Post-Black Friday analysis: the five metrics to check on Monday
From the work/ 14 min read
The Monday after Black Friday: five metrics, where each one lives in Shopify and GA4, how to compare them and a Node script that turns them into a December list.
- Analytics
- CRO
- Shopify
In this article (10)
A post-Black Friday analysis fits in five metrics: conversion rate by device, average order value, drop-off at each checkout step, zero-result searches and payment errors. Compare each one with a normal weekend and with last year, and whatever came out worse becomes your work list for December.
Black Friday 2026 is Friday, November 27, so the Monday analysis lands on the 30th. Below is where each number lives in Shopify and GA4, two ShopifyQL queries, and a Node script that compares the three periods and builds the list. Every example number is illustrative.
Why run the Black Friday analysis on Monday
By Monday, Friday, Saturday and Sunday are closed, and you still remember what changed: which app updated, which discount went live late, what time something got stuck. A week later those details are gone and the analysis shrinks to "we sold more" or "we sold less", which tells nobody what to fix.
A number on its own says nothing. Read every metric against two baselines, always with the same days of the week:
- A normal weekend: Friday, November 13 to Sunday, November 15, 2026, if you ran no promotions. If you ran a campaign that weekend, pick an earlier one with no offers. The week right before Black Friday is a poor baseline: many stores already run early deals and shoppers wait.
- Last year: Black Friday 2025 was Friday, November 28, so the range is November 28 to 30, 2025. Compare Friday with Friday, not date with date.
One warning about comparing with 2025. Shopify's behavior reports page says the Shopify Analytics session measurement rollout ran from September 21 to 23, 2026, that it can shift sessions, conversion rate and other session-based metrics, and that this doesn't necessarily mean your traffic, orders or customer behavior changed. Sessions in 2025 and 2026 are not measured the same way. Against last year, compare orders, average order value, searches and payments; compare conversion against the normal weekend, which already falls after the change.
These are the five metrics, with their formula and where to get each number:
| Metric | Formula | Shopify | GA4 |
|---|---|---|---|
| Conversion rate by device | sessions that completed checkout ÷ sessions | Sessions by device and Conversion rate breakdown | Funnel exploration broken down by Device category |
| Average order value | (gross sales − discounts) ÷ orders | Average order value over time | Not needed: Shopify owns the orders |
| Checkout completion | sessions that completed ÷ sessions that reached checkout | Conversion rate breakdown | Funnel exploration from begin_checkout to purchase |
| Zero-result searches | searches with no results ÷ searches | Top online store searches with no results | Custom search_no_results event |
| Payment errors | failed payments ÷ orders × 100 | Orders > Abandoned checkouts, in the Timeline | Not visible: a failed payment never arrives as a purchase |
Scroll the figure sideways to see all of it
Conversion rate by device
A store-wide conversion rate hides what happens on each screen. Shopify defines it as the percentage of sessions that resulted in a purchase and splits it by device in two Analytics reports: Sessions by device, which shows the devices people used to reach the store, and Conversion rate breakdown, the funnel of sessions with cart additions, sessions that reached checkout and sessions that completed checkout. Your order count and your completed-checkout sessions may differ, because one session can hold more than one purchase.
If your admin's report pages show the query editor (Shopify announced it in its changelog on June 11, 2025), you can pull everything into one table with the ShopifyQL sessions schema. The query follows the documented syntax (version 2026-10); we did not run it against a store:
FROM sessions
SHOW sessions, sessions_that_reached_checkout, sessions_that_completed_checkout, conversion_rate, checkout_conversion_rate
GROUP BY session_device_type WITH TOTALS
SINCE 2026-11-27 UNTIL 2026-11-29
ORDER BY sessions DESC
FROM sessionsreads the online store sessions schema.SHOWasks for sessions, sessions that reached checkout, sessions that completed it, and two rates:conversion_rate(completed ÷ sessions) andcheckout_conversion_rate(completed ÷ reached checkout).GROUP BY session_device_type WITH TOTALSreturns one row per device type (Desktop, Mobile, Tablet) plus a totals row.SINCE 2026-11-27 UNTIL 2026-11-29is Friday to Sunday. Dates are unquoted, inyyyy-MM-ddformat.ORDER BY sessions DESCputs the busiest device first.
