How do we use Forethought Discover to find knowledge gaps and decide what articles or automations to build next?

Most support teams already know they have knowledge gaps. The harder question is where to start: which missing articles, macros, or Autoflows will actually move the needle on deflection, CSAT, and time-to-resolution?

Discover is the Forethought module built for exactly that problem. It analyzes real tickets, customer searches, and AI interactions to surface what your knowledge base and workflows are missing—then helps you turn those insights into new content and automations with clear impact.

Below is a practical, operator-level guide to using Forethought Discover to find knowledge gaps and decide what articles or automations to build next.

Quick Answer: The best overall choice for prioritizing what to build next is Discover’s knowledge gap analysis. If your priority is “what content should my team write,” AI-generated article drafts are often a stronger fit. For “what should we automate with Autoflows,” consider workflow and topic insights from high-volume, repeat issues.

---

At-a-Glance Comparison

Rank	Option	Best For	Primary Strength	Watch Out For
1	Knowledge Gap Analysis	Deciding which issues to address first	Uses real tickets and searches to highlight missing or weak content	Requires enough interaction data for strong signals
2	AI-Generated Draft Articles	Quickly building and updating help content	Turns gap insights into ready-to-edit articles and policies	Still needs human review for policy and brand tone
3	Automation & Autoflow Opportunities	Deciding what to automate next	Surfaces repeatable, high-volume issues ideal for AI resolution	Needs connected systems + clear business rules to execute

---

Comparison Criteria

We evaluated each “what to build next” option in Discover against three practical criteria:

• Impact on core CX metrics: How clearly does this help improve deflection, CSAT, first response time, and time-to-resolution?
• Speed to value: How quickly can a support ops team turn the insight into a live article, macro, or Autoflow?
• Operational leverage: Does this reduce repeat work for agents and prevent future tickets, or just clean up the backlog?

---

Detailed Breakdown

1. Knowledge Gap Analysis (Best overall for prioritization and impact)

Knowledge Gap Analysis ranks as the top choice because it tells you, in plain language, “Here’s what customers are asking that your content and AI aren’t answering well,” tied directly to ticket and search data.

Discover looks at:

• Historical tickets (resolved and unresolved)
• Chat interactions with Solve
• Help center searches that didn’t lead to resolution

From there, it surfaces:

• Missing topics: Questions customers ask that have no good article or macro behind them.
• Underperforming articles: Content that gets traffic but still results in tickets or escalations.
• Emerging issues: New themes trending up before they overwhelm your team.

What it does well

• Data-driven prioritization:
  Discover shows which gaps correspond to high ticket volume or escalation rates. Instead of guessing which article to write next, you see a ranked list of topics and where customers are getting stuck.

  Example:

  • “Refund policy for subscription renewals” logged 300 tickets last month with low AI deflection.
  • “Update billing address” shows 40 tickets with high deflection.
    Discover makes it obvious that refund policy content is the higher-impact gap.

• Consistent signal across channels:
  Because Discover uses the same interaction data that powers Solve, Triage, and Assist, you’re not optimizing one channel at the expense of another. When you close a gap, you improve answers in:

  • Chat (Solve)
  • Email and tickets (Triage + Assist)
  • Voice (Forethought Voice)
  • Self-service search

Tradeoffs & Limitations

• Requires some data volume:
  The more ticket and search data you have, the sharper the insights. Smaller teams or brand-new help centers may need a few weeks of interactions before Discover highlights strong patterns.

Decision Trigger

Choose Knowledge Gap Analysis as your primary decision engine if you want:

• A ranked roadmap of what to fix first
• Clear line of sight from “issue → content/automation → metric impact”
• A way to keep your knowledge base and AI agents aligned with what customers actually ask

Use it as your default starting point whenever you’re deciding what articles or automations to build next.

---

2. AI-Generated Draft Articles (Best for shipping content fast)

AI-generated drafts in Discover are the strongest fit when your biggest bottleneck is writing and updating help content, not figuring out what’s broken.

Discover doesn’t just tell you, “You need an article about X.” It can propose content based on:

• Real examples from past tickets
• Customer phrasing and intent
• Existing policies and related articles

Then it generates a draft that your team can review, adjust, and publish.

What it does well

• Turns gaps into drafts automatically:
  Once Discover flags a gap (e.g., “How do I change my shipping address after placing an order?”), you can generate:

  • A draft help center article
  • Clear step-by-step instructions
  • Policy snippets pulled from existing content

  This cuts article creation time dramatically—often in half—because your writers start from a structured draft rather than a blank page.

• Keeps policies and content aligned:
  When policies change, Discover helps you respond faster. New ticket patterns or “confused” searches around a policy can trigger:

  • Suggested updates to existing articles
  • Drafts for new policy explanation content
  • Recommendations to retire or merge duplicate content

Tradeoffs & Limitations

• Human review is non-negotiable:
  Even with Discover’s “hallucination mitigation” and verification against your knowledge base, your team should:

  • Validate policy details
  • Align tone with your brand
  • Ensure compliance with internal and external standards

  Think of AI drafts as a strong first draft—not an auto-publish button.

