By The D2C Agent Desk team · Updated October 2026

The weekly ecommerce KPI digest: one page, five numbers, one decision

Most dashboards go unread. A short weekly digest that says what changed, why, and what to do next gets read every Monday.

Founders of small and mid-sized Shopify brands rarely lack data. Shopify analytics, the ad platforms, Klaviyo and the helpdesk all report something. What is missing is a single page, delivered at the same time every week, that compares this week with last week and turns the numbers into a decision. This guide covers what to put in it and how to automate it without letting an AI make up numbers.

The rule: code computes, AI explains

Every figure in the digest should come from a query or a spreadsheet formula you can trace back to its source. AI is useful for the narrative ("revenue fell 8%, almost entirely from fewer paid-social sessions; conversion rate held steady") and for suggesting next actions. It should never be the thing that calculates revenue. If you cannot click through from a number to where it came from, the digest will not be trusted, and an untrusted digest does not get read.

The five core numbers

Start with five and resist adding more until people ask for them:

1. Net revenue (after discounts and refunds), week over week and versus the same week last year if you have it.
2. Orders and average order value. Revenue moves for one of these two reasons; show both.
3. Sessions and conversion rate. Did fewer people come, or did fewer of them buy?
4. Marketing efficiency: total ad spend and a blended measure such as revenue divided by ad spend (often called MER). Platform-reported ROAS is useful context, but platforms tend to over-credit themselves, so blended numbers are the steadier anchor.
5. Contribution margin or gross margin estimate, if you can compute it reliably. Even a rough version beats revenue alone.

Optional sections (add one at a time)

• Top products: top five by revenue, plus any product with a big week-over-week change.
• Inventory risk: SKUs with fewer than N weeks of cover at the current sales rate.
• New versus returning customers: the share of revenue from repeat buyers.
• Support: ticket volume, top three contact reasons, CSAT, and for AI-assisted support the automation rate and top handover reasons.
• Email/SMS: revenue attributed to flows versus campaigns, with the caveat that attribution windows overlap.

The format

A digest that works fits on a phone screen without much scrolling:

WEEK OF OCT 19–25 vs prior week

Net revenue   $18,240   (−6%)
Orders        402       (−4%)   AOV $45.37 (−2%)
Sessions      21,900    (−9%)   CVR 1.84% (+0.1pt)
Ad spend      $4,950    (−3%)   Blended MER 3.7 (−0.1)

WHAT CHANGED
Fewer paid-social sessions (−2,300). Conversion held.
Linen shirt sold out Wednesday; it was 14% of last week's revenue.

ONE DECISION
Restock date for linen shirt? If >3 weeks, pause its ads.

WATCH
Two SKUs under 3 weeks of cover: canvas tote, wool socks.

The figures above are an illustration of the layout, not real results. The structure is the point: numbers, then a few lines of explanation, then one decision.

How to automate it

1. Pull data on a schedule. Shopify (orders, refunds, sessions), ad platforms (spend), and optionally your helpdesk and email tool. Tools like n8n, Make or Zapier can pull these into a Google Sheet; larger brands may already have a warehouse.
2. Compute in the sheet or a script. Week-over-week deltas, AOV, CVR and MER are simple formulas. Lock the definitions in a "definitions" tab so they do not drift.
3. Generate the narrative. Pass the computed table (not raw data) to an AI model with instructions: explain the biggest changes, suggest at most one decision, do not introduce numbers that are not in the table.
4. Review, then send. For the first few weeks, have the digest land as a draft for a human to check before it goes to the team. Once it is reliable, send it automatically every Monday morning.

Common pitfalls

• Time zones. Make sure your "week" uses the store's reporting time zone, the same one Shopify uses, or your numbers will not match the admin.
• Refund timing. Decide whether refunds count in the week of the order or the week of the refund, and stick to it.
• Attribution double counting. Adding up revenue from every channel's own report will exceed your actual revenue. Use Shopify as the revenue source of truth.
• Too many metrics. If the digest needs a table of contents, nobody will read it.
• No owner for the decision. Each "one decision" line should have a name next to it.

Adding support metrics if you use AI agents

If you run an AI support agent, add a small section with automation rate, the top five handover reasons, reopen rate on AI-closed tickets, and CSAT split between AI and human tickets. Those four numbers show whether the agent is really resolving issues or just closing them. Our Gorgias AI Agent audit checklist explains how to use them.

Why weekly, not daily

Daily numbers for a small store are noisy: one wholesale order or one viral post swings everything. Weekly comparisons smooth that out while still being fast enough to act on. Keep daily alerts for true emergencies, such as checkout errors, a top SKU going out of stock, or ad spend spiking, and leave the analysis to Monday.

A narrative prompt you can adapt

You are writing the weekly digest for a Shopify brand.
Input: a table of this week's and last week's metrics with deltas.
Write: (1) "What changed": at most 3 bullets, biggest moves first,
explaining which component drove each (traffic, conversion, AOV,
a product). (2) "One decision": a single concrete question for the
founder. Use only numbers present in the table. If the cause of a
change is unclear from the data, say so instead of guessing.

— The D2C Agent Desk team