Run the same query with SINCE 2026-11-13 UNTIL 2026-11-15 for the normal weekend. How to read it: if mobile drops while desktop rises (illustrative: 1.9% against 2.2% on mobile, 3.9% against 3.4% on desktop), the extra traffic arrived but something on the phone stopped the purchase: a promo popup covering the button, a product page weighed down by banners, a payment method that does not show. If the problem is the form itself, review mobile checkout autocomplete and keyboards.
Average order value
The Average order value over time report in Shopify's sales reports calculates AOV as gross sales minus discounts, divided by orders, excluding post-order adjustments such as edits or exchanges. Shipping and taxes are not part of it. The report can compare date ranges, and in ShopifyQL the comparison goes in the same query with COMPARE TO (also not run against a store):
FROM sales
SHOW orders, gross_sales, discounts, average_order_value
TIMESERIES day WITH TOTALS, PERCENT_CHANGE
SINCE 2026-11-27 UNTIL 2026-11-29
COMPARE TO 2026-11-13 UNTIL 2026-11-15
ORDER BY day ASC
FROM salesreads the sales schema, whereaverage_order_valueis (gross_sales−discounts) ÷orders.SHOWbrings orders, gross sales and discounts next to AOV, so you can see where a change comes from.TIMESERIES day WITH TOTALS, PERCENT_CHANGEreturns one row per day, the range totals and the percent change against the comparison.COMPARE TO 2026-11-13 UNTIL 2026-11-15is the normal weekend with fixed dates. For last year, replace the line withCOMPARE TO previous_year_match_day_of_week, which compares against the same weekday a year earlier.ORDER BY day ASCkeeps the days in order.
A lower AOV with deeper discounts is expected. What matters is separating the two: gross sales per order and discount per order. In the illustrative example, AOV was 982.52 against 859.81 on a normal weekend (+14%) and only +1% against last year, with discounts at 22% of gross sales against 6% normally. If AOV drops by more than the discount grew, people bought fewer items per order: review bundles, cart recommendations and the free shipping threshold.
Scroll the figure sideways to see all of it
Checkout abandonment, step by step
Conversion rate breakdown tells you how many sessions reached checkout and how many completed it. With the Open Funnel toggle off, the funnel is closed: it only counts sessions that went through every step in order. What it does not tell you is which step inside checkout people left on.
For that, use a GA4 funnel exploration. It takes up to 10 steps; for checkout, use the recommended GA4 ecommerce events:
begin_checkoutadd_shipping_infoadd_payment_infopurchase
Leave Make open funnel off so it only counts people who enter at the first step, put Device category in the breakdown, and turn on Show elapsed time to see how long people take between steps. Each step shows its abandonment rate, the share of users lost before the next one. To compare with the normal weekend, change the date range and note both results.
Before you build it, confirm which events arrive. Shopify publishes standard checkout events (checkout_started, checkout_shipping_info_submitted, payment_info_submitted, checkout_completed and others), but which ones reach GA4, and under which names, depends on your integration. Check the events report; if one is missing, the way to verify it in DebugView is in the GA4 add_to_cart event article. And keep sources apart: GA4 counts users and Shopify counts sessions, so their numbers will not match (the reasons are in why GA4 and Shopify don't match). Use each source for its own comparison.
How to read it: a big drop between add_shipping_info and add_payment_info is usually a shipping cost or delivery time that surprises people; between add_payment_info and purchase, it is the payment (next section). Fix the step first, then recover what is left with a well-timed abandoned cart email sequence.
Zero-result searches
On Black Friday people search for the promotion's name, for something they saw in an ad, or for a product that already sold out. Shopify has the Top online store searches with no results report, and its searches-by-query report warns that data can lag by up to 72 hours: check this one Monday night or Tuesday. The metric is searches with no results divided by total searches.
If you send GA4 your own event for empty searches, like the one built in zero-result searches in GA4, its weekly report of the ten most common terms already gives you the list. Illustrative: 14% of searches with no results against 8% on a normal weekend. December fixes usually come in three kinds: synonyms for the promotion's terms, a redirect for searches like "black friday" or "deals", and visible alternatives on sold-out products.
