Agent escalations now end in a written record

When an agent raises its hand, the triage now ends in a record anyone can open instead of a conversation in chat. Clients with more than one website can switch between them from the portal.

Most of today went into the machinery around our agents: what happens when one stops and asks for help, and what we can see afterwards about what it did.

For our partners

A client with more than one website on the platform can now switch between them from the portal. The business and the website the portal is about sit at the top of the sidebar, and picking a different one moves the campaign pages, the website pages and the editor with it. With one website, nothing changes.

Campaign scope lines now name what they are. A campaign with four blog posts plus two content pieces of any type reads as four blog posts and two additional content pages, instead of the same line printed twice, and the client’s page and our own view of it say the same thing.

Behind the scenes

1. An escalation now ends in a written record anyone can open, rather than in whatever conversation happened to be going on.
2. Changing an ignore rule rescores a crawl instead of re-running the audit, so an issue someone wrote down keeps its identity.
3. What a run filed reads in one place: the tasks, plans, asks and escalations it produced, what is held for review, and every count opens the items themselves.

1. Triage that leaves a record

When an agent raises its hand, the useful part is the explanation of why. That explanation used to live wherever the person working it out happened to be typing. There is now one path: read the escalation and everything the agent had in front of it, say what kind of thing happened and what caused it, decide whether anything needs fixing, and write the result where the escalations list links to it. A report that blames a defect and points nowhere is refused. Three real escalations went through it today.

2. Issues that keep their identity

Ignoring an issue used to re-run every check over a stored crawl, which handed every issue on it a new identity. Now the rules are applied to the issues that already exist and the score is worked out again, so the issue quoted this morning is the same issue this afternoon. That came out of the first triage runs, which is the point of having them.