Dimension Labs / Causal Brief
The Home Depot · Atlanta Metro

Dimension Labs · Causal Brief

Home Depot

Atlanta metro · 24-store fleet review · Jan 2025 – Jun 2026

The one thing to know

Two Home Depots, same city, same week. One turns away 1 shopper in 7. The other, 1 in 3.

The difference isn’t the economy — every store here shares one city, one housing market, one brand. It’s whether anyone helps you on the floor. No survey asked. Two thousand strangers told us anyway. Here is the whole chain, in one figure: from raw voice, to proven cause, to money.

The Causal SpineRaw voice → signal → cause → measured effect → the stake
VOICESIGNALDRIVEREFFECTSTAKE 2,354 shoppersspoke.One complaint risesabove all the rest.The cause ison the floor.And it isnot a coincidence.The one leverthey still control. 2,354597 16.5× 597 of 2,354one in four voices carries it Every public Google review,read and structured.24 stores · one Atlanta metro.“No one wouldhelp me on the floor.”the most common complaint, by far.Too few peopleon the floor to help.Thin staffing coverage.+0.3%same-store sales growth.Growth has stalled.~$24Bstaked on Pro growth — itonly pays back if the floor delivers. A “no one helped” review is 16.5× more likelyto be a 1–2 star review. n = 597 · p < 0.001THE CONTROL · SAME MARKET14.4%32.9%best storeworst storeAn 18-point spread the economy can’t explain.Share of unhappy shoppers, across all 24 stores.

Every public Google review for these 24 same-market stores was read and structured — 2,354 in all — then each recurring complaint was tested for how strongly it predicts an unhappy review and how often it actually occurs. Because all 24 share one city, one economy, and one brand, the wide store-to-store gap can’t be the market. Financial figures are Home Depot’s own published results · Dimension Labs analysis

The place is full of folks up front but no one on the floor — no help to be found, even when you ask.

1-star review · 4325 New Snapfinger Woods Dr, Decatur · the store the analysis flags to visit first

Home Depot’s growth has stalled. In its most recent fiscal year, same-store sales (sales at stores open at least a year) rose just +0.3% even as the number of transactions fell −1.0% and profit margin slipped to 12.7% from 13.5%. When the market isn’t handing you growth, the in-store experience is no longer a nice-to-have — it’s the growth itself.

And the market is the same for every store here. All 24 share one city, one set of interest rates, one housing cycle, one brand, and one corporate playbook — so the gap between a five-star store and a one-star store can’t be blamed on the economy. It’s about how each store is run.

The thing that moves the needle is having enough staff to help shoppers. Reviews that mention not getting help are 16.5× more likely to be 1–2 star than reviews with no complaint, and this issue drags down the whole fleet more than any other. That backs up Home Depot’s standing ~$1B commitment to associate pay.

The biggest thing to fix

Not enough staff to help

“Couldn’t find anyone to help,” long waits, and rude or unhelpful staff show up in 597 reviews and drag the fleet down 4.3× more than the next-biggest issue, online orders and pickups falling through.

The biggest thing shoppers love

A great employee by name

67.6% of the praise in happy reviews is for a helpful — often named — employee. The best store in the fleet stands out for its people, not its inventory.

1 in 3
leaves the worst store an unhappy review — versus 1 in 7 at the best-run twin.
32.9% vs 14.4% · same city
18 pts
the gap between best and worst store, with the whole market held constant.
across all 24 stores
16.5×
more likely a review is 1–2 star when it says no one helped on the floor.
n = 597 · p < 0.001

01 — These aren’t random reviews — the stores really are different.

Online reviews tend to run either glowing or furious. The point isn’t that they’re loud — it’s that the loudness doesn’t land evenly. The share of unhappy 1–2 star reviews runs from 14.7% at the best store to 32.8% at the worst — too wide a gap to wave off as noise.

What we wanted to know. Do these stores actually split into good and bad experiences, or is it just a noisy pile of 5 star and 1 star reviews? We use each review’s star rating as the score, and group reviews by the kind of problem or praise each one describes. Full method in Methodology & Sources.

