Dimension Labs · Causal Intelligence

BriarPatch
Food Co-op

A causal read of the customer experience at both BriarPatch stores, built from your member surveys and the public reviews. Prepared for BriarPatch and National Co+op Grocers.

Members love this co-op, and it shows. About 83% give a perfect five stars, quarter after quarter. When a visit does fall short, the reasons are few and specific, and the complaint you hear most is rarely the one that costs you most. This is what to fix, what each fix is worth, and where each is heading.

2
stores analyzed
2,839
member surveys read
217
public reviews added
21
signals from every comment
45%
of visits that hit the sharpest driver end in a low rating, vs ~3% normally
What we did. BriarPatch measures its members through the SMG survey, and that survey is the foundation of this analysis. We read every one of those member comments, added the public reviews for both stores as an extra layer, and turned each into 21 structured signals, then measured which ones actually move the rating.
1,502
Grass Valley surveys
+ 98 public (72G / 26Y)
1,337
Auburn surveys
+ 119 public (86G / 33Y)
01 · The action plan

The four things that actually drive a low rating, and the loud one that does not.

BriarPatch is loved: about 83% of members give a perfect five stars, and only about 3% of visits are rated three stars or below. The question is what pulls those few down. Each driver below is ranked by how often it ends in a low rating: of the visits that raise it, the share rated three stars or below, against about 3% of visits overall. Click any row for the members’ own words.

How to read the number. “45% end in a low rating” means that of the visits where staff friction came up, 45% were rated three stars or below, versus about 3% of visits overall. It is the share of the affected visits that end poorly, not a change to your overall average. Your average stays high because these problems are rare; the point is that when one does hit, it does real damage. Causally proven marks the drivers we further confirmed with a formal causal test, so they are a real cause, not a coincidence.
The real drivers, ranked by how often they end in a low rating rare, but when they happen the visit is usually rated poorly
Driver & who owns itHow often it ends in a low ratingHow often raisedTrend, 2024–2026
Staff friction
Own it: Store leadership & training
45%
rated 3★ or below · vs ~3% normally
53
1.8% of all comments
holding steady
0.5%→0.5% of surveys
2024–2026, per quarter
In members’ and reviewers’ own words
“The staff made me feel unwelcome. I was followed around the store, then the cashier and bagger ignored me.”
Auburn · Google · 1★
“Every time I go here the employees are rude to me. They always have an attitude.”
Grass Valley · Google · 2★
“There is a person in the meat department who is really rude. I wonder why he works there.”
Grass Valley · SMG · 3★
How to treat it
The sharpest single hit: when staff friction is named, the visit almost always ends low. Rare, so it is a training and floor-leadership fix, not a fire drill.
Deli
Own it: Deli operations, Auburn
26%
rated 3★ or below · vs ~3% normally
86
2.8% of all comments
▲ complaints rising
1.2%→1.9% of surveys
2024–2026, per quarter
In members’ and reviewers’ own words
“It takes over 20 minutes for one sandwich at the deli, and you can't call in an order.”
Auburn · Google · 1★
“For the past three months I've gotten food poisoning from the deli four times.”
Grass Valley · Google · 1★
“I ordered a turkey sandwich and there was no turkey on it, and I found out once I was already home, twenty minutes away.”
Grass Valley · Google · 1★
How to treat it
A strong driver, concentrated at Auburn: long waits, no call-ahead, an order made wrong. An operations fix at the newer store.
Freshness CAUSALLY PROVEN
Own it: Fresh & produce ops, Auburn first
18%
rated 3★ or below · vs ~3% normally
87
2.9% of all comments
▲ complaints rising
0.8%→1.9% of surveys
2024–2026, per quarter
In members’ and reviewers’ own words
“The half chicken was so nauseating I brought it back for a refund. The spine was covered in blood.”
Auburn · SMG · 2★
“Low quality products. Out of date products. Very high prices.”
Auburn · Google · 1★
“The produce I wanted wasn't available, and what was in stock wasn't fresh.”
Grass Valley · SMG · 3★
How to treat it
Severe per visit, formally proven, and climbing at Auburn. Small in the survey today, but it is your fastest-moving risk and the worst of your public reviews. Get ahead of it.
The price story CAUSALLY PROVEN
Own it: Pricing & member-value comms
12%
rated 3★ or below · vs ~3% normally
163
5.4% of all comments
▲ complaints rising
3.8%→5.2% of surveys
2024–2026, per quarter
In members’ and reviewers’ own words
“They recently did a price increase on many items, as much as $3 an item. I won't be able to shop here as much.”
Auburn · Yelp · 2★
“Very expensive, and not nearly the variety of Trader Joe's.”
Auburn · Google · 2★
“Overpriced, and they want you to sign up to be a member. I'm not sure what the value is.”
Auburn · Google · 3★
How to treat it
A real, proven driver and your most common one, drifting up. You move it by telling the value story: membership, sales, and why a co-op costs what it does. Not by cutting the shelf price.
Loudest, but not a driver: hold raised more than anything else, yet it ends in a low rating no more often than an average visit, and is easing on its own
Driver & who owns itHow often it ends in a low ratingHow often raisedTrend, 2024–2026
Selection
Own it: Category & buying
3%
rated 3★ or below · vs ~3% normally
283
9.3% of all comments
▼ complaints easing
13.9%→10.8% of surveys
2024–2026, per quarter
In members’ and reviewers’ own words
“Frozen raspberries on sale but out of stock. Strauss Greek yogurt out the last two times I shopped.”
Auburn · SMG · 3★
“There is a lack of items that used to be available in this store.”
Grass Valley · SMG · 3★
“On Wednesdays, milk or bread have been out of stock.”
Grass Valley · SMG · 3★
How to treat it
The loudest thing you hear, but it barely moves a rating and is easing on its own. Keep stocking the obvious requests; spend nothing chasing the long tail.
Do this first

