Period comparison is the measurement of a metric in one time window against the same metric in an earlier window of the same length, expressed as a difference or a percentage change. Week over week, month over month, quarter over quarter, and year over year are the standard windows; a custom period is any two windows a seller defines, such as the 30 days before and after a price change.
A comparison is only as honest as the alignment between its two windows. Four things break alignment on marketplace data.
Month length. February has 28 days, January has 31. A seller with a flat daily run rate reports revenue down 9.7 percent in February and has changed nothing. Compare daily averages, or use a trailing 28-day window.
Weekday composition and events. A month with five Saturdays sells differently from one with four. Amazon's mid-year sale event, Black Friday, and category holidays move revenue into specific days, and a window that contains one of those days compared to a window that does not is an event measured against a baseline.
Settlement lag. Amazon settles roughly every two weeks on its own schedule; Walmart's Marketplace Learn site describes payment cycles set by the seller's agreement; TikTok Shop settles after delivery. A month built from deposits contains the tail of the prior month and omits the tail of its own. Comparisons built on order date, with the open settlement accrued, remove this.
Structural change in the business. YoY compares a catalog of 240 SKUs to one of 180, and this year's fee schedule to last year's. The season aligns; the business does not. Decompose a YoY move into what the same SKUs did and what new or retired SKUs added or removed.
Illustrative numbers. Take an Amazon seller of home goods with June revenue of $84,000 and July revenue of $96,000. Month over month, that is up 14.3 percent, and the seller reads it as growth.
Three corrections in sequence.
First, month length. July has 31 days to June's 30, so on daily averages the comparison is $3,097 a day against $2,800, up 10.6 percent. Still growth.
Second, events. Amazon's mid-year sale event ran for two days in July and produced $14,500. The other 29 days produced $81,500, or $2,810 a day, against June's $2,800. Up 0.4 percent, which is flat.
Third, the year-over-year check. July of the prior year was $98,500 with the same event, which contributed $16,200 that year. This July is down 2.5 percent in total and, on non-event days, down about 1 percent against the prior year's $82,300 over 29 days.
Month over month said plus 14 percent. Year over year, with the event isolated, said minus 2.5 percent on an event that itself shrank by $1,700. The second reading is the one to act on, and the action is a look at which SKUs lost ground against last July, not a celebration.
The reverse error is as common: August compared to an event-laden July reads as a decline when the daily rate held.
The comparison report in ConnectBooks puts two periods side by side on reconciled, order-date data, by channel and by SKU, so settlement lag is removed before the comparison starts. Crunch, the analytics AI inside ConnectBooks, compares periods as the second step in its chain on every question: week over week, month over month, quarter over quarter, or a custom range. Ask whether a product had better margins in any month this year than it has now and it selects the windows, compares them, and names the cost lines that differ. It does not know a given week held a sale event unless the data shows the spike, and if you ask for month over month when the honest answer is year over year, it answers month over month.
Yes, when both months are free of events and the comparison is on daily averages. It is the wrong comparison across an event month, across February, or when the question is about seasonality.
Usually because one is on order date and the other is on settlement or deposit date. Amazon's sales dashboard runs on order date. A ledger that books deposits as revenue runs on settlement date. Pick one basis, order date, and build both views on it.
Two to three on each side of the change, at minimum, with event weeks dropped. One week each side confuses the change with ordinary variance.
Neither YoY nor QoQ works, because the earlier window is empty. Use WoW from launch, or a custom window against a comparable SKU's first weeks.
It runs the comparison you ask for and can run several. It does not know which one your question needs unless you say, and it does not know a spike was an event unless the data or you tell it.
Running an e-commerce business comes with plenty of challenges, but ConnectBooks is here to make your life easier. With real-time insights, seamless integrations, and detailed tracking of your profitability and inventory, you can stay ahead of the game. Whether you’re selling on Amazon, Shopify, Walmart, TikTok or eBay, ConnectBooks helps you manage your finances with 100% accuracy and confidence, so you can focus on growing your business.
Ready to level up? Start making smarter, data-driven decisions every step of the way. Try ConnectBooks Free Today or Schedule a Demo
