In the SaaS and AI fields, pricing strategies are key to achieving commercialization and profitability for enterprises. This article delves into the pricing strategies of plugins and AI Agents through actual cases of the author’s SaaS company for your reference.
As a SaaS company established for 10+ years, we have been promoting the “strategy” of plug-in and AI applications for more than a year.
At the same time, as a commercial institution with the pursuit of profit as the core, commercialization is the “only way”, and charging for value-added products is one of its most basic ways.
Lately, we’ve been stuck on how to price it:
- Are plugins part of a standardized SaaS product? Is there a fee?
- If the plug-in is paid for by a customer, is there a fee? If so, is it based on the original price or the marginal cost price?
- If it is an AI Agent, should it be charged per Agent per module or per independent Agent?
- Wait a minute.
So, let’s talk about pricing of products today, especially SaaS products.
Three common pricing methods: on-demand, cost-based, and value-based
The first type: on-demand pricing: Refers to dynamic charging based on the customer’s actual usage (such as the number of calls, storage capacity, computing resources, personnel size, etc.) to flexibly adapt to user needs. Its advantage is that the price is transparent and the threshold for use is low, while the disadvantage is that the more it is used, the higher the cost.
For example, SaaS products have five modules: A, B, C, D, and E, and are priced according to the module + personnel size (as shown below).
The second type: pricing at cost: refers to the production cost (development, operation and maintenance, manpower, etc.) of the product or service plus fixed profit margin pricing.
How can product managers do a good job in B-end digitalization?
All walks of life have taken advantage of the ride-hail of digital transformation and achieved the rapid development of the industry. Since B-end products are products that provide services for enterprises, how should enterprises ride the digital ride?
View details >
It is expressed as follows:Unit price = (fixed cost / expected sales volume + variable cost per unit) × (1 + profit margin)。
For example, if you develop a set of SaaS products, the estimated annual R&D/sales/marketing cost is 20 million, the annual service cost is 8 million, and the cost of changing servers for each customer is 1,000 yuan, the expected profit margin is 30%, and the number of target customers that can be effectively reached is 1,000 per year, then the unit price per year = (28,000,000/1000+1,000)x (1+30%) = 37,700 yuan.
The third type: pricing by value: Refers to pricing based on the value that customers perceive for a product or service.
For example, Starbucks has created a unique sense of brand value through the concept of “third space” and “Starbucks experience”, so when pricing products, they can be priced according to the price that consumers can afford (38 yuan), rather than the cost price of coffee (8 yuan).
Of course, no matter which method is adopted, we will also refer to the price of competing products on the market and the spending power of the target group.
How are plugins priced?
What is a plugin?
A Plugin is a type ofAdd-ons that can extend or enhance existing software functionality。 It relies on host software (such as Chrome browser), and users can choose to install or uninstall (such as ad blockers or automatic translation) according to their needs, avoiding the main program being too bloated.
SaaS products often promise continuous free iterations of features to meet customer needs。
But when we consider plugins, the question arises: Do these add-ons that rely on host software and can be flexibly installed and uninstalled also fall under the category of free commitments?
Let’s understand this question with an example: a SaaS product with 10 standard features and 2 optional plugins (P1 and P2). After the customer purchases, the number of standard features is fixed at 10, while the plug-in features are based on availability.
For example, if customer A does not open any plug-ins, there are 10 available functions; If customer B has enabled plug-in P1, there are 11 available functions (10 standard functions + 1 plug-in).
From the customer’s point of view, they naturally hope that the plug-in function can also be iterated for free. After all, their subscription is meant to meet business needs, whether they are implemented through standard features or plugins, which are the services they deserve.
However, from the perspective of SaaS vendors, the situation is more complicated. The free iterations they promise are usually for standard features. The plugin ecosystem is built to introduce more diversified functions and resources, and if plugins are also included in the free iteration, it may have an impact on the market mechanism and is not conducive to the healthy development of the ecosystem.
The final solution may be:A select few official self-developed plugins are provided free of charge, and most of them are paid on demand.
Since there is a fee, how to set the price?
As plugin developers, we have recently encountered some challenges in exploring plugin pricing strategies.
Initially, we considered three common pricing models:
- On-demand pricing: Refer to the pricing logic of SaaS products, charging based on plugin type and enterprise employee size. For example, plug-in A costs 1,000 yuan for a 50-person enterprise and 2,500 yuan for a 200-person enterprise.
- Per-Cost Pricing: R&D cost pricing based on plugins. For example, the R&D of plug-in A invests 10 person-days, and the price is 20,000 yuan per person per day.
- Pricing by Value: Pricing based on the value that the plugin brings to customers. For example, Plugin A helps customers save 2 person-days per month while being compliant, priced at $15,000 at this value.
We first excluded on-demand pricing (option one) because the plugin enablement is not absolutely positively correlated with employee size and is not conducive to the external presentation of price. For example, a company with 500 people purchased a subsidy calculation plugin, but only 50 people actually used it, and it was difficult for customers to accept charging 500 people, and it was difficult for us to effectively control it by charging 50 people.
