Build a store health report in Slack

A new Slack report is a common request for Athena: she writes the job, posts a sample for review and puts the report on a schedule. In this walkthrough she builds Store Health 2.0 for Harbor & Pine, a daily card laid out like the Meta optimizer's cards. Each card shows:

• conversion rate on a 30-day chart against the prior 30 days

• the day's key errors and the checkout steps where shoppers drop off

• numbered notes on the chart for launches, promotions and site changes

• one or two proposed tests for the landing pages with the most Meta spend, based on what Clarity shows

The tests are proposals for a person to review. The report does not change the site.

3. Build a new Slack report

3. Build a new Slack report

Before you ask#

1. Find the report you want the new one to look like. The demo request points to the channel where the Meta optimizer posts its cards. You can also link one card: hover over it, open More actions (the three dots) and choose Copy link.

2. Decide which conversion rate the report shows. Store Health 2.0 uses Shopify orders divided by GA4 sessions, and the card says so. Naming one definition keeps the chart from mixing measures.

3. Decide where the report posts and when.

Ask Athena#

4. Open #gd-athena-engineering, tag @Athena and describe the report: its name and brand, the report to copy, what each card shows, where the numbers come from, and how you will check it. The request from the demo:

   @Athena build Store Health 2.0 for Harbor & Pine, laid out like Mandy's optimizer cards in #mandy-meta-optimizations. Each daily card should show:
   - conversion rate on a 30-day chart against the prior 30 days
   - the key errors and checkout drop-off points found that day
   - annotations on the chart for site changes and other big updates (launches, promos, theme edits)
   - one or two proposed landing-page tests for the top-spending Meta landing pages, based on what Clarity shows (rage clicks, dead clicks, scroll depth)
   Use Harbor & Pine's demo data and name the source of every number. Post one sample card in this thread for review, then schedule it for 8:00 AM PT daily in #harborpine-reports.

   

5. Send it. Athena replies in a thread under your request. Her status message shows the model, the command she is running and running counts, and it updates as she works.

Follow the build#

6. Each test run gets its own reply, and her notes explain the choices she makes about the data. Store Health 2.0 took 20 minutes, 68 commands and 4 test runs, all passing.

7. To add detail, reply in the thread and tag her, for example to say which landing pages count.

Check the sample card#

8. The sample card posts in the thread as Nova, the agent that runs the daily reports. Check it against your request:

   • the chart shows the conversion rate you named, for the right periods

   • the errors and drop-off steps are specific enough to act on

   • the numbered notes sit on the right dates

   • each proposed test names the page, the change to try and the Clarity finding behind it

   • every number names its source

   

   

9. Read her answer. It links the sample and confirms the schedule: Store Health 2.0 replaced the existing daily store-health report at 8:00 AM Pacific in #harborpine-reports, starting the next morning. She shipped it as a commit on the machine, the way that machine ships changes; there was no pull request.

10. She set the schedule in the same task, right after the sample, because the request asked for both. If the card needs changes, reply in the thread and tag her. If you would rather approve the sample before it goes on the schedule, say so in the request, for example "wait for my OK before you schedule it".