Mercury logo black

Why your AI messaging platform should not be charging per agent

profile stefan trockel

Stefan Trockel

Paying for agents who have to take over when the AI can't do the job does not make a lot of sense, right?

agent

In today’s increasingly digitized economy, customer experience must be at the heart of every business. When the full user journey, all of the transactions, and the hopefully evolving relationship with the customer is happening purely online, the quality of the experience is a critical part of the business model. This is especially true for eCommerce. 

Expectations of customers are constantly rising. Every seamless solution for a purchase, a payment, or a service request they have with other businesses will add to their understanding of what is possible. It becomes their standard and what they expect from you as well.  

Whether on your website or through channels like Facebook Messenger or WhatsApp, messaging is among the most promising additions to the customer experience playbook. While live messaging has consistently been showing excellent results in the past, it is Conversational AI in the form of chatbots that unlocked this channel for large-scale operations. 

The direct and personal messaging experience in the app that your customers use anyway creates closeness, an individual relevance and the seamless usability of not being forced to switch channels. But there’s a more profound insight than just the old mantra of “be where your customers are.” 

To reap the benefits of messaging along the customer journey, most certainly a mix of human live chat and AI automation will yield the best results. How human support agents and AI chatbots work hand in hand and combine their strengths is where most implementations take a terrible turn.   

While a chatbot's dialogue capabilities are a topic for another post, we want to highlight one fundamental problem with most messaging and chatbot software vendors that is often overlooked. It heavily impacts not only the quality of conversational automation but your ROI for messaging at large.

In order to understand the problem, we need to look at business models of SaaS platforms that offer messaging and chatbot service - and at their value proposition. As is common in subscription-based software, there typically is a “consumption” metric used for pricing. The more of that variable unit you consume, the higher your monthly price. We can broadly classify platforms into two groups. 

Charging by the number of contacts: This is very common with platforms that originate from a marketing automation background or sell mainly to marketers. The thinking here is that user data is the crucial output of the tool, and that the main value lies in generated leads. The more contacts, that is the more value, the higher the charge. This approach makes sense for lead-generation chatbots, and that is where these platforms are most commonly used. But these tools offer only limited features to address needs further along the customer journey. At this point the value does not come from adding more users, but from nurturing the relationship with existing customers through delightful customer service. 

Charging by the number of support agents: This is a typical pricing approach in CRM, helpdesk, and customer service tools. The rationale is the same as with other instruments used by professionals to get a particular job done. It focuses on the number of people performing the job and then selling “per seat.” In customer service, this means that given your regular volume of service requests, you need a specific amount of live-chat agents to handle the work. Paying per agent puts you in the position to give all these agents the tool they need to address the volume. 

As this pricing model builds on the number of agents that manually handle the support volume, it is interesting to highlight what happens when chatbots become part of the equation. Let’s look at what chatbots typically do in pay-per-agent messaging platforms. They either perform low-level FAQ automation that curbs the request volume just enough to keep your agent team growing constantly, or they do pre-qualification and agent routing to better utilize your available agent resources.

They typically do not fully automate service requests so that the volume that hits the agents is significantly reduced. And why would they? This would cut into their revenue, as it reduces your “consumption.” When we now think back to the importance of combining human live-chat agents and AI chatbots to provide an intimate and personalized customer experience along the whole customer journey, we see a fundamental flaw in this approach. 

So how can we price a messaging platform that does not disincentivize effective automation?

At Mercury.ai, we adopted an approach that incentivizes an ideal collaboration of live agents and AI automation and maximizes chatbot ROI. This approach follows two fundamental principles.

First: Unlimited agents. We do not charge for live agents.  

And we mean it: There is no charge per number of seats or active users.  

We believe that live chat agents “on duty” are enablers for quality, not the workforce to manage quantity. So they should not be a cost factor. 

Moreover, the adoption of messaging as a customer experience channel should not be hindered by considerations about how many of your agents “really need access.” Instead, the conversation should be about discovering those requests that can be automated. Then quickly and seamlessly plug in AI components to take over. 

Second: Pay for functional scope. The price increases with the value that is delivered. 

We believe that the value of AI assistants lies in the tasks they can perform. An AI assistant that can answer questions, create a ticket, and schedule a meeting is more productive than one that can only answer questions. Accordingly, functionality is our “consumption” metric. 

The basis of this approach to pricing lies in the modular nature of our AI chatbots. It allows for simple upgrades of bots, extending their functional scope. Many task modules are already available in our library, to be added to a bot with no training required. 

As a customer, this means you’re only paying per module. So when analytics shows that your live-chat agents spend time on a set of recurring requests regarding return shipments, you switch on the “returns” dialog module. Instead of investing hours of your agents, you pay a flat fee for the module. 

The more capabilities you add to a bot, the more it will be able to automate - and the more you will reduce manual service volumes as a result. The difference is, you won’t be paying more for getting the desired outcomes of your AI chatbot! 

Don’t pay for live agent seats – instead, pay for automation value.

Sounds too good to be true? Talk to us, and we’ll run the math for your use case.