Every operating location read through its live guest signal, then translated into what it means on the floor. Built for Todd Garton ahead of leadership.
Twelve operating restaurants. One closed five weeks ago. One opening in about three weeks. The averages are healthy and almost entirely uninformative, which is the first finding.
Carrollton
North Dallas
Arlington
Thunderbird Pies
Fort Worth
Deep Ellum
White Rock
Frisco
Lake Highlands
Sachse
Addison
Fort Worth
Frisco
Stripe colour is my read on where attention is needed, not a Google metric. Cow Tipping shows its final live rating before the listing was marked closed. Verified ratings and counts read directly off live Google Maps listings on 20 Aug 2026. Complaint counts are defined in section 10.
Zoli's Addison is the cleanest operation in the group. Twenty consecutive recent reviews with not one complaint in the text, a 4.5 across 2,296 reviews, and owner replies landing inside twelve hours. Nothing else in the portfolio has a clean sheet.
Cane Rosso Frisco has the strongest hospitality signal, with a caveat worth reading in section 4.
Cane Rosso Carrollton. Six complaints inside a seven day window, against a portfolio norm of about one per week. Five of the six are the same complaint: the food takes too long. Verified
Cane Rosso North Dallas. Only two complaints, but one of them is the highest consequence item in the entire dataset: a guest was asked to read her debit card number over the phone after the payment link failed, and then received two wallet-add security alerts. Verified
I cannot answer this honestly from one pull, and neither can anyone else. A Google rating is a lifetime cumulative average. A single snapshot has no direction in it. Anyone who tells you a store is trending from one look at a star rating is guessing.
What I can show you is why the rating will never tell you either. Carrollton sits at 4.5 across 3,067 reviews. If it took twenty one-star reviews tomorrow, its published rating would be 4.48, which Google displays as 4.5. Fort Worth, at 3,363 reviews, is the same story. Your four largest restaurants are mathematically incapable of showing you a bad month. Verified arithmetic, checkable against the counts above.
Lake Highlands, at 313 reviews, would move from 4.5 to 4.3 on the same twenty reviews. So the rating is most responsive exactly where you have the least at stake, and frozen where you have the most.
Pace, and how it is communicated. Ticket time is the most repeated complaint across Cane Rosso. But the sharpest finding is that the same wait produces a one-star and a five-star at the same restaurant in the same week, and the variable is whether anyone said anything. Section 5, pattern 2.
Named people. Guests name your staff constantly and by first name: Maggie, Caleb, Shae, Priscilla, Kristina, Nikki, Ashley, Kyler, Jordan, Polo, Cameron, Carmen, Georgia, Aaliyah, Kimberly, Randy, Austin, Rebecca, Cian, Tony. One Carrollton guest went out of the way to name Arron Nguyen, the dishwasher. That is a culture signal you cannot buy and most groups this size never generate.
Thunderbird Pies and Cane Rosso White Rock share a parking lot and do not share a trade result. Both sit at 7328 Gaston Ave. White Rock collected its twenty most recent reviews in 21 days. Thunderbird needed roughly 60 days to collect twenty. Same corner, same guests walking past, roughly a third of the review velocity. Verified
Stop North Dallas taking card numbers verbally, today. Everything else in this document can wait a week. That one cannot.
Both matter for how the portfolio gets counted.
Cow Tipping Creamery Frisco is closed. Its final day was 14 July 2026 and the Google listing now reads Permanently closed. PILF is converting the space at 3685 The Star Blvd into The Imperial Room, an adults-only cocktail lounge, opening early September 2026. Jay Jerrier's stated reasons were the ten year lease expiring and ice cream being an incredibly seasonal business. Verified via the live listing plus CultureMap Dallas, 17 July 2026.
Worth noting without overclaiming: at 3.8 across 313 reviews, Cow Tipping was the weakest rated asset in the portfolio by a wide margin, and its final months carried product drift complaints, including a guest four months ago writing that the texture had gone grainy and not pleasant to eat. I am not claiming the reviews drove the decision. The stated reasons were lease and seasonality. But the guest signal and the real estate decision pointed the same direction, and the guest signal got there first.
