The Revenue You're Losing After the Click: Forrest Anderson's Case for AI Customer Service
The economics of affiliate marketing are front-loaded. You optimize the click, the creative, the offer. By the time a customer converts, most programs consider the hard part done.
Then they call in.
They've been on hold for 15 minutes. The agent who picks up is working from a script they don't fully understand, handling a product they've never seen, following a refund policy they may or may not agree with. The customer is already frustrated. The call doesn't go well. The refund goes through anyway, or the customer hangs up and disputes the charge with their bank instead.
Forrest Anderson spent over a decade on the paying side of that problem. He's been an advertiser, a traffic buyer, a client of customer service companies, and he watched the same failure repeat across programs of every size. So he built an AI to fix it: Customer2.ai, known as C2, a platform that replaces unpredictable human agents with AI that follows your policy every single time, at any scale.
What he found when he started looking was worse than he expected. What he built in response is what this conversation is about.
The Audit That Started Everything
The origin of Customer2.ai isn't a tech story. It's a frustration story.
Forrest and his team were running offers and fielding a growing volume of email complaints about customer service. They decided to investigate. Instead of forwarding calls directly to their customer service provider, they intercepted and recorded them for a period, then looked at what they found.
Hold times ranged from 10 to 20 minutes. Despite paying for domestic agents during business hours, almost all calls were being routed offshore regardless of when they came in. Agents were either caving immediately on refund requests or arguing with customers and ignoring the policies Forrest's team had set. Neither outcome was acceptable.
The problem wasn't that customer service was bad. The problem was that it was unpredictable. The same call, handled two different ways by two different agents, produced two different results, neither one necessarily aligned with how the business wanted to operate.
So they built something that would follow the script every time.
Revenue Protection, Not Revenue Generation
C2's positioning is deliberate. Forrest is clear that his platform isn't in the business of driving conversions. That work is already done by the time a customer picks up the phone. What C2 does is protect the revenue those conversions represent.
The mechanism is straightforward: faster resolution, consistent policy adherence, no hold times, no language barriers, no agent who's been on the phone for eight hours and has stopped caring. When a customer calls in with a question and gets a quick, accurate answer, the frustration that drives refund demands disappears before it can build.
"The immediate bottom line is we started giving about 10% less refunds almost right away."
That number comes from ease of use as much as anything else. Customers who reach someone quickly, get their question answered, and feel heard don't escalate. The ones who wait 15 minutes and then fight with an agent who sounds like they're reading from a card do.
Call handling times dropped as well. Because the AI pulls up a customer's full account history the moment it receives their number, there's no intake process, no repeat yourself three times. The call starts where the customer is, not at the beginning of a script.
What Happens When a Customer Is Happy
Not every call is a complaint. Some customers call to track an order, or just to ask a question, and they're satisfied. C2 handles those calls too, and Forrest has configured his system to use them.
If a customer calls about a brain supplement order and they're clearly happy with the product, the AI can offer the complementary sleep supplement, explain why they go together, take the credit card information, and process the sale. The customer gets something they want. The program gets revenue it wouldn't have seen otherwise.
The same logic applies in reverse: if a customer isn't happy with a product, the AI can offer a store credit, a partial refund, or a product swap before reaching for a full refund. For programs with thin margins on refunds, that flexibility has a direct impact on the bottom line.
A Policy That Never Bends Unless You Tell It To
The core differentiator Forrest keeps returning to isn't the AI itself. It's the consistency.
Human agents wear down. A customer who calls back three times, escalates their tone, and threatens to dispute the charge will eventually find an agent willing to break policy just to end the call. It happens in every call center. The cost shows up in the refund rate and the chargeback rate, and it usually takes a month of data to notice.
For programs with aggressive refund policies, that's a meaningful protection. For programs that want to give a full refund the moment someone threatens to call their bank, that's configurable too. C2 can be set to detect specific trigger phrases, bypass the standard refund flow, and issue an immediate refund to prevent a chargeback. The cost of that refund is lower than the chargeback fee, and the customer doesn't become a dispute on your processor's record.
Keyword monitoring extends to reporting: if a customer mentions a specific phrase, the system can flag the call automatically and send an email to the client. Forrest has seen it used for everything from legal threats to product quality complaints. The program gets visibility into calls it would otherwise never review.
Traffic Spikes Don't Have to Break Your Support Operation
One of the structural problems with human customer service in the affiliate space is volatility. A strong month of affiliate traffic means more customers, more questions, more calls. A slow month means idle agents you're still paying for.
Staffing customer service for an affiliate program means picking a number and living with it being wrong in both directions.
C2 bills per minute, comparable to offshore call center rates, but without the headcount. A program that goes from 100 calls a week to 10,000 doesn't need to hire anyone. A program that drops back down doesn't need to cut anyone.
"If you scale overnight to 100 times what you're doing today, they'll just eat it and be just fine. If you scale down the next day to almost nothing, nobody needs to be fired because nobody was hired for it."
For performance marketers accustomed to watching their metrics shift dramatically week to week, that kind of operational flexibility changes the math on what customer service costs.
Getting Started Without the Risk
Forrest's recommendation for programs that are skeptical is simple: start small.
Pick the lowest-volume offer on your roster, the one where the downside of a bad call is minimal, and let C2 handle it. Set the instructions according to your current policy. Let the system run, monitor the metrics, and see whether the results match what he's describing.
C2 offers a self-service portal for programs that want to configure agents themselves, as well as a full onboarding team for those who'd rather hand it off. Their own website runs a live demo agent in the bottom right corner that new prospects can test before they ever speak with a sales rep.
The Revenue That Never Had to Leave
The affiliate industry spends an enormous amount of time thinking about the front end of performance: traffic quality, creative, offers, attribution, conversion rates. The back end, what happens after someone buys, gets a fraction of that attention.
That imbalance is expensive. A customer who converts and then has a bad service experience doesn't just generate a refund. They potentially generate a chargeback, a negative review, and a permanently closed door to any future purchase. The value of that customer, across a lifetime, goes to zero.
Forrest built Customer2.ai because he was tired of watching revenue leave through a door he should have been able to close. Ten-plus years in the space gave him a clear view of every place that door opens: hold times, language barriers, off-script agents, inconsistent policy enforcement, the slow accumulation of small failures that add up to a meaningful number on your monthly P&L.
The fix isn't complicated. It's consistent, fast, policy-compliant customer service that scales with your program instead of fighting it. Everything else follows from there.
