Shopify’s analytics can show you almost anything about your store. That is the problem. A merchant on the Shopify Community described the dashboard as “click-heavy” and asked how others manage; the honest answer is that most people either look at everything and decide nothing, or look at total sales and call it a review. Neither helps.
A useful weekly review answers four questions in a fixed order: what changed, why, what is about to happen, and what will we do about it. It fits on one screen. It takes fifteen minutes on a Monday, and since the Summer 2026 Edition, Sidekick can produce most of it as a table and a chart from one prompt.
The five numbers, and why only five
Every week, compare these five for the last seven complete days against the seven before:
- Total sales. The headline, but never alone.
- Orders. If sales rose and orders did not, the average order value did the work, which means a different action.
- Average order value. Watch this after any change to shipping thresholds, bundles or discounts.
- Conversion rate. Sessions that became orders. The number most sensitive to site problems, stockouts on best sellers and traffic quality.
- Sessions. Traffic. Sidekick cannot tell you where it came from beyond Shopify’s channel and referrer data, but it can tell you whether the volume moved.
Why not more? Because five numbers give you a story and twenty give you a spreadsheet. Refund rate, returning-customer share and channel mix are worth a monthly look, not a weekly one, unless something in the five points at them.
Complete days, same timezone, same weekday alignment
Three small rules stop the review lying to you. Use complete days, so Monday morning’s review ends on Sunday; a partial day makes the current week look worse than it is. Use the shop’s timezone, so orders placed late in the evening land on the right day. And when you look at a single day, compare it with the same weekday a week earlier, because a Saturday and a Tuesday are different businesses.
The prompt that does all of this:
Act as my store’s operations analyst. Review the last 7 complete days against the 7 before, in the shop’s timezone, using the store’s own analytics only. Pull total sales, orders, average order value, conversion rate, sessions and returning-customer rate for both periods in one request. Show a table with this period, previous period and percentage change. Never compute a percentage change from a zero baseline; say when a period has too little data.
That is the first half of /weekly-review.
What counts as a real change
A 4% move in weekly sales on a store doing 60 orders a week is noise. A 4% move on 3,000 orders is a signal. A rough rule for small stores: treat anything under 10% week on week as weather unless it repeats for three weeks, and treat anything over 20% as something to explain today.
Explaining it means looking at the other four numbers. Sales down, orders down, conversion flat, sessions down: you have a traffic problem, and it lives outside Shopify in your ads, your email or a search ranking. Sales down, sessions flat, conversion down: something on the site changed, a best seller sold out, a discount ended, or the checkout has a problem. Sales up, orders flat, AOV up: a bundle or threshold is working; find out which and do more of it.
Products: the top five and the falling five
The second half of the review is about products, because that is where the actions usually are.
Pull sales by product for both periods. Show the top 5 products by revenue with revenue and percentage change, then the 5 products whose revenue fell most. For each of the top 5, check current inventory and flag anything with fewer than 14 days of stock at the current sales rate.
Two things to look for. A best seller with under two weeks of stock is the most expensive thing on the page: reorder it today, before the review is over. And a product falling for three weeks in a row is either seasonal, out of stock in a popular variant, or has lost its ad or its search position; each has a different fix, and the first step is noticing.
For a deeper look when something is off, /top-products ranks by revenue and units across 30 days and flags stock on every winner.
Turning the review into three actions
A review with no actions is a report. The rule we use: three actions, each tied to a number on the screen, each naming a product, channel or segment and a verb.
From the numbers above only, write three specific actions for this week. Each names a product, channel or segment, the number behind it, and what to do: restock, feature on the homepage, discount, email a segment, or fix a page. Do not create discounts, edit products or send emails; recommend and wait.
Examples of what a good action looks like: “Restock the Classic Tee in M and L before Thursday: 18 units left against 31 sold last week.” “Feature the Tote on the homepage: revenue up 48% and 140 units in stock.” “Email the 212 customers who bought the Enamel Mug in the last year: it fell 41% this week and there is a new colour.” Generic actions like “improve marketing” mean the numbers did not support anything specific, which is also useful to know.
When the review shows a flat week with nothing obvious, /boost-sales is the follow-up: it searches the last 30 days for opportunities you would not spot week to week, such as products with high revenue but few orders that suit a bundle.
The monthly additions
Once a month, add three checks to the same session:
- Returning-customer share and second-purchase rate. If fewer than one in five of your buyers ever comes back, retention is the cheapest growth you have; the win-back skills start there.
- Refund rate by product. A product with a refund rate double the store average has a sizing, description or quality problem.
- Dead stock. Cash sitting on the shelf. /dead-stock values it and proposes a clearance action per product.
What Sidekick cannot show you
Sidekick reads Shopify analytics only. It does not see Google Analytics, Meta Ads or Klaviyo, so return on ad spend and email revenue by campaign are not in this review. If the five numbers point at traffic, the next stop is your ad platform, or the Shopify connector for ChatGPT or Claude with an export of both, which can hold two datasets at once; our Sidekick vs ChatGPT vs Claude comparison explains when that is worth it.
The fifteen-minute version
Monday morning: run /weekly-review. Read the table. Decide whether the change is noise or signal. Read the top five and falling five, and reorder anything under two weeks of stock. Read the three actions and put them in your task list. If the week was flat and you have five minutes left, run /boost-sales. Done.
The first time you do this it takes longer, because you are learning what your store’s normal looks like. By the fourth Monday you know what a good week and a bad week look like on one screen, and that is the whole point.
Three mistakes that make the review useless
Comparing with last year on a small store. Year-on-year comparisons are useful once you have a few thousand orders a year. Below that, a single wholesale order or a viral post from last October makes this October look like a collapse. Compare with last week and, for seasonality, with the same week last year only as context.
Reacting to one week. A single down week is not a trend. The rule of three above exists because most merchants who change their ads, their pricing or their homepage after one bad week undo a thing that was working. Note it, watch it, act on the third repeat.
Reading sales without reading stock. The most common cause of a falling best seller is that a popular size or colour sold out ten days ago. Every product line in the review carries a stock flag for this reason; if a product is falling and a variant is at zero, the fix is a purchase order, not a marketing plan.
One more, for stores that sell in person too: the review blends online and POS by default. If the channels behave differently, ask Sidekick to split them, or the online number will be masked by a good Saturday in the shop.