Short answer
Keep your weekly dashboard to seven lines, with comparisons, an agreed alert threshold and an owner for each. Spend Monday discussing exceptions.
The answer in detail
A weekly KPI dashboard needs no more than seven lines. Compare each with the previous week, the corresponding week last year, and the target. Agree on an alert threshold beforehand and name an owner. Move everything else into the monthly report.
The benchmark
The KPI selection guide covers which measures to follow. This page concerns the format: columns, thresholds, and Monday's review.
| KPI | This week | vs. last week | vs. last year | vs. target | Alert threshold | Owner | What changed |
|---|---|---|---|---|---|---|---|
| Human sessions, excluding bots | |||||||
| Revenue per visit | |||||||
| Conversion rate | |||||||
| Average order value | |||||||
| Orders | |||||||
| Bestseller availability | |||||||
| Unattributed traffic |
Keep the final column to one sentence. For availability, measure the share of key products with common sizes in stock. A product available only in XS and XXL may technically be in stock while having little remaining sales potential.
The rule
Each comparison answers a different question. Last week reveals a break, but a promotional week will generally outperform an ordinary preceding week. Last year helps with seasonality when you compare Monday-to-Sunday weeks and annotate occasions whose dates move, such as Easter. The target shows whether execution is behind plan. Use the promotional and holiday calendar of the market being measured; French statutory sales periods are not an international calendar.
Read lines together. In April 2026, an apparel brand saw sessions fall 30.5% and transactions fall 40.6% from March, with average order value stable near €133. Separate readings might have launched two investigations. The stable basket helped direct attention toward incoming traffic: paid-social activity had been reduced 94%, removing traffic to two bestsellers and some high-intent visits. These original euro figures are case evidence, not global benchmarks.
Set each threshold as a relative change before reading the results, using normal variation over the preceding eight to twelve weeks. Discuss only threshold breaches, with the named owner bringing an explanation.
The steps
- Choose one source per line. Use Shopify for orders, sales, and average order value; GA4 for unattributed traffic. Choose one session source consistently. The Shopify-versus-GA4 guide helps. Align the property and store time zones.
- Calculate thresholds. With twelve weekly values in B2:B13, oldest first, Google Sheets formula
=MAX(ARRAYFORMULA(ABS(B3:B13/B2:B12-1)))returns the largest week-to-week variation. Exclude an exceptional promotion from the comparison and recalculate. If normal maximum variation is 9%, an illustrative threshold is 10%. This is a practical rule, to revisit after a quarter, and assumes nonzero denominators. - Configure alerts. GA4 supports up to fifty custom insights per property with email notifications. The source's October 2026 route is Home > Insights > View all insights > Create > Start from scratch, with Editor access. Choose weekly evaluation, the metric, a percentage-decrease condition, and the owner's address. Weekly anomaly detection needs thirty-two weeks of history; on a younger property, prefer an explicit threshold.
- Review every Monday. Allow twenty minutes for flagged lines, one explanation or hypothesis per owner, and an entry in the operating log.
- Freeze the format. During major trading events, keep lines and thresholds stable unless measurement breaks.
The pitfall
Dashboards grow because additions rarely require removals. Six months later, an important break disappears among twenty-five metrics. Remove a line whenever you add one.
A single stock threshold can also mislead. One basics-and-lingerie brand shared inventory between its website and wholesale. A proposed 70% sell-through alert missed the risk that one retailer replenishment order could move a product from 35% to above 70%. For those families, Monday is too late: lower the threshold or monitor daily.
Shopify's September 21–23, 2026 measurement update changed sessions and conversion-rate definitions, while orders and sales remained unaffected. Until late September 2027, affected year-over-year comparisons straddle definitions. Flag the break above the dashboard; consult the update guide.
What to tell your team
This example is fictional:
“We are moving from twenty-two KPIs to seven, each with a threshold and an owner. This week only revenue per visit crossed its threshold, falling 14% against normal variation of 6%. Its owner will explain the change on Monday.”