When staff friction happens, nearly half of those visits end in a low rating, the single sharpest driver. But it is rare, so it is a targeted training and floor-leadership fix. The biggest operational levers are freshness and deli execution: common enough to matter, the most fixable, and worst at Auburn, where freshness is also climbing. Start there. Manage the price story rather than the price. Members tolerate a premium but react to a jump, so the lever is communication and targeted value, not across-the-board cuts. And do not spend the quarter chasing shelf requests: selection is the loudest thing you hear and does not drive low ratings.

45% → 12%
of the visits that hit a real driver end in a low rating, from staff friction (45%) down to price (12%), against ~3% normally. These four are what to fix.
3%
selection, the loudest theme at 9% of all comments, ends in a low rating no more often than an average visit. Loud, not a driver. Hold it.
The same five, on two axes
How often each ends in a low rating against how fixable it is. Bubble size is how many members raise it. Top-right is where a cheap fix protects the most visits; the big low bubbles are loud but not costly.
02 · What to do about it

The specific fixes, in the members’ own words.

Naming a driver is only half the job. For each one, here is the pattern members actually describe, drawn from the survey and review text, and the concrete moves that address it. Not generic best practice, but what these comments are asking for.

Staff friction
Store leadership & training · Auburn checkout
What members are telling you
  • Checkout is the hotspot, and it is Auburn. nearly every checkout complaint is at the newer store, not Grass Valley
  • Being ignored or not greeted. shoppers passing through unacknowledged, almost all at Auburn
  • A few named, recurring behaviors. a specific person or counter, not the whole team
What to do about it
  1. Set a checkout courtesy standard at Auburn: greet, make eye contact, thank. Nearly every checkout complaint in the data is at the newer store.
  2. Coach the floor-greeting habit. The “ignored” and “followed around” comments are Auburn-specific and are about acknowledgement, not staffing.
  3. Address the one or two recurring behaviors by name (e.g. the Grass Valley meat-counter interaction). This is targeted coaching, not a systemic problem.
Deli
Deli operations · Auburn
What members are telling you
  • Long waits for a single item. 20–30 minutes for one sandwich, with no way to order ahead
  • Prepared food quality. cold, soggy, or “boring” hot-bar and grab-and-go
  • Orders made wrong. the wrong or an incomplete order, discovered at home
What to do about it
  1. Staff the counter at peak and add call-ahead or online ordering. The marquee complaint is a 20-to-30-minute wait for one sandwich and “you can’t call in an order.”
  2. Add an order read-back at handoff. Several members got the wrong or an incomplete order, and a ten-second confirmation catches it.
  3. Tighten hot-bar and grab-and-go turnover. Cold and soggy prepared food is a rotation and timing fix, not a recipe one.
Freshness
Fresh & produce ops · Both stores, Auburn first
What members are telling you
  • Produce quality is the top complaint. wilting, browning, or simply “not fresh” in the case
  • Items past date reach the shelf. members finding expired or out-of-date goods
  • Meat & poultry draw the sharpest words. “bloody” and “nauseating,” the most visceral reviews
What to do about it
  1. Run a daily produce cull-and-rotate pass at both stores before open. Produce is the single most-raised freshness complaint, and it is the cheapest to fix.
  2. Add a twice-weekly pull-by-date sweep on refrigerated and dated goods. Members are finding expired items on the shelf; a standing date check stops it.
  3. Put a final visual QA on meat and poultry at case-fill. The rare but severe “bloody chicken” complaints are the ones that end up in a one-star public review.
The price story
Pricing & member-value comms · Grass Valley-led
What members are telling you
