Same-Store Sales Decomposition in a KPI Tree

Total Sales Grew 4.8 Percent and Comparable Sales Grew 0.6 Percent

Same-store sales decomposition splits a retailer's sales growth into the part that came from stores it already had and the part that came from changing the store base. The two numbers move apart routinely, and the gap is where most retail reviews lose the thread.

The Home Depot reported first quarter fiscal 2026 sales of 41.8 billion dollars, up 4.8 percent, while comparable sales rose 0.6 percent, read on August 12, 2026.¹ Both figures are correct. They describe different populations of stores.

Read one line further and the comp number moves again. Foreign exchange added roughly 55 basis points to total company comparable sales in the same quarter.¹ A 0.6 percent comp carrying 55 basis points of currency is close to flat once currency is stripped out.

None of this is unusual. It is what a comparable sales number looks like when nobody decomposes it.

What Is Same-Store Sales Decomposition?

Same-store sales decomposition splits total sales growth into four additive contributions: comparable stores, non-comparable stores still trading, newly opened stores, and closed stores. The four sum exactly to the reported change in total sales. Below the comparable line, comp growth splits again into a transaction effect and an average ticket effect.

The metric has three common names and one shared purpose. United States retailers say comparable sales or comps. United Kingdom and European retailers say like-for-like. Corporate Finance Institute defines same-store sales as growth in revenue from locations open at least one year, which isolates organic performance from store count changes.⁴

The purpose is to answer a question total sales cannot: is the estate we already own getting better or worse.

The metric is not defined by any accounting standard. In the United States it is a non-GAAP measure disclosed under company-specific rules. In Europe it is an alternative performance measure, which ACCA describes as a disclosure that lets readers compare retailers pursuing different expansion strategies.⁵

That freedom is the whole problem. Two retailers with identical trading can report different comps because they wrote different rules. The decomposition below is what makes those rules visible instead of assumed.

The Comp Base Is a Set, Not a Filter

The hardest part of a comp sales tree sits above the arithmetic, not below it. It is deciding which stores belong.

Treat the comp base as an explicit set with a membership rule, evaluated once per period and stored as a flag on every store row. A filter applied at query time drifts. A stored flag can be audited.

Best Buy's rule is a useful reference because it is written down. Comparable sales covers stores, websites and call centers operating for at least 14 full months. Revenue from stores closed more than 14 days, including relocated, remodeled, expanded and downsized stores, is excluded until at least 14 full months after reopening.²

Other retailers set the bar lower. Destination XL Group has disclosed a 13 month threshold for comparable stores.⁷ One month of difference moves which cohort of immature stores lands inside the base.

Write the rule into the metric definition and freeze it for the year. Changing membership mid-year makes every prior comp in the series unusable, the same failure mode as the twelve pitfalls that break a KPI tree after month two.

The Bridge From Total Sales to Comparable Sales

A hundred store chain. Prior year total sales 500.0 million dollars. Current year total sales 521.0 million dollars, up 4.2 percent.

Split the current year into four populations and value each against the same population in the prior year.

Comparable stores, 90 of them: 450.0 million prior, 463.05 million current. Contribution plus 13.05 million. Comp growth 2.9 percent.

Non-comparable stores still trading, 6 remodeled during the year: 32.0 million prior, 34.95 million current. Contribution plus 2.95 million.

New stores, 7 opened this year: zero prior, 21.0 million current. Contribution plus 21.0 million.

Closed stores, 4 shut during the year: 18.0 million prior, 2.0 million current. Contribution minus 16.0 million.

Add the four. 13.05 plus 2.95 plus 21.0 minus 16.0 equals 21.0 million. That is the reported change in total sales, to the dollar.

The closure drag of 16.0 million is the term that goes missing most often, because closed stores leave the current period file entirely and a naive join drops them.

The four contributions must sum to current total sales minus prior total sales. If they do not, a store is either absent from every population or counted in two. Both errors are common, and both are silent.

Why Does a 2.9 Percent Comp Add Only 2.6 Points to Total Growth?

Because the two percentages have different denominators. Comparable growth of 2.9 percent is measured against 450.0 million dollars of comp base sales, not the 500.0 million dollar company total. Multiply 2.9 percent by the comp base share of 90 percent and the result is 2.61 points of total growth. The remaining 1.59 points come from openings and closures.