Atlanta shoppers aren’t describing one Home Depot — they’re describing two. At one end, 2525 Piedmont Rd NE turns 80.1% of its reviews into happy 5-star ones and keeps unhappy reviews down to 14.7%. At the other, 1480 Satellite Blvd lets unhappy reviews climb to 32.8%. Even if you set that store aside because it has just under 70 reviews (our confidence cutoff), the next-worst stores we’re fully confident in — 4325 New Snapfinger Woods Dr, 1032 Research Center Atlanta Dr, and 650 Ponce De Leon Ave NE — still sit near a 30% unhappy rate.

A weak market can pull every store down a notch. It can’t explain why one store turns away two shoppers in seven while another turns away just one.

This gap matters precisely because the market is the same for all of them. High interest rates, postponed remodels, and an estimated ~$22B in skipped home-improvement spending hit every Atlanta store equally. So anything that’s the same fleet-wide simply can’t be what’s making one store better than another.

The mix of review types says the same thing. The biggest clearly negative group is “couldn’t get help on the floor” at 14.7% of all reviews — well ahead of online order/pickup failures at 5.5%, price complaints at 1.7%, returns problems at 1.5%, and installation problems at 0.6%. On the happy side, “a great employee helped me” alone is 22.8% of all reviews. The story is about people before it’s about anything else.

Exhibit 1.1How shoppers rate these stores (star breakdown)

Most reviews are either 5-star or 1-star — but the gap between which store you walk into is the thing worth acting on, not the shape of the chart.

1 star4812 star1233 star1804 star1235 star1,447

Source: 2,354 Atlanta-metro Google reviews, Jan 2025 – Jun 2026 · Dimension Labs analysis

Exhibit 1.2What unhappy and happy reviews are really about

Read each side on its own. The biggest reason shoppers are unhappy is not being able to get help on the floor; the biggest reason they are happy is a great employee. The same thing — people — sits at the top of both.

WHY SHOPPERS WERE UNHAPPYCouldn’t get help on floor14.7%Online order / pickup5.5%Price / checkout1.7%Returns & refunds1.5%Installation service0.6%WHY SHOPPERS WERE HAPPYA great employee helped22.8%

Source: 2,354 Atlanta-metro Google reviews, Jan 2025 – Jun 2026 · share of all reviews by main reason · Dimension Labs analysis

02 — Not enough staff on the floor is the number-one problem — and the data says it plainly.

The finding isn’t that “can’t get help” hurts most when it happens — installation and returns sting more in the moment. It’s that “can’t get help” combines real pain with happening far more often, making it the single biggest fixable drag on the whole fleet, by a wide margin.

The test. We compared five common problems on how much each raises the chance of a 1–2 star review, then weighted that by how often each actually happens. A “bad outcome” is a 1–2 star rating; the comparison point is reviews that mention none of these problems. Full method in Methodology & Sources.

A review that mentions not getting help is 16.5× more likely to be 1–2 star than one with no complaint at all (597 reviews; highly significant, p<0.001). Online-order and installation problems are about as painful — or worse — when they happen, but none come close on overall impact because they happen far less often. Combining how often it happens with how badly it hurts, “can’t get help” scores 9,848.9, versus 2,277.7 for online order/pickup, 1,096.7 for returns, 548.3 for installation, and 358.5 for pricing.

That ranking is the whole report in one chart. Price complaints are loud, but they’re not what separates good stores from bad. Installation breakdowns are brutal, but too rare to explain the fleet-wide gap. “Can’t get help” wins because it’s both common and strongly tied to a bad review.

He was rude and made me feel like I was stupid for wanting to replace my sink. No one offered to help — I left and drove to Lowe&rsquo;s.

1-star review · 5300 Windward Pkwy, Alpharetta · one of the six stores to visit first

This holds up no matter how you slice it. Among the few reviews that say who the shopper is, “can’t get help” reviews are 1–2 star 76.9% of the time for DIY homeowners and 90.5% for Pro/contractor shoppers. Across large, mid, and small stores, that rate stays tightly between 76.4% and 80.1%. This doesn’t prove anything about actual staffing levels — reviews can’t see payroll — but what shoppers experience stays bad regardless of store size.

The same thing that drives the worst experiences — staff on the floor — also drives the best ones.