  • Membership value is unclear. “I’m not sure what the value is,” a communication gap, not just a price level
  • Members want visible deals. asking for sales and member pricing, not lower everything
  • Reaction to recent increases. members tolerate a premium but react to a jump
What to do about it
  1. Make member value legible at shelf and register: what membership and patronage actually return. A large share of price comments literally say they don’t know what the value is.
  2. Run a visible, consistent sales program on staples rather than across-the-board cuts. Members ask for deals more than they ask for lower prices everywhere.
  3. Explain increases when they happen, with brief signage on why (supplier costs, the co-op model). Members react to the jump, not the level.
Selection: hold, don’t chase

Almost entirely recurring out-of-stock on items that should be there (milk, bread, yogurt), not a call for new assortment. Selection is the loudest theme but the mildest driver, and it is easing on its own.

  1. Fix par levels on the handful of staples that keep going out of stock. This is an inventory tune, not an assortment expansion.
  2. Keep a simple bring-back log and act only on the items requested repeatedly. Do not spend the quarter chasing the long tail.
03 · The reshuffle

What fills your bad reviews is not what causes them.

The issues most common in low-rated feedback are not the ones most likely to cause it.

A sensible first cut is to read your low-rated reviews and rank the issues that show up in them. Do that and price leads, in a quarter of them, followed by deli, with selection close behind. It looks like the to-do list. But that ranking is inflated by sheer volume. Price and selection fill your worst reviews mainly because members raise them so often, not because they cause the low rating.

Rank the same issues instead by their low-rating rate, the share of the visits raising each one that actually end low, and the order turns over. Staff friction and freshness climb, price and selection fall. When price comes up, only 12% of those visits end low. When staff friction comes up, 45% do. Staff appears in fewer of your bad reviews, yet it is the far stronger signal that a visit went wrong.

What fills your bad reviews is not what causes them
Left: the five issues ranked by how often they appear in your low-rated reviews, at three stars or below. Right: the same five ranked by their low-rating rate, the share of the visits raising each issue that end low. Each line links one issue between the two views.
The left percentage is the share of your low-rated reviews that mention the issue. The right percentage is its low-rating rate: of the visits that raise it, the share that end low. Price and selection are common in bad reviews but rarely cause one. Staff and freshness are the reverse.
04 · The proof

These are causes, not coincidences.

Why each driver holds up as a real cause of a low rating.

Section 01 ranked the drivers by how often they end in a low rating. This section asks a stricter question: are they real causes, or coincidences? To answer it we measure how much each theme actually lowers a rating once store, source, member status and comment length are held constant, then try to break the result with formal tests.

Freshness and price pass. Each survives refutation tests and a sensitivity check strong enough that no realistic hidden factor would overturn it, so they are marked causally proven. Staff friction and deli point the same way on smaller numbers, so they are read as strong directions, not settled constants. That is the difference between this and a topic report. A topic report counts complaints. This shows which ones cause the low rating.

How much each driver lowers the rating, once everything else is held equal
The causal effect in stars. This is the magnitude behind the drivers; Section 01 is where to act.
The control, at every claim. Same co-op, same region, same window, so a spoiled item costing a star is the item, not the market.
05 · What’s moving

Freshness complaints are climbing, and selection gripes are fading.

