You Coached the Agent. You Should Have Fixed the Article.

A QA fail lands on your desk. You book 20 minutes with the agent. You play the call. You walk through the right steps. They nod. You both leave feeling like the job got done.

Next month a different agent fails the same way. Same process. Same gap. Same coaching conversation with a new name on it.

That is not a people problem that keeps recurring. That is a page problem you keep converting into coaching.

Coaching is the expensive way to fix a sentence

I have run a lot of those sessions. Some of them were necessary. The agent rushed. They skipped a mandatory question. They knew the process and didn’t follow it. Headphones on, that is a behaviour conversation, and no article writes that out of someone.

A lot of them were not that. The article was vague at the exact fork they hit. Two pages disagreed. The screenshot was from the old screen. The exception wasn’t written, so they guessed. They searched, opened the first result, and that result was last year’s version still sitting there looking official.

In those cases the coaching was theatre. You corrected one person on a call that already happened. The next person to hit the same page will do the same thing, because the page did not change.

Multiply that by a floor and you get a quality program that looks busy and a process that never moves. Agents start to treat coaching as weather. It happens to someone every week. It doesn’t change what they open on the next call.

Ask one question before you book the meeting

When you find a miss, look at the article before you look at the person.

Would a careful reader, following this page cold, have done the right thing? If the answer is no, or “maybe, if they already knew”, the page is the defect. Fix that first. Then decide whether this agent also needs a conversation.

That sounds obvious. It is the step that gets skipped, because coaching has a calendar slot and a form, and rewriting an article feels like extra work that isn’t in the QA process. So the system rewards the meeting and starves the edit.

The edit is the bit that scales. One clearer paragraph saves you the next five sessions on the same miss. Required reading on the new version at least asks the rest of the floor to look. A tick still isn’t proof they understood. It is better than hoping they overhear the coaching you did with someone else.

If three people have failed the same step this quarter, stop stacking coaching notes. You have a content bug.

What the page fix actually is

One current version. If search still returns the old one next to the new one, you haven’t fixed anything. You’ve added a second opinion.

Write the step they missed, in the order they hit it, in the words they would search. Not the policy language that survived legal. The thing they needed in the sixth minute of the call.

Put an owner and a review date on it. A page you corrected once and abandoned will be wrong again, and you will be back in the same meeting with a new agent, wondering why the floor “isn’t getting it”.

Publishing that update is not the same as the floor landing it. Do both. The rewrite without the landing is how you get a beautiful page and the same fail next month.

The honest bit

A tool can show you the article they had open. It can show you that three other people searched the same thing and bounced. It can hold the corrected version and tell you who hasn’t read it. I built those things because I got tired of having the same conversation in a small room.

A tool cannot sit in the coaching session and decide whether this miss was sloppy or the page was bad. That judgement is yours. And some misses really are the person. Pretending every fail is a content bug is how you stop holding people to a standard.

The tell is repetition. One miss is a call. The same miss on a third agent is a page. Fix the page, then coach whoever still needs it. The other way around is how quality stays busy and the floor stays the same.

The KnowledgeScout Team