Unique Research · Insight · April 18, 2024 · Adapted from a16z (Sarah Wang, Tugce Erten, Alex Immerman); edited by Qiuping
When commercializing a new technology fast, companies must balance quick revenue against durable growth. For generative AI (GenAI) the stakes are higher: the technology can create unprecedented value for customers, but the cost to serve each incremental customer can also be very high.
Especially in the growth stage, founders need to watch unit economics and margins closely. We often hear the same questions: How do I quantify the value GenAI creates? Should I absorb GenAI feature cost or pass it through? Will customers actually pay for GenAI — and how much?
We are still early. Until adoption curves and costs stabilize, there will be no fully proven pricing or packaging frameworks. Even so, we outline how we think about pricing and packaging for B2B SaaS and prosumer companies commercializing new GenAI capabilities — and how others are handling the same problem — so AI founders can place their current strategy more clearly. (Prosumer: e.g. in energy, from centralized fossil generation to distributed renewables on buildings — yesterday’s consumer becomes tomorrow’s producer.)
Triangulate early usage, customer profile, and product mission
When setting pricing and packaging, founders should first clarify:
- How much value do the features create, and for which users?
- What does it cost to deliver them?
Then weigh those findings against where you expect GenAI to sit in the overall product portfolio. Because founders are still discovering value and cost, this is often more art than science — but early usage and customer profiles help on both fronts.
1. Beta and early usage
In beta and early use, learn who uses the product, how often, what it costs to serve them, and how much they will pay for GenAI features. Dig into:
- Does it expand total addressable market (TAM)? (e.g. from serving 10 customers to 100?)
- Does GenAI lift free→paid or paid→higher-tier conversion?
- Is there a core cohort that drives most usage — and how does that hit COGS?
2. Customer profile
Figure out who will pay and who will not. Do all customers benefit from GenAI, or only a segment?
Good ways to learn profiles and segments: interviews, surveys, and sales-team data.
Interviews: with a small base, talk to customers about who is interested in buying and what they may want next.
Surveys: with a larger base, ask which potential features matter most, then link answers to industry or function.
Sales data: sellers talk to customers daily and usually know which features different customer types need to succeed.
One caution: watch for “AI tourists” — people who sign up because their company mandated GenAI experimentation (B2B) or because they personally want to try new GenAI features (prosumer). Even when they pay to try, they can be hard to retain. (That said, more enterprises are moving GenAI spend from innovation budgets into standard software budgets — a sign GenAI is becoming table stakes.)
3. Product mission
This is where vision and mission matter. Even if only a few customers are GenAI-enthusiastic today, you may believe GenAI will reshape the product experience and deepen the value proposition.
You may still be exploring how GenAI helps customers — today it may be an attractive add-on for some segments. That judgment is qualitative and vision-led: you decide how central GenAI is to the roadmap and value prop.
Once you understand value and cost, and have a working view of how GenAI fits current offers and future plans, you can set concrete packaging and pricing.
Packaging: core, upgrade, or add-on?
We typically see B2B and prosumer GenAI features land in one of three places: in the core product, as part of an upgrade tier, or as an add-on. Many non-GenAI prosumer and B2B SaaS best practices still apply — past packaging playbooks still guide GenAI go-to-market.
1. As core product
If nearly all customers are enthusiastic and willing to pay, early usage shows GenAI lifts adoption and conversion, and GenAI is central to the value proposition, putting GenAI in the core product is wise.
You may not need to monetize the GenAI feature directly: expanding TAM and conversion can create strong second-order gains. Example: a product that once served 10 customers may now serve 100 with GenAI.
In today’s land-grab phase of GenAI adoption, core inclusion also differentiates against incumbents and other startups. Still, because every segment may need the capability, some companies will raise core package prices to reflect GenAI delivery cost.
2. As upgrade
Optional upgrade works well when GenAI is an attractive value-add that doesn’t rewrite how customers use the product. Some GenAI companies, for example, put broader datasets behind an upgrade.
