Finding knowledge gaps in Genesys Cloud

What can Genesys Cloud already tell you about knowledge?

Quite a lot, and it is worth using before you add anything. A typical Genesys Cloud contact centre has three sources of knowledge signal built in:

• Knowledge usage. Genesys Knowledge reports on how articles are used by bots and agent-facing knowledge surfaces, including queries that returned no good answer. This is the clearest view of where existing content is hard to find.
• Speech and Text Analytics. Transcription turns calls into text, and topics count the conversations that contain phrases you define. If you suspect a problem with "direct debit failed", a topic will tell you how often customers say it.
• Performance data. Queue and interaction views show handle time, hold, after-call work, transfers and wrap-up codes. Spikes in any of them often point at a topic agents are struggling with.

Teams that use these well already know their busiest topics and their most-opened articles. That is a strong base. The question is what sits outside it.

Where does native reporting stop?

Each of those sources answers a narrower question than "what is our knowledge base missing?"

• Knowledge usage only sees queries. An unanswered query requires someone to ask the knowledge base. Agents who answer from memory, ask a team leader or have stopped looking never create one. We cover this blind spot in failed searches vs knowledge gaps.
• Topics only find what you named. A topic is a phrase list you wrote. It is excellent at measuring a known issue and silent about a new one. Nobody writes a topic for a question they haven't heard of yet.
• Wrap-up codes are chosen under time pressure. They describe the category of the contact, usually from a fixed list. "Billing enquiry" doesn't say which billing question went unanswered.
• Nothing joins demand to content. Handle time tells you a queue is slow. It doesn't tell you that the slow calls share an intent with no matching article in Genesys Knowledge.

None of this is a shortcoming of Genesys Cloud. A contact centre platform records and routes conversations, and a knowledge base serves articles. Gap detection is the step between them: reading what customers needed and checking it against what the knowledge base holds.

How does gap detection connect to Genesys Cloud?

Knowledge Quest connects with a Client Credentials OAuth grant. Your Genesys administrator creates a dedicated OAuth client under Admin, Integrations, OAuth, and enters four values in Knowledge Quest: region, organisation ID, client ID and client secret. A test confirms the connection before anything is read. Credentials are write-only and stored in Azure Key Vault.

Use a client that serves Knowledge Quest alone. Sharing one with another integration makes permissions and audit harder to follow. Pick the region that matches your Genesys organisation, not your office; a wrong region is the most common reason a healthy connection reads nothing.

Once connected, Knowledge Quest reads:

• Transcripts for voice, and the content of email, messaging and callback interactions.
• Interaction records for every conversation, so counts are complete even before a transcript is available.
• Queues, which you map to knowledge areas so each team's conversations are checked against the right articles.
• Timing and wrap-up data such as talk, hold and after-call work, used to weight each gap by how difficult its conversations were.

Interaction metadata is kept to an allowlist of non-personal fields, and agents are referenced by their Genesys user ID. Your routing, flows and agent desktop don't change.

What happens to each Genesys conversation?

Every conversation goes through the same four stages:

1. Analysis. A summary, whether the contact was resolved, partly resolved or unresolved, and whether a callback was needed.
2. Sentiment. A satisfaction score for the conversation.
3. Classification. The conversation is placed in your knowledge taxonomy, so volume and outcomes can be seen by topic.
4. Intents. Each question the customer raised is extracted in plain language, and each one is checked against your articles.

The last stage is where gaps are found. For each intent, the Knowledge Gap Evaluator retrieves candidate articles, compares them side by side, and decides whether one answers the question or whether this is a gap. The decision is recorded with the candidates it considered and a confidence score, so a reviewer can see why a question was marked as covered or not.

A single call can contain three intents: the reason for calling, a follow-up the agent handled well, and a second question nobody could answer. Gap detection keeps all three. Topic counts and wrap-up codes usually keep one.

Which Genesys signals make a gap expensive?

A gap that agents handle well costs little. A gap that sends every call to hold, transfer or callback costs a great deal. Genesys Cloud records the signals that separate the two, and they are what turn a list of gaps into a ranked backlog:

Handle time is broken into talk, hold, after-call work and connect time, so you can see where the minutes go by queue and by topic. Each gap is then ranked by impact: the conversations it affects, weighted by the friction those conversations carried. Our knowledge gap analysis guide sets out the ranking in full, and the cost of a knowledge gap shows how to put a dollar figure on one with your own cost per contact.

How do fixes get back into Genesys Knowledge?

Finding gaps is half the loop. The other half is getting a reviewed answer in front of agents and bots, in the knowledge base they already use.

Knowledge Quest syncs both ways with Genesys Knowledge:

• Pull. Your existing Genesys Knowledge articles come in, so every conversation is checked against the content agents actually see. Edits made in Genesys come back on the next check.
• Draft. For each gap, Knowledge Copilot drafts a new article or an update from the conversations that exposed it, with every statement linked to its source.
• Review. Your subject-matter experts approve the draft. Six quality checks (brand alignment, readability, validity, duplicates, contradictions and complexity) run before anything goes out.
• Push. The approved article goes to Genesys Knowledge. Categories map to your taxonomy, so it lands in the right place.

A sync dashboard shows which articles are in sync, changed on one side, changed on both, or missing from Genesys, and lets your team push, pull or resolve each one. Knowledge Quest never deletes an article because it disappeared from Genesys; it flags the difference and leaves the decision with you. Only the default variation of a Genesys article is synced, which suits most knowledge bases.

Teams that use other knowledge tools alongside Genesys Cloud can export approved articles, including a Genesys-compatible format, or sync to a supported knowledge base instead. Adding gap detection to your knowledge base compares the two routes.

What does the first month on Genesys Cloud look like?

1. Connect and map. Create the OAuth client, test the connection, and assign each queue to a knowledge area. Queues without a knowledge area aren't analysed, so this step matters.
2. Bring in your articles. Your Genesys Knowledge content is imported so gap matching runs against what agents read today.
3. Let a few weeks of conversations build up. A month covers billing cycles and weekly patterns, which is enough for a reliable ranking.
4. Review the top gaps. For each one, read the evidence and choose to create, update, link or dismiss.
5. Close and measure. Push approved articles to Genesys Knowledge and watch transfers, callbacks and repeat contacts on each topic fall, using a fair before-and-after comparison.

Nothing about your Genesys Cloud estate is replaced. The conversations you already record become the evidence for what to write next, and Genesys Knowledge gets the answers customers were asking for. That is what it means to make your Genesys knowledge first class. See how Knowledge Quest finds the gaps.