A practical guide to Intercom article suggestions

Trying to keep up with customer questions can feel like you're running on a treadmill that just keeps speeding up. As your business grows, the volume of support tickets inevitably follows. The default solution is often to hire more support agents, but that gets expensive fast and isn't a sustainable way to scale.

This is where self-service support comes in. The idea is to give customers the tools they need to find answers on their own, which frees up your team for more complex problems. Intercom has a built-in feature called article suggestions that aims to do just that by automatically recommending help center content.

It’s a solid feature and a great first step into self-service. But what happens when your needs get a bit more complicated? This guide will walk you through how Intercom’s feature works, where you might start to feel its limitations, and what a more powerful, integrated AI solution can bring to the table.

What are Intercom article suggestions?

At its core, Intercom article suggestions is a system that lives inside the Intercom Messenger. When a customer starts typing a question or lands on a specific page, the system automatically suggests relevant articles from your help center. The goal is simple: give customers instant answers so they don’t have to wait for a human agent, reducing the number of conversations in your team's inbox.

A view of the Intercom Messenger where article suggestions would typically appear to assist a customer.

Intercom has refined this over the years, evolving it from a feature called "Smart Suggestions" into a more advanced AI-driven component of their AI agent, Fin. But the fundamental concept hasn't changed. It's about using automation to help customers help themselves, so your team can focus on the stuff that really needs a human touch.

How Intercom article suggestions work

Getting article suggestions up and running in Intercom is pretty straightforward. It’s designed to be an accessible way to deflect common questions without needing a team of engineers. Let’s take a look at how it actually operates.

Setting up Intercom article suggestions

As an admin, you can add an "Article search" card directly into the Messenger Home, putting your help content front and center. When a customer opens the chat, Intercom looks at a few things to decide which articles to show. It uses machine learning, data about the user (like their subscription plan), and the specific pages they've been looking at to personalize the suggestions.

You can also set up audience rules to get more specific. For instance, you could decide to only show articles about advanced reporting features to customers on your enterprise plan. This helps make sure people are seeing content that’s actually relevant to them.

The AI improvement loop

This is where things get a bit more advanced, particularly with Intercom's "Suggestions" feature powered by its Fin AI Agent. The system is designed to get smarter over time by learning from the conversations it can't solve.

Here’s how that works. Imagine Fin, the AI agent, tries to answer a customer's question but fails. The ticket gets escalated to a human agent who then resolves it. The system sees this happen and analyzes the conversation to figure out what went wrong. Based on that analysis, it will suggest ways to plug the knowledge gap it discovered. These suggestions might include:

• Creating a new help article. The system might notice multiple customers asking a question that isn't covered in your knowledge base.

• Updating existing content. It could flag an article that might be confusing or out of date based on customer interactions.

• Cleaning up duplicates. It can also identify when you have multiple articles saying the same thing or when information is contradictory.

This feedback loop is definitely useful, but it's important to recognize that it's a reactive process. The system only learns and suggests improvements after a customer has already hit a dead end and a human had to get involved.

Key limitations of Intercom article suggestions

For smaller teams just dipping their toes into self-service, Intercom’s built-in tools can be a big help. But as your company grows and your support operations become more complex, you'll likely bump into a few limitations that can make it harder to scale effectively.

Your knowledge is stuck in a silo

Probably the biggest challenge with Intercom’s native tools is that they can only suggest articles from one place: the Intercom Help Center. Now, be honest, is that where all of your company's knowledge actually lives? I'm guessing not.

This diagram illustrates how a dedicated AI can pull from many knowledge sources, unlike the siloed approach of some built-in tools.

Think about it. The detailed technical guides your engineers wrote are probably sitting in Confluence. Your company’s official policies and procedures are likely stored in a dozen different Google Docs. And what about all the incredibly useful troubleshooting tips and workarounds that get shared in Slack every single day? All of that valuable information is completely invisible to Intercom’s suggestion engine.

This means your AI is working with one hand tied behind its back. It has an incomplete picture of your business and can only answer questions based on the content you've manually curated in one specific location. This leads to more escalations and a frustrating "I don't know" for your customers when the answer actually exists elsewhere.

