How to adjust year-over-year retail comps for a 53rd week

Every five or six years, a retailer's fiscal year runs 53 weeks instead of 52, and the comp table you've built for the last four quarters suddenly doesn't line up. The quarter looks great. Or it looks terrible. Either way, the headline number is lying to you a little, and you need to know how much before you put it in a note.
Why the extra week shows up at all
Most retail chains close their books on the NRF 4-4-5 calendar, which fixes the fiscal year-end to the Saturday closest to January 31 rather than to a fixed date. Because 52 weeks is only 364 days, the calendar drifts against the actual solar year by about a day annually, and every five or six years that drift adds up to a full extra week. Companies tack it onto the fourth quarter almost always, since that's the easiest place to absorb it without disrupting the holiday selling season reporting structure. So you get a 14-week Q4 instead of 13, and a 53-week fiscal year instead of 52.
If a chain under your coverage ran a 53-week year two years ago, this year's comp is being measured against an inflated base, and next year's will be too, depending on when the next extra week lands. The distortion recurs on roughly a five- to six-year cycle, so rebuild the adjustment every time the calendar shifts rather than treating it as a one-off fix.
The comp you get if you don't adjust
If you take reported Q4 revenue in the 53-week year and divide it by reported Q4 revenue in the prior 52-week year, you're comparing 14 weeks of sales to 13. That extra week usually falls in early February, picking up post-holiday clearance volume that the prior-year quarter never had. The naive year-over-year number comes in hot, and if you don't catch it, you'll flag a comp acceleration that's really just an extra week of cash register tape.
The year after is the mirror problem. Now you're comparing a normal 52-week year against a 53-week prior year, and the naive comp looks soft even if underlying demand is fine, because the denominator is carrying one extra week of sales that the numerator doesn't have a match for.
Three ways to adjust the comp
Calendar-align the weeks. Pull the 53-week year down to the same 13-week (or 52-week, for the full year) window as the prior year, matching by week number or by proximity to a fixed anchor like the Saturday nearest a given date, rather than by the raw reported period. This is what the NRF calendar itself is built to support, and it's the cleanest method if you have weekly or store-level data to work with.
Use the company's own footnote. Most chains that report a 53-week year will disclose a "comparable 52-week" figure somewhere in the 10-K or the earnings deck, specifically because they know analysts will otherwise misread the headline. Check the footnotes before you build your own adjustment from scratch; sometimes the work is already done for you.
Run-rate it. When no company disclosure exists and you don't have weekly granularity, divide the 53-week figure by 53 to get an average weekly rate, then multiply by 52. It's a blunt instrument since it assumes sales are spread evenly across the quarter, which they're not during a holiday period. Use it when nothing else is available. It still gets a usable comp into your model.
Whichever method you use, note it in your model assumptions. If a 53rd week shows up in your coverage universe this cycle, odds are good a sell-side peer is going to publish the naive number anyway, and you want your own comp framework written down somewhere you can point to.
Watching the trend without waiting on the calendar
Fiscal calendar shifts are one more reason not to lean too hard on a single quarterly comp print. If you're tracking parking lot traffic week over week through the quarter, you can see whether the extra week's volume looks like genuine incremental demand or just a shift in timing, well before the earnings release forces you to reconcile it. That's part of what the Sales Nowcast traffic series is built to give you: a daily count that runs on its own calendar, not the retailer's. Worth a look before the next print.