Mailchimp is a useful analogy: most users may not need GenAI in the core — their main job is building and maintaining an email list — but GenAI content, segmentation, and analytics can improve the experience. Others use GenAI as an upsell lever: higher likelihood of moving to a higher tier, while helping amortize GenAI delivery cost.
3. As add-on
Optional add-on is smart when a high-willingness-to-pay cohort gets outsized value and you want direct margin control. As an add-on it can:
- Monetize innovation directly for more sustainable near-term margins — though if AI will be the core differentiator you may later migrate packaging models.
- Expand TAM: charge more for those who will pay more while keeping base prices for a broader base.
- Give excited, feedback-rich users a way to test the feature.
As a “power-user package” add-on, a small head cohort can support higher prices. In enterprise deals, some companies also sell GenAI add-ons as part of a whole-solution bundle — partly to discourage login sharing even when only a few users need the extras.
That can cover cost short term, but not every buyer will accept forced purchase of features they don’t need; satisfaction and market acceptance can suffer. Price and sales design must weigh near-term revenue against long-term relationships.
Some companies also tier GenAI by service level: basic AI in core or lower tiers for most users; higher tiers with stronger GenAI (advanced algorithms, faster processing, more customization) at higher prices — segmenting by usage intensity and willingness to pay while leaving a clear upgrade path.
Pricing: subscription or hybrid?
Traditional B2B SaaS and prosumer products are human-facing software, so subscription (not pure usage) has been natural — customers don’t want to estimate how much GenAI they will use. But in GenAI, pure subscription can create a large gap between actual usage and revenue.
If GenAI is sold per seat, heavy and light users pay the same. That creates a perverse incentive: the company may wish power users would use the product less, because the most important customers erode margin while paying the same as near-zero users.
In the GenAI era, companies need pricing so revenue tracks usage while protecting margins. How best to align vendor and customer interests is still open this early in adoption.
In practice, core and upgrade strategies usually default to seat-based pricing: GenAI is either in the base or an option on an existing subscription.
Others are testing hybrid subscription–usage models for add-ons to balance cost and monetize head users — e.g. Box (credits that burn down with use) or Adobe Creative Cloud (base plan plus overage for usage beyond the included amount). Hybrids try to keep subscription predictability with usage flexibility. When adding usage elements to subscriptions, plan for price predictability and overage handling.
How 31 GenAI companies price and package
We surveyed 31 companies shipping new GenAI features on how they price and package them. The charts below summarize what we observed. (4 CN images — packaging / survey visuals; attach on ingest.)
How GenAI pricing and packaging will evolve
1. Outcome-based and other pricing metrics gain traction
As GenAI delivers clear productivity lifts, more companies look at outcome-based pricing — charging for results the software produces, not only for the software itself.
Implementing outcome pricing is still hard: founders are still quantifying GenAI’s concrete value. But if GenAI sharply raises enterprise productivity, seat-only pricing may stop making economic sense. We expect outcome pricing to hit human-facing software hardest — especially workflow and HR tools.
The upside is better vendor–customer alignment. Agreeing on what “outcome” or “solution” means is hard and requires confidence in GenAI’s solving power. Even so, products like Intercom’s Fin Chat are already experimenting — and we are optimistic about the model’s future.
2. Be ready to change pricing
As inference costs stabilize, open source rises, and model providers cut prices to win users, AI companies should be ready to retune pricing as API costs fall.
Near term, price in a way that roughly balances with business returns, while expecting delivery costs to keep falling (they already are) and expanding future margin room. If you bet GenAI is core but results stay thin for a long stretch, revisit pricing and packaging without hesitation. Stay flexible to market and product reality.
GenAI can deliver unprecedented value to end users of software. Growth-stage founders’ job is to maximize that value while keeping unit economics and margins healthy. There is no universal playbook — successful founders combine past experience with live learning into a clear, flexible pricing and packaging strategy that communicates roadmap value, then keep iterating so GenAI’s potential meets market demand with durable returns.
Original authors: Sarah Wang, Tugce Erten, and Alex Immerman · Editor: Qiuping
— END —