The reactive improvement process

As we touched on earlier, Intercom's AI gets better by learning from its failures. This means you have to wait for customers to run into problems and for your agents to step in before you can even identify a knowledge gap. You end up constantly playing catch-up, patching holes in your knowledge base after customers have already fallen through them.

A more proactive approach would be to train an AI on your entire history of customer conversations from the very beginning, learning from both the wins and the losses. Intercom also doesn't have a strong simulation environment, which makes it tough to test changes and see how they’ll perform before you roll them out to your entire customer base.

The surprising pricing for Intercom article suggestions

While the basic article suggestion feature is part of the platform, unlocking Intercom's more powerful AI with the Fin AI Agent comes with a pricing model that can make budgeting a real challenge. On top of the monthly fee you pay for each support agent, Intercom charges $0.99 for every resolution the AI handles.

Let that sink in for a second. As your support volume increases, or as your AI gets better and resolves more tickets, your bill goes up. You're effectively penalized for your own success. This kind of unpredictable, usage-based fee makes it incredibly difficult to forecast your costs and might even make you hesitant to automate as much as you'd like.

A more powerful approach with a dedicated AI platform

These are the exact kinds of problems that dedicated AI platforms are built to solve. A tool like eesel AI is designed to overcome these challenges by plugging into the helpdesk you already use, like Intercom, and giving it a major boost. It works with your existing tools, not against them.

Unify all your knowledge for better Intercom article suggestions

Instead of being locked into a single help center, eesel AI connects to over 100 different knowledge sources right out of the box. You can connect your Intercom Help Center, of course, but you can also pull in knowledge from places like Confluence, Google Docs, past support tickets, and internal wikis. This gives your AI access to your company's complete collective wisdom, leading to more accurate answers, higher resolution rates, and happier customers.

Improve Intercom article suggestions with proactive learning and simulation

From day one, eesel AI starts training on your entire history of past support tickets. It automatically learns your brand's voice, understands the common issues your customers face, and identifies the solutions that have worked before. There's no need to wait around for failed conversations to begin the learning process.

Even better, eesel AI comes with a powerful simulation mode. Before you switch the AI on for your customers, you can test it against thousands of your past tickets. This shows you exactly how it would have responded, giving you an accurate forecast of its resolution rate and potential cost savings. It completely removes the guesswork and risk from launching AI support.

Take full control of Intercom article suggestions with a customizable workflow engine

Answering questions is only part of the puzzle. eesel AI provides a full workflow engine that lets you define custom actions, create a unique AI persona, and decide exactly which types of tickets you want to automate. You can start small, maybe by having the AI only handle simple "password reset" requests while escalating everything else. As you get more comfortable, you can gradually give it more responsibility. This level of fine-tuned control just isn't possible with simpler, built-in tools.

An example of a visual workflow builder that allows for detailed customization of AI behavior and actions.

Pricing comparison for Intercom article suggestions: Intercom vs. eesel AI

Clear and predictable pricing is a must when you're choosing a tool to build your support strategy on. Let's see how the two platforms compare.

Intercom article suggestions pricing

Intercom uses a hybrid model that mixes a monthly per-agent fee with a usage-based fee for its AI.

Source: Intercom Pricing Page

eesel AI pricing

eesel AI keeps things simple with monthly plans based on AI interaction volume, with no hidden fees.

Go beyond basic Intercom article suggestions

Intercom article suggestions are a great starting point for companies wanting to explore self-service support. The feature is easy to use and fits nicely within the Intercom ecosystem.

But for growing teams that have knowledge scattered across different platforms, need predictable costs, and want powerful, risk-free tools for testing, a dedicated AI platform is the next logical step.

eesel AI isn't here to replace your helpdesk; it's an upgrade that plugs into the tools you already love to unlock their true potential. By bringing all your knowledge together and giving you complete control, it helps you build a support experience that can truly grow with you.

Ready to unify your knowledge and automate support with confidence? Get started with eesel AI in minutes.