There is no East Dallas or Fort Worth Cow Tipping to verify. With Frisco closed, the concept has no operating storefront in the portfolio. Treat Cow Tipping as dormant rather than as a fourth active brand. Inferred
So the count is 12 operating restaurants across three active concepts, with a fourth venue opening in about three weeks. That is the number to walk into leadership with.
For each location: what the number says, what the guests actually said, and what that most likely means on the floor. Complaint counts are my manual read of the 20 most recent reviews, counting only reviews whose text contains a clear complaint.
Five of six complaints in seven days are the same thing: the food takes too long, and nobody tells you why.
Small cheese pizza takes >30 mins.Superman Kryptonite, 3 days ago
Service was horrible. Food was good but took way too long to come out. Waiter ignored us multiple times. We had to get our own to go boxes and my wife's pizza was not even made until my daughter and me got our food.Raymond Cate, 1 star, 4 days ago
Order ready time is given as 30 minutes; I have never had it take less than 45, even if the restaurant is empty.C S, 1 week ago
Three separate mechanisms are tangled here and they need separating. Raymond's plates arrived at different times, which is expo and fire sequencing, not oven speed. C S is describing a quoted time that is systematically about fifteen minutes optimistic, which is a quoting problem, not a kitchen problem. Curtis reported burnt pizzas alongside feeling rushed, which is what happens when a bake gets pushed to catch up. Inferred from guest description; the ticket data would confirm which is dominant.
The staff is almost always on point and even if the pizza oven queue gets slow when they are busy they always communicate well. Great with the kids and provide quick and attentive service.Connor Osburn, 5 stars, 4 days ago
Same restaurant, same week, same oven queue. Connor names the slowness explicitly and gives five stars because someone talked to him. This is the single most useful sentence in the entire dataset.
Carrollton has the strongest complaint recovery behaviour in the group. Every negative review is answered inside two days and routed to a named manager, Michael Adame, with a real email address. Only one other location does this.
Patio is the top guest topic at 86 mentions, the highest patio count in the portfolio. A guest named Arron Nguyen, the dishwasher, in a public review.
Placed an order over the phone. The link to pay did not work. I called back and was told to give my debit card info over the phone. Immediately after my card was charged for the pizza I got two "Add to Wallet Request" security [alerts]Leah McCaskill, 1 week ago
Two failures stacked. The payment link broke, and the recovery path a team member reached for was reading card numbers aloud over the phone. Whatever caused the wallet alerts, a restaurant should never be the place a card number gets spoken. The owner reply took four days, which is the slowest response to a serious complaint anywhere in this dataset. Verified
They have a new QR code they use to pay. You can still just hand a card to your server but I already said I'm fine with the QR code. That was until I saw there was a $0.99 [fee]Michelle Chappell, 1 week ago
A payment change is live at North Dallas and both paths are generating complaints: the fallback path is unsafe and the default path carries a fee guests did not expect. These are one problem, not two.
4.2 is the lowest rating of any operating location, but the recent window does not look like a 4.2 restaurant. Eighteen of the twenty most recent reviews are positive, servers Austin and Randy are named warmly, and replies land inside a day. Inferred: the 4.2 is carrying older damage and current execution is better than the number. With only 578 reviews this store can actually move its rating, unlike the big four.
One structural note: noise level appears as a Google review topic here with 7 mentions, which it does not at most other locations.
Arlington collected its twenty most recent reviews in about five days, faster than any other location. High traffic and high visibility, which cuts both ways.
A one-star review from four days ago is unanswered. Five-star ratings from the same day were answered. That is the wrong way round.
I was done eating and I was trying to get someone's attention to try and place a to go order, by the time I got someone's attention it was 9:43. I was told by the general manager (Hannah) [that] the kitchen closed at 9:45 and couldn't [take it]Marissa Domignuez, 1 star, 4 days ago, no reply
A guest already seated and spending was refused an add-on order two minutes before close. That is a policy call a GM is entitled to make, but it produced a public one-star that nobody has answered. Verified
75% of the crust was burnt. Server said this was normal.Joy Janneck, 3 days ago
Two problems in one sentence: a bake that went too far, and a server who normalised it instead of remaking it. See pattern 1.