Payment errors
Here you need to know where to look. The Shopify Help Center says a payment that fails at checkout usually doesn't create an order: it is logged as an abandoned checkout, with the attempt in its Timeline, and only counts as abandoned after more than 10 minutes incomplete. That is why it shows up in no sales report, and GA4 never sees it as a purchase.
Go to Orders > Abandoned checkouts, open the weekend's checkouts and expand the failed payment event: the Message field carries the bank's decline, if there was one. Sort what you find into three piles:
- Bank declines (insufficient funds, expired card): normal on any weekend; a concern only if they jumped.
- Payment provider errors clustered in one hour: check Shopify's status page for that hour.
- Fraud blocks ("Blocked due to high risk of fraud" in the Timeline): if they were real buyers, review your rules.
The metric: failed payments per 100 orders. Illustrative: 12.9 against 5.6 on a normal weekend. Counting them by hand on a Monday works once; to know while it happens, checkout conversion alerts check failed payments every few minutes.
Turn it into the December list
With the three periods written down, the script compares every metric and flags the ones that came out more than 10% worse. It runs on Node 22 with no dependencies:
// bf-review.mjs: the Black Friday weekend against a normal weekend and last year.
// Illustrative numbers: replace them with the ones from your reports.
const periods = {
blackFriday: {
sessions: { mobile: 9400, desktop: 2600 },
completed: { mobile: 179, desktop: 101 },
reachedCheckout: 760, orders: 286, grossSales: 362000, discounts: 81000,
searches: 1500, zeroResults: 210, failedPayments: 37,
},
normal: {
sessions: { mobile: 3100, desktop: 1100 },
completed: { mobile: 68, desktop: 37 },
reachedCheckout: 262, orders: 107, grossSales: 98000, discounts: 6000,
searches: 520, zeroResults: 42, failedPayments: 6,
},
lastYear: {
sessions: { mobile: 7600, desktop: 2500 },
completed: { mobile: 167, desktop: 95 },
reachedCheckout: 640, orders: 265, grossSales: 330000, discounts: 72000,
searches: 1200, zeroResults: 130, failedPayments: 19,
},
};
const THRESHOLD = 0.1; // flag a metric that is more than 10% worse
const sum = (o) => Object.values(o).reduce((a, b) => a + b, 0);
const ratio = (a, b) => (b ? a / b : NaN);
const pct = (v) => `${(v * 100).toFixed(1)}%`;
const metrics = [
...Object.keys(periods.blackFriday.sessions).map((device) => ({
name: `conversion ${device}`, better: "up", sessionBased: true, fmt: pct,
calc: (p) => ratio(p.completed[device], p.sessions[device]),
})),
{ name: "average order value", better: "up", fmt: (v) => v.toFixed(2),
calc: (p) => ratio(p.grossSales - p.discounts, p.orders) },
{ name: "checkout completion", better: "up", sessionBased: true, fmt: pct,
calc: (p) => ratio(sum(p.completed), p.reachedCheckout) },
{ name: "zero-result searches", better: "down", fmt: pct,
calc: (p) => ratio(p.zeroResults, p.searches) },
{ name: "failed payments / 100 orders", better: "down", fmt: (v) => v.toFixed(1),
calc: (p) => 100 * ratio(p.failedPayments, p.orders) },
];
const change = (now, base) => (now - base) / base;
const worse = (m, c) => (m.better === "up" ? c < -THRESHOLD : c > THRESHOLD);
const signed = (c) => (c === null ? "n/c" : `${c > 0 ? "+" : ""}${(c * 100).toFixed(0)}%`);
const todo = [];
for (const m of metrics) {
const [bf, normal, lastYear] = ["blackFriday", "normal", "lastYear"].map((k) => m.calc(periods[k]));
const vsNormal = change(bf, normal);
// Shopify changed how it measures sessions in September 2026: no session-based
// comparison with last year.