A ranking is a snapshot. The survey has run every quarter since mid-2024, so it also shows direction, which is where the early warnings live. Two lines matter most: the most severe theme, freshness, is trending up, especially at Auburn, while the loudest theme, selection, is trending down.

Share of surveys raising each issue, by quarter
Member survey only, the consistent instrument, from 2024-Q3 forward. Use the toggle to isolate a store. Hover any point for the exact share.
Read the two bold lines. Freshness (red) is the most severe thing a member can hit, and it is rising, fastest at Auburn, so it is your fastest-moving risk even though it is small in the survey today. Selection (grey) is the most-mentioned theme and it is easing, further reason not to chase it. Price sits mid-pack and drifts up slowly.

Overall satisfaction has held remarkably steady. Roughly four in five members give a perfect five stars every quarter, and that has barely moved in two years, with a slight softening in the two most recent quarters worth watching rather than alarming over. The movement to act on is underneath the average, in the mix of what members raise, not in the top-line score.

Members rating five of five, by quarter
Share giving a perfect score. Member survey only, both stores.
High and flat near 82–84%, with a two-quarter dip to about 80%. The story is stability at a high level, which is why the underlying mix, and its direction, is where the leverage is.
06 · The two stores

One co-op, two different stores.

The stores look alike in the survey and different in the public reviews.

The two stores are different businesses. Grass Valley is the fifty-year flagship with a dense, loyal, close-in following. Auburn, open since 2023, is still finding its footing. In the member survey both score beautifully and look nearly identical, within six hundredths of a star. The public reviews are where the difference shows: they place Grass Valley a little ahead, and Auburn carries a larger share of the sharpest freshness and deli complaints. It has been hard to see from the survey alone, and it points to exactly where Auburn can close the gap.

Alike in the survey, apart in the public reviews
Mean rating by store, within each source.
Both stores earn strong survey scores. The public reviews, a smaller and more varied sample, place Grass Valley a little ahead. Read the size of the gap as a direction, not a settled number.

Grass Valley

50-year flagship
Member surveys1,502
Survey five-star80%
Public rating4.24★
Clearest leverPrice & selection

Auburn

opened 2023
Member surveys1,337
Survey five-star86%
Public rating4.05★
Clearest leverFreshness & deli
07 · Where your customers come from

Two stores, two different trade areas, drawn for the first time.

In 2026 the survey began asking for a home ZIP code. Mapped against the two stores, it draws each store’s trade area, something a rating average cannot show and something NCG has never had from survey data.

The two catchments
Each circle is a home ZIP, sized by how many members shop there, colored by the store they mostly shop. Dashed lines mark members who pass a nearer store to reach the other.
Grass Valley pulls a tight local ring: 77% of its members live within four miles. Auburn draws from a wider, more dispersed area, and its reach extends into Grass Valley’s own territory, with a slice of members traveling fourteen to twenty-three miles. Based on 575 responses with a ZIP, all from 2026, so read it as a first directional map.

This is a growth and marketing lens, not a problem to fix. Grass Valley is saturated locally, with little room to widen its ring. Auburn is still building its catchment, and because that catchment already overlaps Grass Valley’s, about one in ten members cross-shop between the two. Knowing which ZIPs each store actually serves, where the two overlap, and where the white space sits, is trade-area intelligence that has never been visible from survey data. It runs on any store in the network, and it points to marketing and siting decisions rather than to the fixes earlier in this brief.

08 · The compounding

A visit survives one complaint. It does not survive three.

The damage is accumulation, not any single miss.

The simplest check on the whole story is the problem ladder. A comment that names no clear problem is almost never unhappy and averages 4.9 stars. Name one and the unhappy share rises to 8%. Name two or more and it reaches 27%, with the mean falling to 3.75. Problems do not cancel out. They stack, which is why fixing the sharp few protects the most visits.

One problem hurts. Two collapse the visit.
Share of comments that are unhappy, by number of distinct problems named.
09 · How we read every voice

Your survey is the foundation. Public reviews widen the lens.

Two complementary sources, weighted the same, seeing more together.

BriarPatch measures its members through the SMG survey, and that survey is the foundation of this analysis: a broad, randomly sampled, first-party read of real members, and the most trustworthy single source you have. Like every satisfaction survey it also leans toward happy visits, which is exactly what you want it to capture.