Based on inertial thinking, we initially opted for cost-based pricing (option two), but this model drew strong opposition from customers and internal partners.
For example, we have developed a plugin that can help users clearly split different overtime hours and subsidies with 0 points as the boundary when overtime encounters statutory holidays (i.e., overtime on weekdays is 1.5 times work, and holiday overtime is 3 times wages).
Due to the complexity of the system and the diversity of scenarios, the R&D cost is high, taking 30+ man-days, calculated at 3,000 yuan/person-day, the cost is 90,000 yuan, which is about 2-3 times the purchase price of a single module. Just like when buying a house, the decoration cost exceeds the house price by 2-3 times, which is unacceptable.
Customer feedback:”This offer is a bit outrageous, is your company short of money?？ This made us realize that pricing based solely on cost cannot reflect the true value of plugins, and it is difficult to be accepted by the market.
As a result, we began to rethink the feasibility of value-based pricing (option three) and actively explored how to accurately evaluate and communicate the value that the plugin brings to our customers.
At lastWe price it according to the value it provides. At the same time, a single plug-in cannot exceed 30% of the price of its module, and if the plug-in cost is high (i.e., more than 30 person-days), it can be premiumed by 5%-10%。
For example, if the price of a single module is 20,000 yuan and the plug-in invests 30+ man-days in R&D, the price should not exceed 6,000 yuan (i.e., 20,000 x 30%), and the maximum should not exceed 8,000 yuan (i.e. 20,000 x 40%).
As an insider, you may be questioning:The marginal cost of the plug-in is close to zero, so why not adopt the strategy of small profits and quick turnover？
For example, if the price is 8,000, only 2 customers will buy it, and the income will be 16,000; and if the price is 2,000, 10 customers will buy it, and the income can reach 20,000. However, under the current circumstances, it is rare for the same plug-in to sell more than 10 customers, and in order to ensure cost recovery, small profits and quick turnover are not a suitable choice.
Note: This case is based on the author’s B-end SaaS enterprise as an example, and may not be applicable to C-end SaaS products.
How is AI Agent priced?
In addition to plugins, we also face another pricing problem: AI Agent.
1. On-demand pricing: Charges are based on the number of agents, number of calls, and function modules required by the customer.
For example: 10,000 yuan for the basic version (1 agent, 50,000 calls, 1 tuning); upgraded version (3 agents, 500,000 calls, 5 tunings) 50,000 yuan; Premium Edition (10 standard agents, 1 custom agent, unlimited calls, unlimited tuning) 200,000 yuan.
2. Pricing at Cost:For example, it takes 15 man-days to develop an agent, and the price is 45,000 yuan based on 3,000 yuan per person-day.
3. Pricing by valueFor example, a data analysis agent can help customer decision-makers effectively call, analyze data and gain insight into trends, with independent value, and can be priced at 20,000 yuan.
Initially we opted for “on-demand pricing” but quickly ran into trouble:If you charge by the number of agents, how do you define what functions an agent includes?
For example, if a SaaS company has 5 modules (organization, performance, approvals, attendance, compensation), you can choose:
- Each module corresponds to an Agent (1 to 1)
- One module corresponds to multiple Agents (1 pair N)
- Multiple modules correspond to one Agent (multiple to 1)
- Multiple modules correspond to multiple agents (many-to-many)
If you charge by the number of agents, you will naturally prefer the “many-to-many” model, as this may lead to more payment opportunities. But for customer and product positioning, it may not be the best choice.
For example, we split the “Data Agent” into five (organization, performance, approval, attendance, salary), and commercially we can sell the money of five agents, but the operation will be very cumbersome for users and the experience will be very poor.
Therefore, pricing cannot simply rely on the number of agents, nor can it be used in one way, but needs to be considered comprehensively.
Our new ideas are:
First, the base tier is on-demand pricing。 It is mainly aimed at the number of user calls and storage space (such as large model token usage and file storage), which is relatively standardized.
Second, the Agent layer is priced by value。 That is, each agent is priced independently, which not only considers product positioning and commercialization, but also takes care of the user experience. Each agent is priced according to its value, and customers can choose to activate it on demand.
Differentiated pricing based on the actual value and product positioning of each agent, such as:
- HR Agent (Personnel Basic Services + Policy Inquiry): 12,000 yuan
- Data Agent (full module analysis and decision-making): 25,000 yuan
- Legal Agent (real-time regulatory case database): 8,000 yuan
- Leave Attendance Agent (full function of holiday management): 10,000 yuan
- Scheduling Agent (intelligent scheduling system): 15,000 yuan
- Wait a minute.
Written at the end
Finally, I would like to say that product pricing is both a technology and an art. There are many pricing methods, and this paper only starts from practical problems and roughly discusses the three common modes of on-demand, cost-based and value-based.
It is important to emphasize that pricing is by no means a “one-man show” for product managers. The above content is more of my personal thinking and sorting as a product manager, and it is also a trial sharing of a “novice” in the field of pricing. It would be my greatest honor to be able to bring you a little inspiration.