Perfect place for my allergies. I actually had a vegan pizza that tasted awesome!! I haven't had a real pizza over a yr.Theda Harjochee, 3 days ago
Arlington also has a named GM, Hannah Wells, appearing in owner replies with a working email. Server Kristina drew a detailed five-star.
after sitting in restaurant and constantly ordering and clarifying to the waiter i was ordering non-pork options, i ordered a non pork pizza to go for my sister. i drive an hour home. i get home, and there is literally pepperoni on the [pizza]Eman Saleem, 1 day ago
This is not a wrong-topping complaint. A guest stated a religious dietary restriction repeatedly, out loud, to a server, and it did not survive the trip to the kitchen. She drove an hour before discovering it. The restaurant apologised with a generic reply and offered no named contact and no recovery path. Verified
The mechanism is almost certainly that the restriction lived in conversation and never made it onto the ticket. That is a fixable process, and it is the same process that protects allergy guests.
Caleb is named in three separate reviews in three weeks, once with "give him a raise". Aaliyah and Kimberly also named. Patio is the dominant topic at 70 mentions. Pasta of the month is pulling genuine enthusiasm, with one guest calling it the reason she comes monthly.
White Rock took 21 days to collect twenty reviews, the slowest of the eight larger Cane Rosso stores. Volume of guests is not the issue given the rating and the patio traffic; this looks like nobody is asking. Inferred
3,363 reviews at 4.5. The strongest food language anywhere in the group: guests repeatedly compare it to Italy.
Best pizza in town, most like our experience in Italy! Garlic cheese bread is to die for. Get the brisket lasagna if it's not sold out!Mette Blockhan, 4 days ago
The eleven most recent reviews are all unanswered, going back four days, including the store's only recent complaint, posted 13 hours ago: "Service could have been A LOT better for a somewhat upscale pizza place." Replies resume only at the four-day mark. Deep Ellum answers inside 18 hours. Zoli's Addison inside 12. Verified
Ashley gave us amazing service! I have been to other locations and service was not great as hers.Gabriella Dupre, 5 days ago
Guests are comparing your locations to each other in public. Ashley, Jordan and Kyler all named this week.
Eight of the twenty most recent reviews name one server. Madison, Hunter, Phillip, Josue, Christopher, Javier and Clint all name Maggie, and an eighth reply confirms her by name. Several reviews contain nothing else at all: the entire review is her name.
Maggie was awesome! Our party was huge!!! Service was impeccable!Phillip Potter, 3 days ago
This is the strongest single hospitality signal in the portfolio and it should be recognised loudly. It is also a concentration risk: 40% of Frisco's recent public reputation is one person's section. If Maggie takes two weeks off, Frisco's review flow changes character and nobody will know why. Inferred
The useful question is not how to keep Maggie. It is what she does at the table that the other servers do not, and whether it can be taught. That is a shift with her, not a survey.
Frisco is 4.4, tied for the second lowest operating rating, while producing the best recent service sentiment in the group. Another example of the lifetime average failing to describe the current restaurant. The Imperial Room opens next door in early September, which will change this store's evening traffic pattern.
Near total reply coverage with a typical turnaround of 18 hours to 3 days. If you want a house standard for reply speed, it already exists here.
Make sure to give these guys some extra love while construction is taking over the street in front of the building. Food is amazing as always, as is the service.diana flurry, 2 weeks ago
Street construction in front of the building. A guest is publicly advocating for the store because of it. Any traffic softness at Deep Ellum over the next stretch has an explanation that is not the operation, and it would be an error to read a dip here as an execution problem. Verified
One harsh crust complaint, covered in pattern 1. Otherwise strong: Carmen named twice, Georgia named, and a vegetarian guest specifically praised for how the team handled her diet, which is the direct counterpart to the White Rock miss.
Google clusters review topics automatically. At Sachse, one of those clusters is "soggy pizza", with 4 mentions. No other location in the portfolio has a negative topic cluster surfaced on its listing. That cluster is visible to every guest who scrolls the reviews. Verified
It's like this place knows what is wrong but refuses to fix it. Without asking, our waitress gave us a disclaimer that the crust is floppy.Eric Thomas, 1 week ago
The server pre-apologised for the product before the guest had tasted it. That converts a style choice into an admitted defect. See pattern 1, which is the most transferable finding in this document.
