends choosing a value metric that lets revenue grow naturally as customers use more, such as API calls, messages, storage or words written. Seat-based pricing was the historical SaaS norm, but usage-based models have been gaining ground with product-led growth. Matching price to value also forces you to understand who you are building for.

Evernote charged $45 a year from the beginning, and surveys revealed the problem. A popular reason for converting was feeling guilty about using it so much, which signals the free version is too good. New users saw it as a pre-installed notes app, while avid users who relied on it everywhere said it was worth hundreds, so a single premium tier shortchanged the business and led to different plans for different personas.

Ionita's first rule is "do not set it and forget it." Evernote went many years before overhauling its pricing, even though teams labor over product design. She suggests treating each meaningful launch, roughly every six to twelve months, as a chance to revisit monetization.

To set an initial price she recommends a pricing committee that spans functions: product, growth and data science at a PLG company, or sales, finance and rev ops at an enterprise firm. That group talks to customers through surveys and interviews. One method is to have users rank features as must-have, nice-to-have or unnecessary, or to spend a hundred points across features, because conversion usually hinges on just one or two of them.

For willingness to pay she points to the Van Westendorp method. Users are asked what price is so cheap they doubt the quality, what feels like a good deal, what is expensive but still acceptable, and what is prohibitively expensive. Plotting those four curves informs the price and pairs demand questions with willingness-to-pay questions.

The payoff can be large. She cites an OpenView survey in which about half of companies that changed pricing saw at least a 25% increase in ARR, and a ProfitWell study in which a 1% gain in monetization had four times the impact of the same gain in acquisition. She has personally seen lifts of up to 10 times, though they are tangled up with product launches and rebrands, and the best results come from rebalancing pricing and packaging holistically rather than just picking a new number.

Ionita tells the story of Envoy, the digital visitor check-in tool. While negotiating with a large hospitality company and sensing good vibes, its founder quoted ten times his usual price, and the executive said sure without a second of hesitation. The lesson, she says, is that he was wildly underpriced and could probably have pushed even further.

In enterprise deals she advises continually asking for more to find the upper bound, and accepting that losing 20 to 30% of deals on price is reasonable. Lenny notes that jumping straight to 10 times loses data, and Ionita agrees these are feedback loops that benefit from some incrementality, as long as you have enough data points per segment.

For PLG companies with public pricing pages, she urges experimentation even though tooling has historically been weak. Invoice2go built heavy in-house metering and experimentation systems and tested value metrics, quota limits, price points and promotions. She warns that pricing tests are hard to isolate and need a consistent experience across product, pricing page, app stores and emails.

Two practical tricks stand out. Her team ran tests in Canada or Australia before rolling out in the US, and they judged results over a long horizon, because something like a year-one discount can only be evaluated by watching year-two churn, retention and ARPU.

On the tension between revenue and growth, Ionita says that if you know you have a bridge to move up market, giving up the long tail of individual users can be well worth it. Figma is her example: it took a while to monetize, and free individual use built enormous community love among designers.

Because design is inherently collaborative, working with engineers, PMs, marketers and executives, users tripped a wire to pay once they used it in a corporate setting. She notes that more than half of Figma users in the enterprise were not designers. Those compounding growth loops produced an exponential monetization curve once team and enterprise packages arrived.

She does not advise ignoring revenue entirely. Early users can be design partners who give tight feedback, perhaps with a year-one discount, but companies often take too long, and in Evernote's case the free version was simply too good. The takeaway is to be deliberate about where the paywall sits. Lenny quips, "Just don't optimize for guilt in your paid driver."

Ionita explains the modern data stack as cloud-native tools for moving and managing data: managed ELT pipelines, a cloud warehouse such as Snowflake or Redshift, a transformation tool like dbt, and a visualization layer. Her thesis area at Menlo, the modern growth stack, is the next layer: the workflows that data enables for product growth and revenue teams working on activation, monetization and retention. It replaces the infrastructure that teams like hers built or bought.

She frames the benefits as data, workflow and impact. Reverse ETL tools such as Hightouch and Census break down data silos, letting business teams self-serve without waiting for engineers or data scientists. Purpose-built software also spares teams from spending "sacred hours" building experimentation, personalization and billing tools in-house, and the strongest products show hard ROI through both cost savings and revenue.

She then walks through layers of the stack. Product-led sales tools, including Endgame and Pocus, use product usage data to point sales teams to accounts ready to upgrade. Experimentation platforms such as Optimizely, Eppo (which ties results directly to warehouse metrics like revenue and margins) and Amplitude help teams know whether a product or pricing change worked. Lenny discloses that he is an investor in several of these companies, and that Eppo and Amplitude have sponsored his podcast.

The last layer she covers is billing and monetization platforms. With SaaS shifting from seats to usage, new entrants like Metronome and Orb handle usage-based billing, and Ionita likes how Orb connects billing with data infrastructure to inform pricing iteration and revenue forecasting. She says she wishes she could have bought tools like these many years ago.

Asked whether usage-based or seat-based pricing is best, Ionita says the answer is not either-or. Only around 5% of SaaS companies have a pure usage-based model, and the vast majority use a hybrid, with good-better-best subscription tiers that include a consumption quota. Hitting the quota triggers an upgrade, or sometimes overage charges.

She cautions that purely usage-based models are not always how CFOs want to buy, because buyers want predictability and budgets. She recalls being surprised at how much she was paying for tools like Mixpanel, Segment and Jira after fast growth. Her advice is to package a fixed component and a variable component so customers can buy more predictably.

On the future of the stack, Ionita says she would not be doing her VC job without mentioning generative AI. She expects AI to help with image, text and code generation in marketing copy and outbound sales, and to produce large cost savings in customer support. She is most interested in uses with attributable ROI, because they touch revenue directly.