Compiler: Shidao AI Group, Editor: Lion Knife, original title: "Top VC "Nanny Level" Tutorial: Pricing Strategy of AI ToB Service", header image from: Visual China
Last week, Seekway shared the 16 major trends of AI To B, a report by venture capital firm a16z, and concluded that the B-side is both a "way to live" and a "way to win" for start-ups. Large corporations are less likely to share private data with giants, preferring instead to find "middlemen". As a result, the "data flywheel" of start-ups has the opportunity to spin – accumulating high-quality data in vertical fields and gradually building technical barriers.
Although the general direction of To B has been clarified, for small start-ups that are "short of money and computing power", the first step of the long march has just begun.
Troubled founders include, but are not limited to, "How to make the AI solution meet the specific scenario problems?", "If so, will I bear the high cost or will I pass it on to the customer?", "If it is passed on to the customer, how much are they willing to spend?", "How good can my AI function be, and can it be sold at a high price"........ This creates a closed loop of problems.
At the heart of the above issue is cost. After all, there is no considerable economy of scale, and the profit needs to be deducted from the computing power, token call and other expenses. Not every AI service can be "hit the nail on the head" and allow founders to see the real money right away.
Therefore, it is not only an art but also a science to adopt an appropriate pricing plan based on its own financing situation and product functions, so as not to be dying due to stagnation, nor to collapse because of burning money too quickly.
Recently, a16z released "Pricing and Packaging Your B2B or Prosumer Generative AI Feature", which gives founders a very detailed overview of the pricing strategy for current GenAI products. The article points out that we are in the early stages of GenAI, and there will be no "set and forget" pricing schemes until the adoption curve and cost are stable. Hope it helps.
1. Thinking: early adoption, customer roles, product vision
Before getting started, founders need to ask themselves two questions:
- How much value does the product's GenAI capabilities provide, and to whom?
- How much does it cost to provide the feature?
Here are three things you can do to clear out your cluttered mind.
1. Beta and early use
Which customers are using your product and how often, how much it costs to serve them, and how much they're willing to spend on GenAI features.
Taking it a step further, the problem can be refined as:
- Will the GenAI feature increase the TAM (addressable market size) of the product? (Can it be scaled to 100 customers from 10 customers?)
- Will the GenAI feature increase the conversion rate of the "Free-Paid-Paid Pro Version" of the product?
Will GenAI's capabilities capture some of the "heavy users" and if yes, what is the cost impact?
2. Customer Personas
Figure out who is willing to pay, and who doesn't, whether all customers benefit from GenAI, or only a subset of customers?
We can find answers through interviews, surveys, and sales team data.
- Interviews: If you have a small number of customers, interviews can give you an idea of who is interested in buying products and which products they might be interested in in the future.
- Surveys: If you have a large number of prospects, surveys can give you an idea of which potential new features are most important to them and tie that information to the customer's company industry and functional positioning.
- Sales team data: Your sales team talks to customers day in and day out, and they are often better able to capture what features different customers need.
You'll also need to distinguish between real customers and "AI visitors" — people who sign up for a product, pay for a "try," but struggle to retain conversions.
3. Product vision
As a founder, you need to think about where GenAI capabilities will be in your product roadmap.
Scenario 1: While a small group of customers are "on the side" of GenAI, you believe that GenAI will eventually reshape the customer experience and provide multiple values.
Scenario 2: GenAI can make some customers "icing on the cake", but you are still struggling with how GenAI can benefit customers.
In summary, if you understand the above questions and come up with feasible hypotheses, you can concretely consider how to package and price your product/service.
2. Positioning: core functions, upgrade options, add-ons
We divide B2B GenAI features into 3 categories: core features, upgrade options, add-ons.
1. Core Features
If all of your customers are "on the side" of GenAI and are willing to pay for it, and early usage data shows that GenAI has significantly increased product adoption and conversion rates, and GenAI is critical to your value proposition.
Then put GenAI at the core!
In this case, you may not be profiting directly from the GenAI feature, but it does have a significant downstream effect on TAM and conversion rates.
Especially when we are in the stage of "grabbing territory" with GenAI, treating GenAI as a core function can make your product "unique". With GenAI capabilities in demand across all market segments, a16z believes that some companies will eventually increase the total price of their core products to better cover the additional costs they incur.
