Why is this lead here? Debugging routing just got a lot easier.

Every account and lead routing system eventually produces a question nobody can immediately answer: Why did this lead end up here?

It's one of the most common questions in RevOps, and one of the most time-consuming to resolve. A rep flags an unexpected assignment. An ops manager spots a pattern that doesn't make sense. A new hire can't figure out why their territory looks the way it does.

Each question kicks off the same manual investigation: open the app, find the record, trace the assignment logic, check the target book, look at the queue configuration, review the history. If you're lucky, you find the answer in 20 minutes.

More often it takes longer. And you're still not totally sure you found the root cause.

The real problem with routing debugging

Routing logic is inherently opaque. Once you've built a system with enough conditional logic to handle real-world GTM complexity — territory overlaps, queue priorities, ACV thresholds, rep capacity, geo rules — even the person who built it can't always reconstruct exactly why a given decision got made. It's the nature of complex logic at scale.

The traditional answer to this is documentation, tribal knowledge, and a slow feedback loop. When something routes wrong, someone with deep system context investigates, figures it out, fixes it, and hopes it doesn't happen again. That someone is usually your most senior or tenured RevOps person, and it's not the highest-leverage use of their time.

What the Gradient Works MCP changes

We recently launched the Gradient Works MCP, a server that connects Gradient Works to any AI agent, letting you query and manage your routing setup in plain language.

The routing debugging use case is the clearest demonstration of what this unlocks.

Before: A rep flags that a lead routed to the wrong person. You open Salesforce, debug the flow, trace the logic, check the lead record, look at the assignment history. Forty-five minutes later, maybe you have an answer.

After: You ask your agent: "Why did this lead route to rep A instead of rep B?" The Gradient Works MCP pulls the relevant routing logic, assignment history, target book configuration, and queue state, hands it to your agent, and your agent gives you a plain-language answer. In seconds.

The canonical workflow looks like this:

1. A rep or ops manager flags a routed object
2. You ask your agent what happened
3. The Gradient Works MCP retrieves the relevant data from your environment
4. Your agent explains the routing decision in plain language, and can even suggest a fix if necessary 

(Learn more about Gradient Works Routing here.)

It's not just about speed

The time savings are real, but the more important shift is confidence.

When your team can quickly understand why routing behaved a certain way, they make better decisions about how to change it. They stop assuming. They stop working around rules they don't fully understand. They start trusting the system because they can see inside it.

There's also a floor-lowering effect: because the MCP lets you query routing logic in natural language, it's no longer just the Salesforce admin who can meaningfully engage with how the system works. The ops manager who inherited a routing setup they didn't build. The CSM who gets questions from customers about unexpected assignments. The AE who just wants to know why their lead went somewhere unexpected. All of them can now get answers without escalating to whoever holds the institutional knowledge.

We use it ourselves

We've been using the Gradient Works MCP internally to help debug customer routing issues and it's changed how (and how quickly!) our CS team operates. When a customer asks why a lead routed incorrectly, we can connect to their SFDC org via the MCP, ask the same question, and get an answer in seconds instead of spending half an hour tracing the logic manually.

How to get started

The Gradient Works MCP is available now in alpha, for all Gradient Works customers (free and contracted). You can connect it to Claude Desktop, Claude Code, ChatGPT, OpenAI Codex or Cursor. Setup takes just a few minutes.  

Full technical documentation here: docs.gradient.works/kb/mcp/overview

If you're an existing customer and want help getting connected, reach out to your CSM. We're happy to walk you through it.

Debugging routing has always been harder than it should be. We think that's finally worth fixing.