Public reviews add the other end of the range. They are fewer and self-selected, the vocal minority who take the time to post, but they are a larger, more varied sample that validates and extends what the survey shows, and they reach voices a survey samples least. We weight the two the same here, because everyone who leaves a review is also a customer. Neither replaces the other. Together they see more.

How each source rates
Where member surveys and public reviews place their ratings.
Two sources, a fuller picture
Share of each source that is unhappy, three stars or below.
Member-owners and non-members inside the survey rate almost identically, so the difference between sources is the sampling, not the member. Both readings are valid; together they cover more ground.
10 · Mirroring your SMG view

We can also reproduce your SMG scorecard, from the comments alone.

Alongside the causal work, the same reading of the open-ended comments reproduces the quarterly topic scorecard you already know, so this sits directly beside your SMG reporting rather than replacing it.

Percent rating five of five, by topic and quarter
Reconstructed from open-ended comments. Darker is better. Grass Valley, then Auburn.
Price and value runs cool at both stores, quarter after quarter, the same signal the causal read ranks by consequence earlier in this brief.
Why this matters beyond BriarPatch

Everything in this brief (the 21 signals, the causal ranking, the impact math, the trend lines) was produced by one engine reading raw comments, with no custom setup for this co-op. The same engine runs identically on any store in the network. That means NCG can compare stores on the same causal yardstick, catch a freshness problem rising in one location the quarter it starts rather than the year it shows up in the average, and give every co-op the read BriarPatch just got, from the survey data they already collect.

The signals

The 21 dimensions read from every comment.

Each comment, member or public, was turned into these structured signals. The full library is what lets a pile of text become a measured, comparable, causal story, and it is what runs the same way on any co-op in the network.

Experience signals
Overall sentimentPositive, mixed, or negative feeling in the comment.
Primary topicThe single subject the comment is most about.
Price and valueWhether price or value is raised, and as praise or complaint.
Product quality and freshnessFreshness and quality of groceries, produce, meat, dairy.
Selection and availabilityAssortment, out-of-stocks, and requests to carry a product.
Staff and serviceAttitude, helpfulness, being greeted, or being ignored.
Checkout experienceLines, open registers, and the wait to pay.
Deli and prepared foodsHot bar, sandwiches, wait times, and taste.
Store cleanliness and environmentCleanliness, layout, crowding, and parking.
Loyalty and values
Community and membershipCo-op belonging, being a member-owner, local pride.
Sourcing and valuesLocal sourcing and mission integrity.
Loyalty signalWill return, shopping less, or will not be back.
Actionable issueA fixable operational problem versus a taste or a compliment.
Entities and evidence
Named staff memberThe specific employee named, in praise or complaint.
Competitor namedA competing grocer the customer compares to.
Specific product requestA named product the customer wants stocked.
Supporting verbatimsFour exact-quote fields for price, staff, quality, and deli, kept word for word.
Record typeSeparates real feedback from survey scaffolding, so nothing is miscounted.

Methodology & Sources

The corpus. 3,054 pieces of customer feedback, two BriarPatch stores, January 2024 to July 2026: the SMG member survey (2,839 responses), Google (156) and Yelp (59). The survey export splits each response across several rows, alternating a field label such as “Why Highly Satisfied” with its answer, so 2,839 real responses had inflated to 9,308 rows. The counts here reassemble one row per respondent and drop the scaffolding.

How effects were measured. Each theme is compared against similar feedback without it, holding store, source, member status and comment length constant. The two strongest drivers, price and freshness, were confirmed with formal causal inference: agreeing estimators, three refutation tests, and E-values of 3.5 and 4.0 for sensitivity to unmeasured confounding. Thin-base themes, staff service, checkout and cleanliness, are labelled Suggestive and read as directions.

On the two sources. The SMG member survey is the first-party foundation and the most trustworthy single source; public reviews are a complementary, larger, self-selected sample used to validate and widen the picture, not to outrank it. Both are weighted equally here because every reviewer is also a customer.

Limits. The public sample is 217 reviews, so the store comparison is directional, not decisive. ZIP codes were collected only from 2026 responses. Survey respondents and public reviewers are both self-selected. Auburn and Grass Valley share a region but differ in local affluence. Grocery net margin, about 1.7%, is an industry benchmark (FMI).