const vsLastYear = m.sessionBased ? null : change(bf, lastYear);
const flagged = worse(m, vsNormal) || (vsLastYear !== null && worse(m, vsLastYear));
console.log(
`${flagged ? "!" : " "} ${m.name.padEnd(30)} ${m.fmt(bf).padStart(8)}` +
` vs normal ${signed(vsNormal).padStart(5)} vs last year ${signed(vsLastYear).padStart(5)}`,
);
if (flagged) todo.push(m.name);
}
console.log(`\nDecember list: ${todo.join(", ") || "nothing above the threshold"}`);
periodsholds the raw counts for the three periods, as they come out of the reports. It stores counts, not rates, so the formulas are identical across all three.THRESHOLDis the 10% tolerance. Raise it if your store has low volume and the numbers jump around on their own.ratioreturnsNaNwhen the divisor is zero instead of dividing by zero.metricsdefines each metric with its formula (calc), whether it should go up or down (better) and whether it depends on sessions (sessionBased). Conversion is generated once per device from the keys ofsessions.changeis the relative change;worsedecides whether it is bad given the metric's direction.- For session-based metrics,
vsLastYearstaysnulland printsn/c(not comparable), because of the September 2026 measurement change. - A metric gets a
!if it came out worse against the normal weekend or against last year, and its name goes on the list.
The output with the illustrative numbers:
! conversion mobile 1.9% vs normal -13% vs last year n/c
conversion desktop 3.9% vs normal +15% vs last year n/c
average order value 982.52 vs normal +14% vs last year +1%
checkout completion 36.8% vs normal -8% vs last year n/c
! zero-result searches 14.0% vs normal +73% vs last year +29%
! failed payments / 100 orders 12.9 vs normal +131% vs last year +80%
December list: conversion mobile, zero-result searches, failed payments / 100 orders
Look at checkout completion: it fell 8%, under the threshold, so it stays off the list, but it is worth a look in the GA4 funnel. Each flagged line becomes a task with an owner, a date and the metric that will say whether it worked. Three or four tasks done well before the December shopping weeks beat twenty written down. If your store is still in a code freeze, each change goes through the exception process.
Scroll the figure sideways to see all of it
Common mistakes
- Comparing with the week before Black Friday, which already had early deals.
- Comparing date with date against 2025 instead of Friday with Friday.
- Comparing 2026 conversion with 2025 without accounting for September's session measurement change.
- Mixing GA4 users with Shopify sessions in the same formula.
- Reading AOV without looking at how much the discounts weighed.
- Looking for failed payments in orders: they rarely create one; they live in abandoned checkouts.
- Checking searches on Saturday, with data that can lag up to 72 hours.
- Ending with a list of twenty items with no owner or date.
Frequently asked questions
Why a normal weekend and not the week before Black Friday?
Because the week before is usually distorted: there are early deals and shoppers holding off until Friday. A weekend without promotions tells you how the store performs under normal conditions, and against that you see what Black Friday changed.
Can I compare conversion rate with Black Friday 2025?
Very carefully. Shopify changed its session measurement between September 21 and 23, 2026, and warns that session-based metrics can move without any change in traffic. Against 2025, compare orders, AOV, searches and failed payments; compare conversion against a normal weekend after the change.
Why don't GA4 and Shopify numbers match?
Because they measure different things: GA4 counts users and events sent from the browser, and Shopify counts sessions and orders on its server. Ad blockers, cookie consent and offsite payments widen the gap. Use each one for its own comparison and do not combine their numbers.
What if GA4 isn't set up for checkout?
Run the analysis with Shopify, which already covers all five metrics; the only thing missing is which step inside checkout people leave on, and that goes on the December list. If you set up the checkout events now, next Black Friday you will have a normal weekend and a previous year to compare with.
Checklist
- The three periods are defined: Black Friday, a normal weekend without promotions and the same weekend in 2025.
- Conversion by device is compared only against the normal weekend.
- AOV is read together with discounts.
- The GA4 funnel uses events that actually arrive, with Make open funnel off.
- Zero-result searches are checked once the data is complete.
- Failed payments come from abandoned checkouts, grouped by reason.
- The script runs with your numbers and the list has three or four tasks with an owner and a date.
Monday's analysis is half the job; the other half is measuring again in December with the same formulas, to know whether the fixes worked.
Saw this on Instagram? Every Reel guide is at ecomlabs.dev/guides.
Written by

Published November 19, 2026 · updated November 20, 2026