How to capture the revenue sitting in your SMB book with AI voice agents

Chris Winkler
Head of Marketing
•
November 17, 2026
•
Updated 
AI Summary

It’s a common problem for large companies.

They often have thousands of small customers and accounts, either current or past, that could be expanded or reengaged. While each conversion might not add a lot to the bottom line, if enough of them are converted they can deliver real revenue.

Yet this list rarely gets worked in any real way. Sales reps (rightly) focus their limited time and attention on new prospects and their biggest existing accounts, and it would cost too much to staff a dedicated team to call the entire book one at a time.

Some companies try to solve this problem with dashboards, CS tools, and reports that flag which of those small customers are most likely to convert, and when. But even this approach depends on a lot of manual effort to call those customers, and it is not viable for the many companies with products that don’t collect user data.

Thankfully, this is now a solvable problem.

AI voice agents can now make sophisticated, consented calls to a company’s entire book of small customers, and offer them expansions and reengagements.

At 11x we’ve deployed voice agents for these types of campaigns at some of the largest companies in the world, and learned what works and what doesn’t.

In this article, we’ll walk you through:

  • How to know if these agent-led campaigns are right for you
  • Whether you should run a signals-based or “brute force” campaign
  • The 3 unique needs of these campaigns vs general voice agent campaigns
  • How 11x’s voice agents can deliver these campaigns for you

How to know if these agent-led campaigns are right for you

Not every company will benefit from running an AI voice agent against their long-tail of small customers, and it is quick and easy to determine if one of these campaigns makes sense for you. Just ask yourself three questions.

“Do we have a big enough list of small customers to call?”

There is no one-size-fits-all answer to this question, as it depends on what you are selling,how long you have been selling it (and accumulating a book of current or past customers), and how quickly you are acquiring new customers. 

However, as a general rule the larger your list of past or current customers, the better. These campaigns have an overall lower conversion rate than other similar voice agent campaigns (for reasons we will explore below) so volume really does matter. 

While it is possible to run one of these campaigns against a smaller list if their average opportunity is large, overall these campaigns work best for large, established enterprises that have been building their book for years or even decades.

“Is there enough ROI from targeting and converting them?”

This can vary a lot depending on the contract size of the small customers, and the cost of building or buying an AI voice agent to work that list. Here are a few quick calculations to give you a ballpark.

  • Pull the total ARR and number of accounts in your SMB tier of current customers. Decide how much you think you could upsell one of them on average, assume you could convert 4% with a voice agent, and take the total.
  • Pull the average ACV and number of accounts in your SMB tier of your past customers. Decide how much you could sell one of them if they bought again, assume you could get 4% with a voice agent, and take the total.
  • Add the two together and get a rough idea of the opportunity you’re sitting on.
“Are we comfortable proactively calling these customers?”

Some companies worry about reaching out to their existing or past customers when most of them are not used to hearing from you, and some of them may have stopped buying from you years ago.

The best way to overcome this blocker is to build your agents to reach out in a responsible manner that customers appreciate (covered below), and to recognize that your reps are already doing this for a small number of accounts and you are simply using AI voice agents to scale those existing activities.

If you worked through these questions and feel a reengagement or expansion campaign with AI voice agents might be right for you, then the rest of this article covers how to run these campaigns effectively.

Your first campaign decision: Signals vs “brute force”

When setting up your campaign, you first need to determine if you are able to identify past customers to call in a targeted manner, or if you just need to call everyone and figure out who is ready to buy on the phone. Here’s how to know which to pick.

If you have a software product that collects a lot of user data, then you likely can create a signals-based campaign, and call them in a targeted manner. For example, you might be able to see which customers are using most of their seats or credits, or have begun to deploy extra features, or have spiked their usage recently. All of these are good candidates to have a voice agent call and offer them an expanded license.

Even if you don’t have this granular data, you likely know when the customer bought the software, or downloaded a free trial, and when their license or trial is running out. In these cases, you can have your voice agent call them before their license is ending and lock-in the renewal, or call them a few days before their trial is ending (or a few days after they signed up and you are fresh in their mind) to capture the paid license.

Some companies will already have this type of propensity model built, giving them a sophisticated way to know who is ready for a call and who isn’t, and all they will need to do is build their voice agent campaign around it. But even if you don’t have a complex model built, a few small, clear signals are enough to start with.

But not every product collects enough user data to build a signals-based campaign around. Some software products collect little-to-no user data, and most physical products collect no data at all. At best, you might know what they bought and what other related products they might benefit from, or when they bought and when they might need to buy again based on remaining useful life.

With products like these, you will need to build a “brute force” campaign and have your voice agent just call everyone cold and see who is open to buying more, or again, on the phone. At the same time, they can ask the user questions about their usage, collect data, and log this into the CRM for future sophisticated campaigns.