The happy reviews close the case. Shoppers praise a specific employee by name 577 times, versus just 97 calling out a bad one. Happy shoppers mostly aren’t praising big selection or low prices. They’re praising a person.

9,849
impact of “can’t get help” — how often × how badly it hurts.
over 4× the next problem
76–91%
turn negative, across every shopper type and store size.
robust to every cut
577 : 97
shoppers name a great employee far more than a bad one.
praise outweighs blame ~6:1
Exhibit 2.1Which problem drags the whole fleet down the most

This combines how often each problem happens with how badly it hurts the rating. “Can’t get help on the floor” wins easily — it’s both the most common and one of the most damaging.

Can’t get help on the floor9,849n=597 16.5×p<.001Online order / pickup fails2,278n=136 16.7×p<.001Returns & refund problems1,097n=57 19.2×p<.001Installation problems548n=27 20.3×p<.001Pricing & checkout359n=40 9.0×p<.001

Source: 2,354 Atlanta-metro Google reviews, Jan 2025 – Jun 2026 · how often it happens × how badly it hurts; significance test on each · Dimension Labs analysis

Exhibit 2.2The “can’t get help” problem holds up no matter how you slice it

Whether the shopper is a DIY homeowner, a Pro/contractor, or unspecified — and whether the store is large, mid, or small — reviews that mention not getting help stay overwhelmingly negative.

60%70%80%90%100%FLEET AVERAGE: BAD REVIEW WHEN HELP IS HARD TO FINDDIY homeowner76.9%Pro / contractor90.5%Not stated77.8%Large stores76.4%Mid-size stores80.1%Small stores78.2%

Source: 2,354 Atlanta-metro Google reviews · split by shopper type and store size · small groups (fewer than 25 reviews) read with caution · Dimension Labs analysis

Exhibit 2.3Shoppers name a great employee far more than a bad one

The same thing — staff on the floor — drives both the best and worst stories. And shoppers praise a named employee almost six times as often as they call one out.

Praised a named employee577Named a bad employee97

Source: 2,354 Atlanta-metro Google reviews · reviews naming a specific employee · Dimension Labs analysis

A wall of voices — getting help on the floor

demasiados trabajadores y les pides ayuda y no te quieren atender

1-star · 3755 Shackleford Rd, Duluth

if you need help don’t expect much from this place

1-star · 2450 Cumberland Pkwy SE, Atlanta

Angel approached it with dedication and integrity, embodying what great customer service should be

5-star · 2450 Cumberland Pkwy SE, Atlanta

Milton was phenomenal! Amazing customer service from start to finish.

5-star · 2525 Piedmont Rd NE, Atlanta

03 — The visit list starts with six stores — and every one has a staffing problem first.

The store ranking turns the finding into a plan: visit the stores with the most unhappy shoppers first, and understand each one through its biggest problem. In this fleet, that problem is “can’t get help on the floor” almost everywhere it counts.

Per store. One row per review. Each store is scored on its share of unhappy 1–2 star reviews and on the share of its reviews that mention not getting help. Confidence cutoff: stores under ~70 reviews are flagged “read with caution.” Full method in Methodology & Sources.

The first visit should be 4325 New Snapfinger Woods Dr. It’s not the single worst store on paper, but it’s the worst among stores with enough reviews to trust (over 70): 31.5% unhappy reviews out of 127, with “can’t get help” showing up in 33.1% of all its reviews. It’s the clearest, most reliable case of staffing breaking the experience.

Next come 1032 Research Center Atlanta Dr, 650 Ponce De Leon Ave NE, and 5300 Windward Pkwy. All three sit near or above 29% unhappy reviews, all have lots of “can’t get help” reviews, and all pile on a second problem — especially online order/pickup failures at Ponce and Windward. The main fix is the floor; the second is getting online orders and pickup right.

The proof is in comparing the top and bottom stores. The worst stores have “can’t get help” in about 29.5% of reviews; the best stores, about 20.2% — roughly a 9.3-point gap within one city, one brand, one way of operating. The difference between stores is not subtle.

The store to copy is 2525 Piedmont Rd NE. It has the fleet’s lowest share of unhappy reviews at 14.7%, the highest share of happy ones at 80.1%, and the most employees praised by name, led by repeated mentions of Milton. For district leadership, Piedmont isn’t just a “good store” — it’s the best in-house example of what great service looks like when it works.