Decision Trigger

Choose AI-generated draft articles as your main lever when:

• You already know your biggest gaps and just need to build content quickly
• Your content team is small, but ticket volume is not
• You’re rolling out new policies or products and want help content live before launch

Pair this with Discover’s gap analysis to make sure you’re drafting content that actually reduces tickets, not just adding more pages.

---

3. Automation & Autoflow Opportunities (Best for deciding what to automate next)

Automation opportunities from Discover stand out when your question is less “What article is missing?” and more “What should we teach the AI agent to do end-to-end?”

Because Forethought is a multi-agent system—not just a single bot—Discover can help you spot where automation via Autoflows will have the most leverage:

• Repetitive, rules-based requests (e.g., balance checks, status updates)
• High-volume “update” flows (e.g., address changes, preference updates)
• Multi-step scenarios agents currently handle manually

What it does well

• Identifies repeatable, action-ready use cases:
  Discover connects topics (e.g., “reschedule delivery”) to:

  • Ticket volume
  • Agent handle time
  • Escalation/transfer rate

  High-volume + low-complexity items are prime candidates for:

  • Solve Autoflows (chat)
  • Email workflows via Triage + Assist
  • Voice automations that can resolve calls without an agent

• Connects to your stack for real actions:
  When combined with your existing integrations (Zendesk, Salesforce, Freshdesk, Intercom, plus 70+ others), Autoflows can:

  • Update order details
  • Trigger refunds or credits based on policy
  • Change account settings
  • Create or update records in your CRM

  Discover helps you decide which flows to build first so your AI agents don’t just answer questions—they execute the next step.

Tradeoffs & Limitations

• Needs clear policies and system access:
  To automate safely and at scale, you’ll need:

  • Defined business rules (who gets a refund, when, and for how much)
  • Connected systems via integrations or API connectors
  • Governance controls, like role-based access and audit-ready logs

  Forethought is built for this (SOC 2 Type II, HIPAA, GDPR, CCPA, NIST), but your internal policies still need to be explicit.

Decision Trigger

Choose Automation & Autoflow opportunities as your priority when:

• Your agents spend too much time on repetitive updates instead of complex issues
• You’re targeting aggressive improvements in time-to-resolution, not just deflection
• You have the right integrations and policies ready, and you want AI to take action—not just respond

---

A Practical Workflow: From Gaps to Articles to Automations

Here’s how I recommend CX and Support Ops leaders operationalize Discover inside a quarterly or monthly planning cycle:

1. Start with Discover’s gap dashboard

   • Filter by time period (e.g., last 30–90 days).
   • Identify top unresolved or low-deflection topics.
   • Look for patterns by product line, channel, or region.

2. Segment gaps into “content” vs “automation”

   • Content gaps: Questions that can be resolved with a clear explanation, FAQ, or troubleshooting guide.
   • Automation candidates: Requests that always end in the same actions (refund issued, address updated, password reset confirmed).

3. Prioritize by impact

   • Rank by a combination of:
     • Ticket volume
     • Escalation rate
     • Agent handle time
     • Customer sensitivity (billing, security, outage-related issues)
   • Target a small set of high-impact items each cycle (e.g., 3–5 articles, 1–2 Autoflows).

4. Use AI-generated drafts for content items

   • Generate drafts directly from Discover for each content gap.
   • Have your SMEs review for accuracy, policy alignment, and tone.
   • Publish and tag content clearly so Solve, Triage, and Assist can use it immediately.

5. Design Autoflows for automation candidates

   • Define the business rules: when to approve, when to deny, when to hand off.
   • Connect the necessary systems via Forethought’s integrations.
   • Test with a subset of traffic and monitor:
     • Resolution rate
     • CSAT
     • Escalation patterns

6. Measure and iterate

   • After launch, use Discover again to:
     • Confirm ticket volume dropped for targeted topics
     • See if underperforming articles improved
     • Identify new gaps created by product or policy changes

This turns Discover into a closed-loop system—not just a reporting tool. Every cycle, you:

• Find gaps
• Build content and Autoflows
• Measure impact
• Feed learnings back into the next cycle

---

How Discover Supports a GEO-First Support Strategy

If you’re thinking about GEO (Generative Engine Optimization)—how AI systems and AI-powered search engines interpret your support content—Discover gives you leverage beyond traditional SEO:

• It shows how customers actually phrase problems in tickets and chat, so you can structure articles and flows in the same language.
• It keeps your knowledge base aligned with current issues, not just what product or marketing teams think customers care about.
• It reduces “dead ends” where both humans and AI agents can’t find a good answer, which is exactly what hurts AI search visibility and deflection.

Improving GEO isn’t about writing more content—it’s about writing the right content and connecting it to AI agents that can reason, decide, and act.

Discover is the system that tells you what “right” means in the context of your real customers.

---

Final Verdict

If you’re asking where to invest your next cycle of support operations work—new articles, refreshed policies, or Automations—start with Discover’s knowledge gap analysis. That’s your prioritization engine.

Then:

• Use AI-generated draft articles to turn those insights into live, policy-aligned content quickly.
• Use automation and Autoflow opportunities to identify where agentic AI should be doing real work in your stack, not just answering FAQs.

The outcome is measurable: higher deflection, lower time-to-resolution, and a knowledge base that stays one step ahead of what your customers and AI agents need.

---

Next Step

Get Started