Restate the four contributions as points of prior year total sales and the arithmetic closes.

Comparable stores contribute 13.05 divided by 500.0, which is 2.61 points. Remodeled stores contribute 0.59 points. New stores contribute 4.20 points. Closures subtract 3.20 points. The four sum to 4.20 points, which is the reported total growth rate.

This is the single most common reading error in a retail review. A comp percentage and a total percentage are placed side by side and the difference is described as store growth, when part of it is simply the comp base weight.

Carry the weight explicitly. A chain where comp stores are 90 percent of prior year sales passes through 90 percent of any comp movement. A chain in heavy expansion, where comp stores are 65 percent of prior year sales, passes through far less, and its comp number tells you correspondingly less about the total.

Store the comp base share as its own node. It changes every year, and it changes what the comp number means.

Below the Comp Line: Transactions and Average Ticket

Inside the comp base, growth has two proximate causes: how many baskets, and how large each basket was.

Comp transactions: 30.0 million prior, 29.4 million current, down 2.0 percent. Comp average ticket: 15.00 dollars prior, 15.75 dollars current, up 5.0 percent. Average ticket is comp sales divided by comp transactions at every node. It is never a stored column, because averages do not add.

The additive split has three terms. Transaction effect is 29.4 minus 30.0, times 15.00, which is minus 9.0 million. Ticket effect is 15.75 minus 15.00, times 30.0, which is plus 22.5 million. Interaction is minus 0.6 times 0.75, which is minus 0.45 million. The three sum to 13.05 million.

Most retailers report two terms, not three. Valuing the ticket effect at current transactions absorbs the interaction: 15.75 minus 15.00, times 29.4, is 22.05 million, and minus 9.0 plus 22.05 is still 13.05 million.

Pick one convention and hold it. Going further, and splitting ticket into units per basket and price per unit, is the three-term price, volume and mix bridge applied to the comp base.

Five Comp Base Design Choices That Change the Number

Every retailer makes the same five decisions. Each one moves the reported comp without any change in trading, which is why the Retail Think Tank has questioned how much a like-for-like figure still tells an outside reader.⁶

Make the five explicit in the metric definition, and the comp becomes readable by someone who did not build it. Leave them implicit and the number is an opinion with a decimal point.

The table states each choice, the settings in common use, what the choice changes, and the risk it creates.

Calendar Effects That Corrupt a Comp Number

A comp compares two periods. If the periods are not the same shape, the comparison is wrong before any store math starts.

Most large United States retailers use the National Retail Federation 4-5-4 calendar, which divides the year into four, five and four week months so that comparable months hold the same number of Saturdays and Sundays.³ That alignment exists precisely so like days are compared to like days.

The 4-5-4 year runs 52 weeks, which leaves a spare day each year. Every five or six years a 53rd week is added, most recently in fiscal 2012, 2017 and 2023, and the NRF restates the 53 week year for comparability against the following year.³

Three corrections follow. Compare 52 weeks against 52 weeks, never 53 against 52. Restate the prior year rather than adjusting the current one, so the series stays stable. And check holiday placement, because a shift of Easter or a New Year week between quarters moves a comp with no change in demand.

Record the calendar convention next to the comp base rule. They fail together.

How Do You Test a Same-Store Sales Tree for Correctness?

Run four checks. The four population contributions must sum to the reported change in total sales. Every store must appear in exactly one population per period. Average ticket must be derived by dividing two summable columns rather than stored. And both periods must cover the same number of weeks under the same calendar.

Test 1. Reconciliation. Sum comparable, non-comparable trading, new and closed contributions and compare to current total sales minus prior total sales. Any gap means the bridge is broken, not approximately right.

Test 2. Exhaustive and exclusive membership. Count stores in each population and confirm the counts add to the full estate. A store that traded in either period and appears in none of the four is a hole. A store in two is double counted.

Test 3. Derived ratios only. Sales and transactions are summable and belong in the data. Average ticket is computed by division at whatever level of the tree you are reading. A stored ticket column silently produces the average of averages, which does not equal the true ticket.

Test 4. Equal periods. Confirm week counts, day counts and holiday placement match across the two periods before reading anything.

Fail any one of the four and the tree returns a number that looks defensible and is not.

