All guides

Operations, Ops Scorecard, about 5 minutes

Reading the Ops Scorecard

The Ops Scorecard shows how each operator uses the platform. It counts the AI drafts they actioned and how fast they answered. It also shows how much of the AI's work is being used at all. It needs a manager login.

Open the scorecard

Why this matters: it is the only screen that measures the people side of the system rather than the properties.

  1. Sign in at app.hostassist.ai with a manager account.
    A specialist login opens the address but sees an empty page. That is a permission limit, not a bug.
  2. Click Operations in the left menu, then click Ops Scorecard.
    You should see a page headed Ops Scorecard, with four tiles and a table below them.
The Ops Scorecard screen with a window selector, four summary tiles for active operators, drafts actioned, draft adoption and pricing adoption, and the operators table underneath.
Step 2 Window selector top right, four tiles, then one row per operator.

Choose the window

Why this matters: every number on the screen is counted inside the window you pick, including the table.

  1. In the top right, click 7d, 14d, 30d or 90d.
    30d is selected when you arrive. The table heading changes to match, for example operators, last 30 days.
  2. Use 7d for a stand up and 90d for a review.
    Short windows on a small team swing a lot, so do not read a 7d dip as a trend.
The window selector with 7d, 14d, 30d and 90d, with 30d selected.
Steps 3 and 4 One control, four choices. Everything below it obeys the choice.

Read the four tiles

Why this matters: adoption is the number that says whether the AI work is being used or quietly ignored.

  1. Read active operators, shown as a count out of the team size.
    0 of 4 means nobody on the team actioned anything inside the window.
  2. Read drafts actioned and the small line under it.
    The line gives the total drafts created in the window, which is the pool the count came from.
  3. Read draft adoption and pricing adoption.
    Adoption is the share of drafts that reached an outcome and were actioned by a person rather than left to expire.
  4. Read the sentence under the tiles.
    It splits the guest lane and the pricing lane into ignored, auto sent or auto approved, and still waiting.
Four tiles: active operators 0 of 4, drafts actioned 0 with 9 drafts in 30 days, draft adoption with no decided items yet, and pricing adoption with no decided items yet.
Steps 5 to 8 No decided items yet means nothing has finished, so a percentage would be made up.
The sentence under the tiles defining adoption as the share of AI drafts that reached an outcome and were actioned by a person, split into a guest lane and a pricing lane.
Step 8 The definition sits on the screen so nobody has to guess what adoption counts.

Read the operators table

Why this matters: this is where a team average turns into a name, which is the only level where you can actually coach.

  1. Find the operator you care about in the left column.
    Each row shows their role badge and their email under the name.
  2. Read approved, edited, rejected and escalated across that row.
    Those four count guest message drafts the operator actioned.
  3. Read the pricing column next to them.
    It counts pricing queue decisions, kept apart from guest drafts on purpose.
  4. Read median response and median pricing.
    Median response is the middle time from a draft being created to the operator acting on it.
  5. Read team share and last action on the right.
    Team share is that person's slice of the team's actioned work in the window.
The operators table listing four accounts with their roles and emails, and columns for approved, edited, rejected, escalated, pricing decisions, median response, median pricing, team share and last action.
Steps 9 to 13 A dash in a cell means no data yet in this window. It does not mean zero performance.

Three things that catch people out

Why adoption is measured, and not just volume

Counting drafts sent flatters the system. If the AI writes a hundred drafts and every one is ignored until it expires, the volume looks healthy and the product is doing nothing.

Adoption asks the harder question: of the work that reached an ending, how much did a person actually use. That number moves only when the AI is genuinely helping somebody.