Highest volume of clean five-star ratings in its window. Nikki named. The meme wall is generating unprompted delight, and one guest called out that the inside dining room is not noisy, which is the opposite of the North Dallas signal.
Only 313 reviews, so this is the youngest listing. It also has noticeably the most human owner replies in the group: varied, specific, not templated. Whoever writes these should write the house guide.
Tried to get a table for 2. Front desk basically told us no because they don't have enough servers for all of their open tables they had. How can a restaurant this busy not be staffed? Very rude service right from the start so we just left.Luke Cannaday, 2 stars, 1 week ago
Open tables, refused cover, guest gone. This is the clearest labour-model signal in the portfolio: the constraint was servers, not seats. One instance is not a pattern, but it is the exact scenario that turns into one quietly. Verified, single instance.
The crust had that nice charred edge that adds a lot of flavor without making it taste burnt.happy fun, 6 days ago
Same product attribute that reads as a defect at Arlington and Sachse reads as craft here. Shae named twice, once with "give her a raise". Polo named.
Twenty consecutive recent reviews, two weeks, not one complaint in the text. No other location in the portfolio comes close. Replies land in roughly twelve hours and use their own distinct format, which suggests Zoli's is being handled by a different hand than Cane Rosso.
Zoli's has no crust complaints at all. Not one, in either location, in the recent window. Guests describe it as light, crispy crust and move on. This is the cleanest evidence that Cane Rosso's crust problem is not a dough problem: New York style delivers the texture guests already expect, so nobody has to explain it. Neapolitan does not, so somebody does. Pattern 1.
Priscilla named twice, both times with the drink or appetizer she recommended, which means she is actually selling. Cameron, Tony, Cian and Rebecca also named. Zucchini fries at 34 mentions and garlic knots at 33 are doing real work as signature items.
Outdoor seating is the dominant topic at 104 mentions, the highest single topic count anywhere in the portfolio, and kids play area appears at 15. This is a family destination with a playground, not a pizza counter. It should be staffed, scheduled and marketed as one.
Delicious pizza and the kids love playing on the playground!Caitlyn Cozart, 5 days ago
Lowest rating of the two Zoli's and second lowest operating. The recent window does not explain it: the only soft note in twenty reviews is a DoorDash order that arrived slightly cold. Inferred that the 4.3 is historical rather than current.
Came back to edit my review. Decided to try Zoli's again after being a little disappointed the first time, and they did not miss this time. Crust was amazing, great cheese, great pepperoni.Chris Davis, edited 6 days ago
A guest who was disappointed, returned, and publicly rewrote his own review upward. That is the outcome every recovery process is trying to produce, and it happened here.
Thunderbird needed roughly sixty days to collect twenty reviews. Cane Rosso White Rock, in the same building at 7328 Gaston Ave, needed twenty-one. Deep Ellum needed fourteen. Thunderbird is running at roughly a third of its neighbour's review velocity in the same trade area. Verified
Review velocity is a rough proxy for cover count, not a measure of it, so treat this as a flag rather than a verdict. But it is the flag I would chase first at this store, because it is the only location where the number of people talking is the finding. Inferred
The crust had a nice crispy exterior, but the dough itself was way too dense for my liking. The Drip Pan was disappointing. It was pretty bland, and the mushrooms did not seem fully cooked.Winterwinds, 3 weeks ago
the pizza we received was dense and seemed poorly made. More than half of it ended up in the trash.April McKnight, 2 months ago
Cane Rosso's complaints are undercooked and soggy. Thunderbird's are dense and overbaked. Both are bake execution, at opposite ends. Detroit style is unforgiving about proof time and pan temperature, which is where I would look first. Inferred
We came in as fans of Cane Rosso and were excited to give this place a try. Unfortunately... we don't plan to return and try again.April McKnight, 2 months ago
Guests arrive at Thunderbird carrying Cane Rosso expectations. That is free traffic and borrowed risk in the same sentence.
Why would you all make a pizza first and then have me wait another 10 minutes for cheese sticks.Tone Def, 1 week ago
the negative about this place is the soda :( has to be the worstZarina Luttrell, 3 weeks ago, alongside praise for the pizza
Takeout items are being fired in the wrong order, and the fountain needs calibrating. Neither is a hard fix and both are actively costing stars.
