Page title and meta description come with every content draft

Content pieces now carry a page title and meta description for whoever publishes them, and ordering an audit can pick a Google Analytics property without leaving the form.

Today was about handoffs: what we hand a partner to publish, what we ask for when an audit is ordered, and what a client sees when they open a request for approval.

For our partners

1. Every content piece now carries a page title and meta description beside the draft, with copy buttons in the editor and on the task page.
2. When a Google account with Analytics is connected, the audit order form lists that account’s properties in place. With none connected, it sends you to Integrations and keeps every answer until you come back.
3. An audit ordered for one location of a larger brand’s website measures that location on its own, and the order form says so before you buy.
4. A website change that covers several pages now opens for clients, with the pages listed together in one row, in the portal and from the emailed review link. A partner reported this one and it is resolved.
5. The Email sending card now says when a domain’s records failed verification rather than asking you to try again later, and Check DNS asks for another look.
6. On the Listings priority list, changing When missing saves.

1. Page title and meta description

The two fields sit in a Meta card in the editor sidebar, with counts that turn when the title passes 60 characters or the description passes 160. They live on the piece itself, so a rewrite or a revision does not lose them, and they appear on the task page next to the featured image. Partners publishing our drafts in their own CMS no longer have to invent them.

2. Analytics while you order an audit

The property is linked to the client’s website when the order is placed, so an order you come back to finish shows it already connected. There is no longer a “we will connect it later” choice: Analytics is connected at order time or it is not, and the audit runs either way.

Behind the scenes

Most of the day went into how a scheduled agent decides what it runs on. An agent now names the service it works and the kind of record a single run takes, and a start that does not fit is turned away with the reason (an agent for campaigns handed an escalation, a campaign whose client is not active). We checked it by generating the next month of scheduled work on the old path and the new one and comparing the two: same work, no differences.

Agents that create their own follow-up work now have to date it. Reading run transcripts, we found a minority of created tasks arriving with no due date, and an undated task sorts to the bottom of the queue. A two-week default was the obvious fix and the wrong one, because the undated tasks were often the urgent ones. The agent is refused, told the rule, and picks a date.

Site speed work got a new tool: our agents can set font loading, defer scripts, make stylesheets non-blocking and add preload hints on a client site using only what is already installed there, for the briefs that rule out adding anything new.