Sales Nowcast

What a two-year stack in same-store sales actually tells you

A 2025 wall calendar on a desk next to a cup of coffee and a small potted plant.

A single-year comp number is easy to misread. If a retailer posted negative 15% comps in 2020 and then plus 18% in 2021, the headline on that second print looks great. Stack the two years together and you're still below where the chain started. That's the whole reason stacking exists.

The basic math

A two-year stack is just addition. You take this year's same-store sales growth and add it to last year's, treating both as if they were simple percentages rather than compounding off different bases. So if comps were -15% last year and +18% this year, the two-year stack is +3%. Some shops will compound it properly instead of adding (multiply 0.85 by 1.18 to get a true two-year index of roughly 1.003, or +0.3%), and the two methods drift apart more the bigger the swings get. Most sell-side models use the simple addition because it's fast and directionally close enough, but check which convention a management team is using before you compare their stack to a peer's.

The point of doing this at all is to strip out the base-period distortion. One year's comp is only meaningful relative to what that same quarter looked like twelve months earlier. If the prior-year quarter was unusually weak (a closure, a weather event, a pandemic lockdown), this year's comp will look artificially strong even if traffic hasn't actually recovered. Stacking two years together gives you a read against a base that's further back and usually cleaner.

Why management loves to quote it, selectively

Here's where it gets a little annoying as an analyst. A chain will happily cite its two-year stack on a quarter when the stack flatters them and go quiet on it when it doesn't. If you're modeling a retailer that had a rough prior-year compare, expect the earnings deck to lead with the stacked number. If the prior year was a gift, expect the single-year comp front and center instead. Pull both years' single comps yourself and do the stack independently so you're not grading on whatever curve the company picked.

When you need a three-year stack instead

Two years isn't always enough of a lookback. Retailers that got hit hard in 2020, recovered unevenly through 2021 and 2022, then settled into a more normal cadence by 2023 are a good case for a three-year stack comps analysis instead of a two-year one. Add three consecutive years of comp growth together and you're measuring against a base from before the disruption started at all, which filters out more of the noise from any single weird year in between. The tradeoff is that a three-year stack is further removed from current trend. It tells you where the business sits relative to 2019, not what's happening in the store right now.

What stacking can't fix

A stacked comp is built entirely from quarters that already closed and reported. The quarter you actually care about right now, the one still in progress, isn't in that math at all. And because same-store sales bundle foot traffic and average ticket into one number, a decent-looking stack can hide a real drop in shopper count if ticket size is propping up the total. A chain can post a flattering stack on fewer shoppers spending more per visit, which is a very different story for the model than more shoppers walking in the door. If you want a read on the traffic side specifically, ahead of the print rather than after it, a daily parking-lot count across the chain's locations gets you there without waiting on the quarterly release or management's framing of it.

Run the stack both ways (simple and compounded), check it against the prior-year base the company is implicitly using, and don't let a flattering two-year number stand in for an actual traffic read.

If you'd rather see the traffic trend building before the stack gets published, that's what Sales Nowcast is for.

Get started

← Back to the blog