Exhibit 3.1All 24 stores ranked by share of unhappy shoppers

Every store, none hidden. Red is the worst end, green the best, the dashed line is the 25.7% average, and stores with fewer than ~70 reviews are flagged to read with caution.

10%20%30%AVERAGE 25.7%1480 Satellite Blvd32.8%n=67 · interpret w/ caution4325 New Snapfinger Woods Dr31.5%n=127449 Roberts Court Rd31.3%n=67 · interpret w/ caution1032 Research Center Atlanta Dr31.1%n=122650 Ponce De Leon Ave NE29.9%n=1375300 Windward Pkwy29.5%n=784136 Jimmy Carter Blvd28.6%n=841670 Scenic Hwy N28.6%n=842295 Lawrenceville Hwy28.4%n=1481200 East West Connector SW27.8%n=1151000 Market Pl Blvd27.4%n=953885 Jonesboro Rd SE27.2%n=924343 Tilly Mill Rd26.2%n=1071100 Bullsboro Dr25.8%n=899037 Hwy 9225.6%n=902450 Cumberland Pkwy SE24.7%n=1584101 Roswell Rd23.1%n=913755 Shackleford Rd22.8%n=79875 Lawrenceville-Suwanee Rd21.8%n=1013605 Sandy Plains Rd19.7%n=716400 Peachtree Dunwoody Rd19.5%n=821580 Holcomb Bridge Rd16.7%n=66 · interpret w/ caution4028 Lawrenceville Hwy16.7%n=48 · interpret w/ caution2525 Piedmont Rd NE14.7%n=156

Source: 2,354 Atlanta-metro Google reviews by store · stores under ~70 reviews flagged for confidence · Dimension Labs analysis

Exhibit 3.2The worst stores aren’t just worse overall — they’re worse at helping shoppers

Proof from inside the same city: the bottom stores have far more “can’t get help” reviews than the top stores — a 9.3-point gap that the market can’t explain.

Share citing "can’t get help"Worst quartile 29.5%Best quartile 20.2%9.4-point gap, same city

Source: 2,354 Atlanta-metro Google reviews · each store’s share of “can’t get help” reviews, grouped by how many unhappy shoppers it has · Dimension Labs analysis

A wall of voices — at the named stores

Worst HD I’ve ever been to. Everybody has a horrible attitude, nothing can get done

1-star · 4343 Tilly Mill Rd, Atlanta

customer is horrible. They are unorganized and rude to contractors.

1-star · 1032 Research Center Atlanta Dr, Atlanta

Alan from customer service was great for helping me out!

5-star · 1670 Scenic Hwy N, Snellville

Anita was great. And very friendly.

5-star · 1480 Satellite Blvd, Suwanee

04 — Online orders, returns, and installation are the next problems — each smaller than staffing.

Once you account for getting help on the floor, the rest of the picture is clear. Online orders and pickup that fall through are the next real chunk of avoidable frustration; returns and installation are smaller but sharper. Each needs a different fix.

How we measured it. Each review is read for the specific problem it describes — an online order or pickup that fell through, a returns dispute, an installation breakdown, a mention of a competitor, or a shopper saying they won’t come back — and we counted each group. Full method in Methodology & Sources.

Online orders and pickup are the second-biggest fixable group, at 129 reviews. This is the “the website said it was in stock, so I drove over for nothing” problem. It shows up a little everywhere, but it hurts most where it lands on top of already-weak floor service — which is why 650 Ponce De Leon Ave NE, 2295 Lawrenceville Hwy, and 5300 Windward Pkwy need more than just more staff.

Returns and refund problems are smaller at 36 reviews, but clearly about process. The complaints aren’t about disappointing products. They’re about being turned away, brushed off, or handled with obvious annoyance — how the returns desk treats people and applies policy, not inventory.

Installation is the classic rare-but-severe problem. There are only 14 installation failures in the whole set, but 78.6% come with the shopper saying they won’t return or warning others off. When a contractor no-shows or a billing dispute lands after the install, the shopper is telling the internet to avoid the brand.

