Floodwatch: The Rise of Protest Tickets, CX-AI Pricing and Fake Volume

There ARE two trend lines rising in CX: the rise of resolution-based pricing and ticket protests. If and when these two trends collide, there’s a very expensive disaster waiting for the teams that aren’t prepared for it.

Resolution-based pricing

With the rise of AI agents, a new pricing model is showing up across tools. This pay-per-outcome system is a logical setup for both sides. For the AI agent provider, it prevents them from having to choose between high, flat-fee contracts to maintain profitability or eating the insane cost of power-user clients with token bills high enough to fund a whole CX team.

For the buyer, it lets them pay for exactly the capacity they use and not a dollar more. It’s a pricing structure that we’ve long advocated for on the staffing side of things. It’s good to see tools reflecting that reality as well.

It’s also easy to run the hard numbers and see the proof.

As of writing this article, Zendesk lists $1.50 per automated resolution. Intercom lists $0.99 per Fin resolution. Salesforce launched Agentforce at $2 per conversation (not per resolution), then added action-level credits at roughly $0.10 each.

Gartner benchmarks put cost-per-contact for self-service tickets at $1.84 and $13.50 for assisted channels. The resolution capabilities of AI agents probably sit somewhere between self-service and assisted channels. Based on all of that, resolution-based pricing dominates in terms of affordability and value, at least for the cases it can handle.

There’s one scenario though, where the math inverts.

The ticket protest

Oftentimes, online movements will send disgruntled users right to the support queue to make their voices heard. Proactive communication from consumers is fine and can be a useful signal for what needs to be improved in a business.

The problem is that it’s all ending up in the wrong place.

The best example of this is the Consumer Financial Protection Bureau, which releases its complaint numbers publicly. It’s a solid representation of what’s happening in the space overall.

The CFPB logged 6.6 million consumer complaints in 2025, double the 3.2 million it took in 2024 and up from 1.6 million in 2023.

When the Bureau overhauled its portal in June 2026, it said that abuse of the process strains the systems of both the Bureau and companies, slows the processing of legitimate complaints, harms consumers, and wastes resources.

They also specifically cited “Social media influencers with questionable expertise encouraging followers to submit complaints,” and “Adoption of new technologies (e.g., ‘AI tools’) that may act as an individual’s agent” among the causes for the volume increase.

It doesn’t take much to find plenty more examples of exactly that kind of behavior happening in other industries. A quick search can reveal plenty of influencers directing their followers to support pages, giving them specific templates to express a grievance, and even showing how the process can be automated with AI tools.

Coordinated false reporting isn’t new. In 2022, Meta removed roughly 2,800 accounts, Groups and Pages in a single Indonesian mass-reporting network.

What is new is the scale of these reports and the tools available to those reporting. Imperva has already found that more than half of all web traffic is automated.

All of these factors combined can end up creating a nightmare scenario.

The protest runs up the meter

Protestors file these tickets as open-ended complaints. Once that complaint is made, they’re likely to abandon the queue as quickly as they came to it. That means their ticket gets closed automatically as an assumed resolution and billed accordingly.

Each ticket satisfies the billing definition and resolves nothing, because there was nothing to resolve.

Add AI tools on the consumer end, and you end up in the expensive (and very strange) scenario where the company and the consumer are essentially paying for their bots to duke it out in a support queue.

For the online brigaders, they’re paying in crowdsourced time and tokens. For businesses, it’s all coming out of the same CX budget. You get hit with thousands of empty tickets, each with a receipt attached.

So, what’s the solution here?

Cost per resolution still makes sense

This protest-ticketing scenario can make resolution-based pricing look riskier than it actually is. For most of what hits a queue, metered pricing works great. Usually, a customer has a problem, automation fixes it, and you pay for that outcome directly. If there’s no automatic solution, the ticket gets passed up to the right person, and you get the best results for your dollars.

This risk is specific and uncommon. Of course, at a large enough scale, “uncommon” is basically a guarantee at some point, which is why it’s worth knowing about. But that doesn’t mean the day-to-day has to change. You just need a plan.

The failsafe is boring, simple, and effective

The solution is to sort the spam so you can service the customers. Build one path that doesn’t run through the meter.

Plain automation, without AI, is inflexible, predictable, and costs virtually nothing to run. That includes:

• Routing and classification: bucket the campaign on arrival.
• Deduplication: give one acknowledgment per account, not per message.
• Keyword clustering: hashtags and common phrases help you understand what’s happening.
• Escalation rules: pull the real problems out of the flood.

It’s the classic machinery CX has been using for years. Automate away the junk, pay for the time it takes to solve what’s left. That’s how you find the needle in the haystack when you need to, without paying for the redundant volume around it.