Enterprises with GenAI as a core feature:
2. Upgrade Options
If your GenAI features aren't bad, but it's a bit "chicken-gagged", you might as well package it as an "upgrade option" as a sales lever to increase the conversion rate of the "Pro version", or cover part of the cost of GenAI.
For example, some companies offer more datasets in their upgrade options, and Mailchimp, where most of its users don't need to add GenAI functionality to their core products, AI-generated email copywriting, segmentation, and analytics are really "delicious" and optimize the user experience.
Businesses that are using GenAI as an upgrade option:
3. Add-ons
If your GenAI feature is only a small group of customers who are "willing to spend a fortune" and you want to see profit in them. Well, package the GenAI features as add-ons.
In this case, GenAI can monetize innovation directly and achieve more sustainable profit margins in the short term (if you believe that GenAI will be a core differentiator of your product, you will need to move to a different package), GenAI can expand TAM by charging more for some customers while retaining customers at existing price points, and GenAI can also provide an opportunity to beta test your ideal user base.
Businesses that offer GenAI as an add-on:
Currently, a16z sees some companies include GenAI features of base performance in their core products or base versions, and introduce more powerful GenAI features in later versions of their products, or offer more GenAI features.
In the above case, the logic of value segmentation remains the same – if GenAI is able to expand the TAM, it can be used as a core feature, and if the stronger GenAI is only available for Pro users, it can be used as an add-on.
3. Pricing: Subscription or Hybrid?
The reason why most B2B GenAI companies are subscription-based, rather than "pay-per-use," is because customers don't want to estimate "how much" of features they can use.
However, the subscription system can make AI companies "lose money" all the way, especially "per head" billing. For example, "heavy users" and "light users" pay the same amount of money, but the former uses 100 times and the latter only uses 1 time, and the corresponding costs are of course very different, which means that your most important customers will eat into the profits of the business.
As a result, some companies try to mix subscriptions, i.e., tiered pricing based on usage quotas, with additional payments for the excess amount, so as not to be dragged down by "heavy users".
Currently, there are two trends in pricing strategies:
1. Outcome-based pricing
Some B2B GenAI companies are starting to think about "extracting" fees based on results, rather than charging customers based on the software itself. However, outcome-based pricing is more difficult to achieve, as founders are still figuring out how to quantify the value that GenAI provides to customers.
Seeway summarizes the pricing models of some companies. For example, Cresta, a startup that started on a subscription basis, has changed to a "pay-per-view" — based on the number of conversations that help contact center employees, customer service company Intercom has released a chatbot Fin that costs 99 cents per customer request made, and startup Hume AI has started charging per minute, per note and per word.
2. Be prepared to be flexible with pricing at any time
As the cost of GenAI inference stabilizes and open-source models flourish, model vendors continue to drive down prices. As a result, companies are adjusting model pricing as API costs decrease.
In this regard, the founders should at least set a price that can guarantee profits in the short term. And with this pricing, margins will increase in the future as long-term service costs fall.
But at the end of the day, there is no one-size-fits-all pricing scheme, and successful founders need to build a clear, flexible pricing framework that connects with the past and looks to the future to communicate the value of their products.
epilogue
If GenAI is compared to a cake, the bottom layer of the cake is the basic model, the middle is the developer tools and infra, and the top layer is the application. A year ago, the general prediction was that a large number of innovative companies would emerge at the application layer as large models continued to advance. But the reality is the opposite, with more model vendors emerging and raising a lot of money, and the application layer seems to be just getting started.
Recently, OpenAI's chief operating officer, Brad Lightcap, predicted that 2024 will be the "year of adoption" of artificial intelligence. It is reported that the demand for ChatGPT Enterprise is growing dramatically. More than 600,000 people have signed up to use ChatGPT for Business, compared to about 150,000 in January.
Startups are good at identifying opportunities and acquiring customers, but operational and competitive differentiation is a challenge, and incumbents are good at integrating functions but difficult to capture new explosive demand. How to find another way in the "pinch" attack of new giants such as OpenAI and Anthropic? Perhaps, we have to go back to the source: enter a very clear scenario as soon as possible, solve specific problems, and continue to build high barriers in the first-mover advantage.
This content is the author's independent view and does not represent the position of Tiger Sniff. May not be reproduced without permission, please contact [email protected] for authorization
People who are changing and want to change the world are all in the Tiger Sniff APP