Shoppers leaving for a competitor reinforces the staffing story. Among named rivals, Lowe’s comes up most, with 81 mentions; 65 say they now prefer the competitor — and the most common reason is still not being able to find help. The leak isn’t mainly about price. It’s about service you can count on.

Home Depot and their contractors robbed me of about $500 with hidden charges and unethical billing practices. Never again.

1-star review · 1480 Satellite Blvd, Suwanee · an installation failure — rare, but reputation-grade

Exhibit 4.1The next-biggest problems, after staffing

Online orders and in-store pickup that fall through are the next real chunk of unhappy shoppers; shoppers leaving for a competitor, returns, and installation drop off from there.

Online order / pickup fails129Left for a competitor81Returns & refund problems36Installation problems14

Source: 2,354 Atlanta-metro Google reviews · review counts by problem type · Dimension Labs analysis

Exhibit 4.2What it would actually take to fix the unhappy reviews

Almost all of the fixable unhappy reviews come down to staffing and training first, then online/pickup and process fixes — the work is focused in one place, not scattered.

DARKER = MORE UNHAPPY SHOPPERS (1–2 STAR REVIEWS)Staffing & training464Online & pickup34Stock & supply28Store upkeep16Process & policy7Pricing & promos6Other/mixed49

Source: 2,354 Atlanta-metro Google reviews · 1–2 star reviews grouped by what would fix them · Dimension Labs analysis

A wall of voices — the other problems

they need to update their website and say that they’re out of generators

1-star · 3755 Shackleford Rd, Duluth

They act as if they are doing you a favour to give you cash back.

2-star · 3605 Sandy Plains Rd, Marietta

These companies are going to go out of business and Amazon is going to take over.

1-star · 3885 Jonesboro Rd SE, Atlanta

I’ll be shopping at Lowe’s from now on — at least someone there will help.

1-star · 650 Ponce De Leon Ave NE, Atlanta

The verdict — what to fix first, by payoff

Four fixes, ranked. One matters more than the other three combined.

This isn’t a “customer experience matters” report. It ends with a short list — ordered by how many unhappy reviews each fix would actually move. The number on the right is how many 1–2 star reviews that fix addresses.

1

Staff the floor & coach service

Most controllable, most reviews, proven by the control. Start here, fleet-wide.

464
unhappy reviews
Largest by far
2

Fix online orders & pickup

Worst where it lands on already-thin floor help — the eight stores weak on both.

34
unhappy reviews
Stacks on the worst stores
3

Retrain the returns desk

About how people are treated, not products. Cheap to fix; quick to feel.

7
unhappy reviews
Small but purely procedural
4

Escalate installation misses

78.6% of these reviews warn others off. Handle as district exceptions.

14
unhappy reviews
Rare, but reputation-grade

Severity × fixability — how to read the list

Staffing the floor sits in the top-right of every read: it is severe when it happens, common across the fleet, and the most controllable thing a district manager has. It is the rare fix that is both high-impact and high-fixability — so it goes first.

Installation is the mirror image: severe but rare and harder to control directly — handle it as a small set of exceptions, not a fleet-wide program. Returns and online/pickup sit in between: cheaper to fix, and quick for shoppers to feel.

DO FIRSTEasier to fix →More severe →Staff the floorOnline / pickupReturns deskInstallation
Methodology & Sources

How this brief was built

Everything in this brief comes from one source: 2,354 public Google reviews of 24 Atlanta-metro Home Depot stores, posted between January 1, 2025 and June 12, 2026. Dimension Labs read every review and tagged it for the experience it describes — the sentiment, the store and department, the specific friction or praise, whether a named employee was involved, and whether the customer said they would or would not return. The structured fields (called dimensions) used to produce every figure above are listed here for the technically inclined reader; they do not appear in the body because they are internal field names, not findings.

Dimensions used

overall_sentimentprimary_topicprimary_frictionstaff_interaction_signalemployee_nameddepartment_or_areaservice_type_mentionedcustomer_type_signalcompetitor_mentioncompetitor_contextrecommendation_signalissue_resolution_signalissue_severityactionability_tieraccessibility_or_courtesy_signalprimary_pain_point_phraseprimary_delight_phrasesentiment_verbatimstore_experience_summary

The analysis runs at one row per review, after collapsing duplicate records on the review’s unique key. Four reviews had no matching enriched record and were kept in the base count with empty dimension values; they do not affect any store-routing conclusion.