Gluten free is a real differentiator here, appearing as a Google topic cluster with 14 mentions and drawing the most detailed positive review on the listing. Detroit style pizza is the dominant topic at 74. Arcade games appear as their own cluster. Thunderbird also has the most consistent complaint recovery in the group: every negative review routes to Johnny Brewington by name.
Direct comparisons, only where the evidence supports one.
The windows differ by store, so raw counts are not comparable. Normalising to complaints per week is. Verified counts, Inferred rate, since Google shows relative dates like "a week ago" rather than exact ones.
| Location | Rating | Reviews | Window | Complaints | Per week |
|---|---|---|---|---|---|
| Cane Rosso Carrollton | 4.5 | 3,067 | 7 d | 6 | 6.0 |
| Cane Rosso Arlington | 4.4 | 1,536 | 5 d | 2 | 2.8 |
| Cane Rosso Deep Ellum | 4.5 | 2,806 | 14 d | 2 | 1.0 |
| Cane Rosso Fort Worth | 4.5 | 3,363 | 7 d | 1 | 1.0 |
| Cane Rosso Frisco | 4.4 | 1,746 | 7 d | 1 | 1.0 |
| Cane Rosso Sachse | 4.5 | 672 | 7 d | 1 | 1.0 |
| Cane Rosso North Dallas | 4.2 | 578 | 14 d | 2 | 1.0 |
| Zoli's Fort Worth | 4.3 | 1,595 | 7 d | 1 | 1.0 |
| Thunderbird Pies | 4.5 | 495 | ~60 d | 6 | 0.7 |
| Cane Rosso White Rock | 4.5 | 1,951 | 21 d | 2 | 0.7 |
| Cane Rosso Lake Highlands | 4.5 | 313 | 14 d | 1 | 0.5 |
| Zoli's Addison | 4.5 | 2,296 | 14 d | 0 | 0.0 |
The finding: Carrollton is generating complaints at roughly six times the portfolio norm, and every store above shares the same 4.5 or near it. Rating cannot see this. Complaint rate can.
This is where the portfolio is least consistent, and it is entirely within your control. Verified
| Location | Typical reply speed | Named recovery contact | Gap observed |
|---|---|---|---|
| Zoli's Addison | ~12 hours | No | None |
| Cane Rosso Deep Ellum | 18 hours to 3 days | No | None |
| Cane Rosso Carrollton | 1 to 2 days | Yes, Michael Adame | None |
| Thunderbird Pies | 1 to 3 days | Yes, Johnny Brewington | None |
| Cane Rosso Lake Highlands | 1 day | No | None |
| Cane Rosso Sachse | 2 days | No | Slower on rating-only reviews |
| Cane Rosso White Rock | 19 hours | No | Severe complaint got a generic reply, no path |
| Cane Rosso Arlington | 1 day | Partial, Hannah Wells | 1-star unanswered, 5-stars same day answered |
| Cane Rosso Fort Worth | Stalled | No | 11 most recent unanswered, complaint included |
| Cane Rosso North Dallas | 1 day typical | No | 4 days to answer the card-over-phone report |
Only where defensible. Two are deliberately left empty.
Zero complaints in twenty reviews, 4.5 across 2,296, fastest replies in the group.
Eight of twenty recent reviews name a server. Concentrated in one person, which is the caveat.
Largest base at 4.5 with the most vivid food language in the portfolio. Service, not food, is its soft edge.
Not the most complaints, the most consequential one. Card data handling outranks ticket times.
Present at Carrollton, Arlington, Thunderbird and Lake Highlands. Concentrated hard at Carrollton.
For the clean sheet and the reply speed. Carrollton supplies the specific play, in Connor Osburn's review.
Six times the portfolio complaint rate, one repeating cause, and the antidote already documented in its own reviews.
North Dallas is the candidate: 4.2 lifetime with a strongly positive recent window. But one snapshot has no direction in it. This needs a second read to become a claim.
Carrollton looks hot right now, but I cannot separate a bad week from a trend without prior weeks to compare. Naming it would be a guess.
Where one concept already solved what another is struggling with.
