Explore Amazon SES sending performance by stream and provider

Build me an app for exploring how our Amazon SES sending is actually performing. I open it, pick a date range, and answer one question: which of our email streams is landing and which is not. It is a drill down surface for investigating, not a scheduled report, so everything on screen reacts to the date range picker.

The top of the app is a date range picker with quick presets (last 7 days, last 30 days, month to date) plus a row of headline tiles for the selected range: sends, deliveries, bounces, complaints, opens and clicks. Pull these from Amazon SES Batch Get Metric Data for the account as a whole. Next to the bounce and complaint tiles show the rates as a percentage of sends, compared against the thresholds Amazon enforces: warn when bounce rate goes above 2 percent or complaint rate above 0.05 percent, and show danger at the levels where Amazon takes action on the account, 5 percent bounce and 0.1 percent complaint. Make both thresholds editable in a small settings area of the app.

Below the tiles, two breakdown tables built from the same metric data. The first is per configuration set: call List Configuration Sets to get every set, then request metric data broken down by configuration set so I can put transactional sending next to marketing sending in one table. Columns are sends, deliveries, bounce rate, complaint rate, open rate and click rate, sortable, with the same threshold colouring on the rate columns. Batch Get Metric Data accepts a limited number of queries per call, so batch and page the requests behind a single handler rather than firing one call per set from the client.

The second table is per mailbox provider, using the same metric data broken down by provider, so I can see when Gmail or Outlook alone is rejecting us while every other provider is fine. Same columns, sorted by volume by default. Highlight any provider whose bounce or complaint rate is materially worse than the account average, because that pattern means a provider specific problem rather than a list quality problem.

A side panel shows account standing and open issues. Use List Reputation Entities and Get Reputation Entity for the sending status and reputation policy of the account and its identities, and List Recommendations for the open deliverability issues Amazon has raised, showing each one's type, impact and status. Let me acknowledge a recommendation with a short note. Store the acknowledgement and note inside the app, keyed to the recommendation, with who acknowledged it and when, and keep acknowledged items visible but dimmed rather than hiding them. Acknowledging must not change anything in Amazon SES.

The same panel lists every sending identity from List Email Identities, enriched with Get Email Identity, showing whether it is a domain or an address, its verification status, and its DKIM status and signing key type. Sort unverified or DKIM failing identities to the top, since those are the ones doing damage right now.

A domain panel shows inbox placement and engagement per sending domain for the selected range using Get Domain Statistics Report: inbox versus spam placement, read rate, delete rate and complaint volume.

Add an Archive snapshot button. Clicking it appends one row to a Google Sheet using Append Values, containing the date range, the headline totals, the bounce and complaint rates, and a timestamp, so the team accumulates the long term history Amazon does not retain. Let me set the spreadsheet and tab in the app's settings, write a header row when the tab is empty, and confirm in the UI when the row lands. Give me an option to append the per configuration set rows for the same range as well.

Call Get Deliverability Dashboard Options when the app loads. If the deliverability dashboard is switched off, the mailbox provider table and the domain inbox placement panel must say so in plain words, explaining that those views need the Amazon SES deliverability dashboard enabled and where it is turned on, rather than rendering blank tables or zeroed charts. Do the same when a range genuinely has no activity: say there was no sending in this range instead of presenting zeros as if they were results.

Keep it exploratory and read only against Amazon SES. The only writes anywhere are the snapshot row in Google Sheets and the acknowledgement notes stored in the app. Changing the date range refreshes the tiles, both tables and the domain panel together.