Where a Same-Store Sales Tree Is the Wrong Tool

A comp tree explains a multi-unit retailer's growth. It is a poor instrument in four situations.

Pure digital retailers. With no store estate, the comp base is the whole business and every contribution collapses into one term. Traffic, conversion and average order value are the useful decomposition instead.

Franchise systems. System-wide sales, franchisee sales and franchisor revenue are three different numbers. A comp built on system-wide sales does not bridge to the franchisor's own revenue, and mixing them produces a tree that reconciles to nothing.

High expansion phases. When new stores are 30 percent or more of the estate, the comp base is a shrinking minority of the business and the comp number stops describing the company. Report it, but do not lead with it.

Attribution. The tree says closures cost 16.0 million dollars and ticket added 22.5 million. It does not say why ticket rose. Assortment shift, list price change and promotional depth each need their own decomposition, which is repeatable root cause analysis on the branch that moved.

Naming the limits is part of shipping the method.

Building the Comp Sales Tree in kpitree.io

kpitree.io is a self-service KPI tree builder for finance, business and product analysts. This is the only section of the article about the product.

The mechanics match the four tests. You upload a CSV of store-period rows carrying store identifier, period, sales, transactions and a population flag. Inside the tree, derived KPIs are computed by dividing two summable columns, so average ticket is sales divided by transactions at every node rather than a stored value. That is what Test 3 requires.

Node identities are addition and subtraction. Comparable, non-comparable trading, new and closed stores are therefore sibling nodes summing to total sales, and the reconciliation in Test 1 becomes a property of the structure rather than a manual spreadsheet check.

Two honest limits. The evidenced ingest path today is CSV upload, so a comp bridge is refreshed by uploading a new file rather than through a live connection. And capabilities the site marks as coming soon are not available.

The same additive discipline appears in promotion effectiveness in CPG, viewed from the manufacturer side of the same shelf.

Frequently Asked Questions

Is same-store sales the same as like-for-like sales?

They name the same idea in different markets. Same-store sales and comps are the United States terms, like-for-like is the United Kingdom and European term. Neither is defined by an accounting standard, and both are disclosed under company-specific rules.⁴ ⁵

Should e-commerce orders sit inside the comp base?

Decide by fulfillment, not by channel. Orders picked, shipped or collected at a comp store belong in that store's comp. Orders fulfilled by a distribution center belong in a separate digital population. Best Buy includes websites and call centers in comparable sales, which is one defensible answer among several.²

How long should a store wait before entering the comp base?

Thirteen to fourteen months is the common range, and the exact figure matters less than freezing it. A store admitted at 13 months carries one more month of opening ramp than a store admitted at 14.² ⁷

Can comparable sales fall while total sales rise?

Yes, and for an expanding chain it is the normal state. New store contribution can exceed a negative comp contribution, which is exactly the case the four-term bridge is built to show.

Why does driver-level decomposition stay rare?

It is not a retail problem. FP&A Trends reported in its 2025 benchmarks, read on August 12, 2026, that only 17 percent of organizations use fully driver-based models.⁸

Closing: Rebuild Last Quarter's Comp Bridge Before the Next Review

The next time a review compares a comp percentage to a total percentage and calls the difference store growth, the fastest way to settle it is a bridge that reconciles.

One CSV is enough. Store identifier, period, sales, transactions, and a flag marking each store as comparable, non-comparable trading, new or closed. Two periods of equal length under the same calendar.

Sum the four populations, check they equal the reported change in total sales, then divide comp sales by comp transactions to get average ticket. If the four sum, you have a defensible answer in an hour. If they do not, you have found a store that is missing or double counted, which is worth more than the answer would have been.

Start with one region and one quarter. Upload that single CSV and decompose the comp number your team already argues about.

kpitree.io is a self-service KPI tree builder for finance, business and product analysts.

Sources

• The Home Depot Announces First Quarter Fiscal 2026 Results, Exhibit 99.1 to Form 8-K
• Best Buy Co., Inc. Form 10-Q, comparable sales definition
• 4-5-4 Calendar
• Same-Store Sales: Definition, Importance, Formula
• Additional performance measures
• Are like-for-like sales figures still a useful and relevant measure of retail performance?
• Destination XL Group, Inc. Form 10-K, comparable sales definition
• 2025 FP&A Benchmarks: Where We Are, Where Leaders Are Going