Across both Zoli's locations there is not a single crust complaint in the recent window. Across Cane Rosso there are crust complaints at four separate locations, plus a Google topic cluster reading "soggy pizza" at Sachse. The difference is not quality, it is expectation. New York style delivers exactly the texture a guest already pictures, so nobody has to explain anything. Neapolitan does not, and so a server ends up explaining it, usually too late and usually as an apology.
Thunderbird sits at the third point of the triangle with the opposite failure: dense and overbaked rather than soft and wet. Three concepts, three crusts, three completely different guest expectation problems, and no shared language for handling any of them.
Both Zoli's locations are effectively free of pace complaints. Cane Rosso carries nearly all of them. That is not surprising given a fired-to-order wood oven versus a New York slice line, but it has a practical consequence: the Cane Rosso wait is structural and permanent, so it can only be managed through communication, never eliminated through speed. Carrollton is trying to solve it by rushing, and burning pizzas in the process.
The named recovery contact. Every negative review at Thunderbird routes to Johnny Brewington by name and email. Carrollton does the same with Michael Adame. Seven other locations do not, and White Rock's dietary failure is the clearest cost of that: the most serious complaint in the dataset got a generic apology and nowhere to go.
One only: a server explaining away the product instead of fixing it. Arlington, "server said this was normal" about burnt crust. Sachse, an unprompted disclaimer that the crust is floppy. Deep Ellum and Frisco, a guest attacking the house explanation itself as a copout. This behaviour is being learned somewhere, and it crosses locations.
Signal, where it appears, what it likely means inside the restaurant, and who should own the next step. Correlation is labelled as correlation.
Servers pre-emptively explaining or normalising the crust; guests reading that explanation as an excuse.
Sachse: "Without asking, our waitress gave us a disclaimer that the crust is floppy." Plus a Google topic cluster reading "soggy pizza", 4 mentions.
Arlington: "75% of the crust was burnt. Server said this was normal."
Deep Ellum and Frisco: the same guest, posting identical text at both, calling the house explanation "a copout for soggy pizza".
Lake Highlands, the control case: "that nice charred edge that adds a lot of flavor without making it taste burnt."
Honesty note: the Deep Ellum and Frisco complaints are one guest duplicated across two listings, not two independent guests. Counted once.
Two distinct things are happening and they need separating. At Arlington and Carrollton the pizza is genuinely coming out too dark, which is a bake problem. Everywhere else the pizza is correct and the framing is failing. A soft, foldable, leopard-spotted centre is what Neapolitan is. Said before the first bite it is craft. Said after, or as a disclaimer, it is a confession. Inferred from guest language.
Give every server one sentence to say when the pizza lands, before the guest picks up a slice. Not a defence, a description. Separately, pull bake times at Arlington and Carrollton, where the char is real.
Director of Operations for the language, Kitchen Leads at Arlington and Carrollton for the bake.
Ticket time complaints clustering at Carrollton, alongside a five-star review from the same week describing the identical wait.
Six complaints in seven days at Carrollton, five of them about pace. Same store, same week, five stars: "even if the pizza oven queue gets slow when they are busy they always communicate well." Pace also appears at Arlington, Lake Highlands and Thunderbird, but at roughly one sixth the rate.
A wood oven has a fixed throughput. The wait cannot be engineered away, so the only variable left is whether a guest knows it is coming. Carrollton's own reviews contain both outcomes with the same underlying delay. Three mechanisms are mixed in: oven queue, table fire sequencing, and a to-go quote that runs roughly fifteen minutes optimistic. Inferred. This is correlation with a mechanism, not proof of cause.
See the play in section 7.
Carrollton GM, with the Kitchen Lead on fire sequencing.
One severe accuracy failure against three strong successes, all in the same short window.
White Rock, the failure: a guest "constantly ordering and clarifying to the waiter" that she needed non-pork, drove an hour home, found pepperoni.
Arlington: "Perfect place for my allergies. I actually had a vegan pizza that tasted awesome!!"
Thunderbird: gluten free surfaced as a topic cluster with 14 mentions.
Deep Ellum: a vegetarian guest publicly thanking the team for accommodating her.
