Article Suggestion

Support Knowledge Base

5 min read

Also known as: Suggested Articles, KB Recommendations, Related Articles

An article suggestion is a knowledge base article surfaced to an agent or customer in real time based on the active ticket, query, or chat context.

Definition

An article suggestion is a recommendation engine output that surfaces the most relevant help center article to whoever is solving a problem right now — either a support agent inside a ticket view or a customer typing into a help widget. The suggestion is driven by intent signals: ticket subject, message body, product area, customer history, or the words being typed in real time.

In practice, your knowledge base sits behind a matching layer that watches what's happening in the conversation and pushes 2-5 candidate articles into the agent sidebar or the self-service search. Agents can paste the link, send the snippet, or mark the article as 'didn't help' to retrain the matcher. On the customer side, suggestions show up as 'related articles' on the contact form to deflect tickets before they're submitted.

Article suggestion is narrower than 'knowledge base search' (which is user-initiated) and broader than 'canned response' (which is a fixed template, not a published article). It's the connective tissue between your KB and your live support surface.

Why It Matters

Every suggestion that resolves a ticket or deflects a contact is time your team doesn't spend typing the same answer for the 400th time. Mid-market support orgs typically see 15-30% ticket deflection and 20-40% faster handle times once article suggestions are tuned, which directly compounds into lower cost per ticket and faster first response. It also makes your KB investment actually pay off — articles that no one ever finds are wasted writing.

Skip this and your knowledge base becomes a graveyard. Agents default to writing custom replies from scratch, answers drift between reps, customers escalate issues that were already documented, and leadership keeps asking why headcount needs to grow with ticket volume. Worst case: you keep funding KB content that never gets surfaced where decisions are made.

Examples in Practice

A SaaS support team turns on article suggestions in their agent console. When a ticket comes in mentioning 'SSO login loop,' three KB articles auto-surface in the sidebar. The agent sends the top one with a one-line greeting and closes the ticket in under two minutes instead of the usual eight.

A DTC ecommerce brand adds article suggestions to the pre-submit step of their contact form. Shoppers typing 'where is my order' see three tracking-related articles before they finish the sentence; roughly 22% close the form without submitting a ticket, cutting peak-season volume meaningfully.

A B2B fintech routes new signups through an onboarding chat. As the customer asks about wire transfer limits, the bot suggests the compliance article and the limit-increase request flow, resolving the question without ever pulling in a human agent.

Frequently Asked Questions

What is an article suggestion and why does it matter?

It's a knowledge base article surfaced in real time to an agent or customer based on the conversation context. It matters because it turns your KB from a passive library into an active deflection and assist tool, cutting handle time and ticket volume without adding headcount. Teams that tune suggestions well often deflect a quarter of inbound contacts.

How is article suggestion different from knowledge base search?

Search is user-initiated — someone types a query and gets results. Article suggestion is system-initiated — the platform reads the context (ticket subject, chat message, form field) and pushes recommendations without anyone searching. Suggestions reduce the cognitive load of knowing what to look for, which is exactly the friction that causes customers to submit tickets instead of self-serving.

When should I use article suggestions?

Turn them on once you have at least 30-50 well-written KB articles covering your top ticket drivers. Below that threshold, suggestions miss too often and agents stop trusting the sidebar. Use them across three surfaces in order of ROI: the agent console first, the pre-submit contact form second, and the in-product help widget third.

What metrics measure article suggestion performance?

Track suggestion click-through rate (agents or customers who open a suggested article), deflection rate (contact forms abandoned after viewing a suggestion), article helpfulness votes, and time-to-resolution on tickets where a suggestion was used versus not. Also watch coverage: the percentage of tickets where any suggestion was offered at all.

What's the typical cost of article suggestion tooling?

It's usually bundled into your help desk or KB platform rather than priced standalone, so the real cost is content production and tuning. Plan on a content writer at roughly 20-40 hours per month for a mid-market KB, plus a few hours weekly from a support ops lead reviewing low-performing suggestions and retiring or rewriting stale articles.

What tools handle article suggestions?

Most modern help desk platforms include native suggestion engines tied to their KB module. Standalone knowledge management platforms and AI support copilots also offer this, often with better matching quality. The category to evaluate is 'agent assist' or 'AI knowledge' — look for tools that match on semantic meaning, not just keyword overlap, and that learn from agent feedback.

How do I implement article suggestions for a small team?

Start by auditing your top 20 ticket drivers and confirming each one has a clear, current KB article. Turn suggestions on in your agent view only, and have agents thumbs-up or thumbs-down each one for two weeks. Use that signal to rewrite weak articles and add missing ones, then expand suggestions to your contact form once accuracy is solid.

What's the biggest mistake teams make with article suggestions?

Turning them on before the KB is ready. If suggestions surface outdated, off-topic, or poorly written articles, agents ignore the sidebar within a week and customers lose trust in self-service. The fix is content-first: audit, rewrite, and retire articles based on suggestion performance data before pushing the feature wider.

Can article suggestions work for outbound sales, not just support?

Yes — the same matching logic applies to sales enablement libraries. When a prospect asks a technical question in a discovery call or replies to an outbound email with an objection, suggestion engines can surface the right case study, one-pager, or FAQ for the rep to send. This is increasingly common in CRM platforms with built-in AI assist.

How often should I refresh the articles being suggested?

Review suggestion performance monthly. Articles with high impression counts but low click-through or helpfulness rates need rewriting; articles that consistently resolve tickets should be promoted and linked from related content. Plan a full KB audit every six months to retire anything tied to deprecated features or outdated policies.

The AMW Suite