Methodology & Sources — continued

Operational lever read-through map

For each operational lever, the underlying tagged fields and the reasoning we drew from them:

LeverFields usedRead-through
Floor coverage / staff trainingprimary_friction = could_not_find_help, unhelpful_or_rude_staff, long_wait_or_understaffed; staff_interaction_signal = negative_unnamed or named_negativeLargest controllable detractor block; first fleet-wide fix
Online inventory accuracy / pickup executionprimary_friction = online_order_or_delivery_failure, out_of_stock_or_inaccurate_inventory; primary_topic = online_order_pickup_curbside, product_availability_stockSecond-largest recoverable cohort, especially where stacked on weak floor service
Returns policy / service-desk behaviorprimary_friction = return_denied_or_friction; primary_topic = returns_and_exchangesSmaller volume but clearly procedural and interaction-driven
Installation-vendor managementprimary_friction = installation_service_failure; issue_severity = serious failure contextLow volume, high reputational damage
Pricing / promotion accuracyprimary_friction = price_error_or_overcharge; primary_topic = pricing_and_promotionsLoud complaint set, but not the dominant separator between good and bad stores
Methodology & Sources — what this brief does not claim

Analytical frame, limits, and external figures

Causal frame. The argument holds the market constant: high rates, deferred remodels, and an estimated ~$22B home-improvement underspend fall on all 24 same-metro stores equally, so they cannot explain the between-store split. What varies store to store, and tracks the detractor gap, is labor coverage as customers experience it.

What the reviews can and cannot see. Reviews surface the customer-side output of staffing — whether a helpful, knowledgeable associate was found — not internal payroll, staffing ratios, or labor budgets. No per-store operating metric (revenue, labor budget, internal NPS) is invented anywhere in this brief.

Confidence handling. Stores under ~70 reviews and customer-type groups under about 25 are flagged “interpret with caution” in-line and never promoted to a headline. Friction-lift figures carry two-proportion z-tests; every cluster reported clears p<0.001. Charts are static and self-contained — no data is re-queried when the page is viewed.

Language. The review set includes 116 Spanish-language reviews; quotes preserve the original language where selected, untranslated, with full store, city, and star attribution.

External figures. Public financial and strategic figures cited in this brief — FY2025 comparable sales (+0.3%), the ~12.7% operating margin, the ~$1B standing associate-pay commitment (a standing FY2023 commitment, not a 2025 announcement), the ~$22B home-improvement underspend estimate, and the ~$24B combined SRS and GMS acquisitions — are drawn from Home Depot’s published financial reporting and investor communications. They are cited as context only; no corporate or per-store financials were estimated or invented.

Methodology & Sources — quality checks

What we verified before publishing

Pass

Every figure traces back to the review set; the underlying data was re-verified, not assumed.

Pass

Public figures cited as context only; no corporate or per-store financials were invented.

Pass

The leaderboard includes all 24 stores, with sub-70-review stores flagged for caution.

Pass

Charts are static and embedded; nothing is fetched or re-queried when the page is viewed.

Pass

Ten figures present, led by the Causal Spine: star distribution, reason split, lift, robustness, named-employee tally, leaderboard, quartile gap, secondary sizing, the fix-it map, and the verdict matrix.

Pass

Every quote is a real customer review, never fabricated, and Spanish reviews are kept in the original language.

Pass

Statistical tests applied to all lift figures; every reported cluster clears p<0.001.

Pass

The document is fully self-contained — fonts, charts, and data are embedded, and it renders identically offline.

The one standing caveat is sample depth at the smallest stores and within the sparsest customer groups; those are flagged in-line and never used to anchor a headline.

Prepared by Dimension Labs as a causal review-intelligence brief. Findings derive solely from the analyzed review set and cited public figures; nothing is re-queried when the page is viewed. This document is fully self-contained — fonts, styles, charts, and data are embedded — and renders identically offline.

© 2026 Dimension Labs · Causal Brief · Home Depot Atlanta Fleet · Window Jan 1 2025 – Jun 12 2026

Dimension Labs
Causal Intelligence