You are good at this three times out of four, which makes the fourth more dangerous, not less, because nobody thinks there is a problem. The failure mode is specific and mundane: the restriction was spoken, acknowledged, and never written anywhere the kitchen could see it. The same gap that produces a pork mix-up produces an allergy incident. Inferred mechanism, single instance.
Any stated restriction gets entered as a modifier on the ticket, and gets read back to the guest at the table. On to-go orders it gets read back at handoff, because the guest cannot check at the window.
Director of Operations, house-wide. Not a White Rock fix.
Reply speed, personalisation and escalation path all vary by location with no visible standard.
Carrollton and Thunderbird route every complaint to a named manager with an email. Deep Ellum and Zoli's Addison answer fast but offer no path. Fort Worth has stopped replying entirely, eleven reviews deep. Arlington answered five-star ratings and left a one-star from the same day untouched. North Dallas took four days to respond to a card security report.
Replying and recovering are different jobs and only two locations are doing the second. A templated thank-you on a five-star costs nothing and earns nothing. An unanswered one-star is the only review a prospective guest will actually read closely. Verified behaviour, directly observable on every listing.
One rule: every review with a complaint in it gets a reply within 24 hours, from a named person at that location, with a way to reach them. Rating-only five stars can wait or be skipped entirely.
COO sets the standard, GMs execute per location.
Two complaints in the same week, both touching the same newly changed system.
The pay link failed and the fallback was reading a debit card aloud over the phone, followed by two wallet-add security alerts. Separately, the new QR pay flow carries a $0.99 fee that a willing guest objected to only when she saw it.
These read as one rollout with an unsafe fallback and an undisclosed cost. I cannot establish what caused the wallet alerts and I am not claiming the restaurant did. What is certain is that the store created the exposure by taking the number verbally. Verified guest report, Unknown root cause.
Immediate: no card numbers taken verbally, at any location, for any reason. Then fix the link, and put the QR fee in front of the guest before they scan rather than after.
COO plus whoever owns the payments vendor relationship. This one is above store level.
The highest confidence issue in the portfolio, converted into something a GM can run tomorrow.
Repeated complaints about ticket times and being ignored during the wait.
Cane Rosso Carrollton.
Five pace complaints in seven days, against roughly one per week across the rest of the portfolio. One five-star from the same week describing the same delay and praising the store for communicating through it.
Three changes, all verbal, none structural.
1. Quote the real number. To-go quotes move from 30 minutes to 45. A guest told 45 who waits 40 is happy. A guest told 30 who waits 45 writes the review C S wrote.
2. Name the oven at the table. When the kitchen is backed up, the server tells the table before they ask, with a number: "your pizzas are about fifteen out, the oven is full." Connor Osburn already told you this works.
3. Fire the table together, or explain why not. Raymond Cate's family got their food at three different times and one pizza had not been started. If plates cannot land together, someone says so before the first one hits the table.
Carrollton GM owns the floor language and the quote change. Kitchen Lead owns fire sequencing on tables of four or more.
Not "faster pizza". The observable change is that guests who waited a long time say so and still leave a good review, the way Connor did. Pace stops appearing next to words like "ignored" and starts appearing next to words like "they told us".
Next Monday, count three things at Carrollton and only these three:
1. Complaints mentioning pace in the week's reviews. Baseline this week is 5.
2. Reviews mentioning a wait without complaining about it. Baseline this week is 1, Connor's.
3. Whether any review repeats the to-go quote gap. Baseline this week is 1, C S.
Ratio two against one is the real metric. It should move before the star rating does, and given 3,067 lifetime reviews, the star rating will not move at all.
Named, so it can be replicated rather than admired.
Maggie, Frisco. Eight of twenty recent reviews. Several reviews are nothing but her name, which means guests who would not otherwise have written anything wrote something because of her.
Caleb, White Rock. Three reviews in three weeks, one of them just "Caleb is a great server give him a raise!"
Shae, Lake Highlands. Twice in a week, also with a raise request. Priscilla, Zoli's Addison, twice, and both times the guest names the item she recommended, which means she is selling and it is landing.