Overall, if you have signals, then you should use them, but both campaigns can work, and both require the same three steps to refine them over time.

3 unique needs for reengagement and expansion campaigns

Building a successful voice agent campaign is a big topic, and one we cover in real depth in our recent guide. You can access that for free here.

But there are a few subtleties and nuances that apply to these types of reengagement and expansion campaigns worth digging into here.

First, your initial goal is to just get customers to stay on the line with AI.

AI voice agents are usually deployed to answer customer support calls, or to instantly call prospects who just filled out a form. In those use cases, the customer or prospect either made the call themselves, or is expecting to hear from you, and so they tend to stay on the line even if speaking with AI wouldn’t have been their first choice.

This is not the case with reengagement or expansion campaigns, where an AI voice agent is calling them cold. They are not expecting to receive any call from you, let alone a call from an AI agent, and this makes them more likely to hang up right away if the call is not structured and handled properly by the agent.

Because of this, you need to set the right expectations. You will have a lower success rate from this type of campaign than you would with other AI voice agents. The value will need to come out of volume, and constant testing to get your rates up.

The first thing to test is your agent’s opener, and how they handle the first 10 seconds, to find out what will make your customers simply willing to talk.

Second, don’t jump into the pitch, start as a support call.

Almost no one will stay on the line if they receive an out-of-the-blue phone call from an AI voice agent that immediately starts trying to sell them something.

However, a meaningful number of people will engage with a call that offers them some sort of help, and expansion or reengagement calls work best when they begin as proactive customer support calls. Have your agent open each call by asking the customer how their use of your products is going, and if they need any help.

If the customer says they do require support, then have your agent offer that help, and work through the issue. We’ve found that some customers will have issues that they never got around to contacting support over. Other customers will say they had already been raising issues and trying to reach someone, and felt they were ignored.

Sometimes, after a customer’s issue is resolved, they are willing to stay on the call to hear the upsell, or to book a call with the sales team. Sometimes they aren’t and they just wanted the support, but that’s still a customer who had a real issue and was helped, and is both less likely to churn and more likely to hear you out in the future.  

Most of your customers won’t have a support issue they need resolved, but they still seem to appreciate being checked on before hearing a pitch. Beginning the call from a support perspective makes them feel heard, puts them into a safe space, and helps them feel more open to talk about buying again, or more.

Third, find the right balance between each element of the sales conversation.

Once a customer is willing to hear a sales pitch, the agent can’t push it too hard.

This is tricky to pull off. On the one hand, the agent needs a clearly defined goal for the sales portion of the call. It needs to know it’s trying to land a renewal, or a paid license, or an expanded contract, or to sell a specific new product or feature.

On the other hand, the agent needs to set up that sale. It needs to perform discovery to find the right way to angle their pitch to the needs of the customer. It needs to be able to handle objections. It needs to work with whatever context you have on that customer, their company, and their history with your organization. It needs to know when to try to close the sale itself, and when to pass the customer off to a rep.

Generally, a single model is not going to be able to do all of this (in addition to handling the customer support half of the conversation). We’ve found it’s best to have models for each element of the sales conversation, as well as models to monitor and manage the “team”, that hand off to each other as the call progresses.

It can be challenging and time consuming to build this multi-model ensemble that operates seamlessly behind a single voice agent. From our conversations most internal teams simply won’t have the bandwidth to build, test, and iterate that setup, and do best by working with a tool that comes with all of it out-of-the-box.

A tool like 11x.

Bringing AI voice agents to your campaigns, with 11x

At 11x, we provide voice agents that are purpose-built for multiple use cases, including expanding and reengaging past customers.

Our agents arrive with the best practices from this article built into them, including the ability to rapidly test and iterate against both simulated and real customers. They also operate in full compliance with all legal requirements around AI calls, including consent by jurisdiction, disclosure that it’s an automated call, and honoring an opt-out request as soon as a customer speaks it.

They are also proven in the field, with real customers and real expansion and reengagement campaigns. On average, across industries, they have delivered:

  • 44% of contacts reached
  • 31% of contacts reached engaged in a real conversation
  • 27% of engaged conversations convert (when there is a concrete ask)
  • 4% - 10% of total contacts convert for certain commercial asks

These are accurate numbers, for commercial calls where the customers were not expecting to receive a call at all, let alone a sales call from an AI agent.

To get a sense of how our voice agents, and an expansion or reengagement campaign, might deliver for you, do the math. Run the above numbers against your current book of customers and see what opportunities are available to you.

Or, if you’d rather work through these numbers, define the opportunity, and hear live examples of our voice agents, schedule a demo at 11x.ai.

Written by
Chris Winkler
Head of Marketing
Share this post