Also named this month: Kristina and Hannah at Arlington. Ashley, Jordan and Kyler at Fort Worth. Nikki at Sachse. Austin and Randy at North Dallas. Polo at Lake Highlands. Carmen, twice, and Georgia at Deep Ellum. Aaliyah and Kimberly at White Rock. Cameron, Tony, Cian and Rebecca at Zoli's Addison. And Arron Nguyen, a dishwasher at Carrollton, named by a guest in a public review, which almost never happens anywhere.
Honey Bastard and the dessert pizzas across Cane Rosso. Brisket lasagna at Fort Worth and Sachse, with guests noting it sells out. The pasta of the month at White Rock, which produced the most enthusiastic single review in the pull. Zucchini fries at 34 mentions and garlic knots at 33 at Zoli's Addison. Spinach dip at Sachse, 17. Detroit style at Thunderbird, 74, the single strongest item-to-concept association in the group.
The pattern under all of it: your strongest asset is not the pizza, it is that guests form relationships with individual people on your floor and then write those people's names on the internet. That is unusual at 12 locations. It is also fragile, because right now it depends entirely on who happens to be working.
Five items. Two minutes. This is the whole thing, every week.
12 operating locations · 3 active concepts · Imperial Room opens in ~2 weeks
North Dallas took a debit card number over the phone. The pay link failed, a team member asked the guest to read her card out loud, and she reported two wallet-add security alerts immediately after. It took four days to answer her.
One instruction, today, every location: nobody takes a card number verbally, ever, for any reason. Then find out why the link failed.
Thunderbird is quiet. It took about 60 days to collect the reviews White Rock collects in 21, and they share a building. Guests who do write are describing dense, overbaked dough, and one arrived as a Cane Rosso fan and left saying she would not return.
No intervention yet. Get the cover counts next to the review counts before deciding whether this is a traffic problem or a prompting problem.
Maggie at Frisco is carrying 40% of that store's recent reputation by herself. Eight of twenty reviews name her, some name nothing else.
Zoli's Addison ran a clean sheet: twenty reviews, two weeks, zero complaints. Nowhere else in the group did that.
Call out both by name this week. Then ask Maggie what she says at a table, because that answer is worth more than the recognition.
Carrollton, say the number out loud. Six complaints in seven days, five about pace, against one a week everywhere else. Move to-go quotes from 30 to 45 minutes, tell backed-up tables how many minutes out they are, and fire tables together or say why not.
Your own five-star review from the same week proves it works. Owner: Carrollton GM.
Empty this week, and it should be. This is the first read, so nothing has been acted on yet and there is nothing honest to put here.
Next Monday this slot holds the Carrollton result: pace complaints against waits-mentioned-without-complaint, week over week. If that ratio does not move, the play was wrong and you will know inside seven days.
So every number above can be checked or challenged.
Source. Every rating, review count, review text, topic cluster and owner reply timestamp was read directly off the live Google Maps listing for each location on 20 August 2026. No aggregators, no third-party mirrors, no estimates. Ratings and counts are exactly what a guest sees today.
The analysis window. The twenty most recent reviews per location, sorted newest first. That window covers 5 days at Arlington and about 60 at Thunderbird, which is itself a finding rather than a flaw. Older reviews were deliberately excluded.
The complaint count is mine, not Google's. It counts reviews in that window whose text contains a clear complaint. It is a manual read and another careful reader might land one either side on a borderline case. It also systematically undercounts, because a large share of Google reviews carry a star rating and no text at all, and a silent one-star is invisible to any text-based method.
The weighted rating of 4.46 is computed from Google's displayed ratings, which are already rounded to one decimal. It is accurate to roughly a hundredth, not better.
What this cannot tell you. Direction. One pull has no trend in it, so "improving" and "deteriorating" are left unfilled rather than guessed at. It also cannot see covers, sales, labour hours, ticket times, or anything a guest did not choose to write down. Where I have inferred a mechanism inside the kitchen from a guest's description, it is labelled Inferred and should be checked against your own data before anyone acts on it as fact.
One correction I made to my own analysis. The harshest crust complaint appeared at both Deep Ellum and Frisco. It is one guest posting identical text at two listings, not two independent guests. Counted once. A naive theme count would have doubled it, and that is exactly how a portfolio report talks a leadership team into fixing a problem that is one person loudly.
Everything above came from data PILF already owns. It is public, it is free, and it is sitting on twelve Google listings right now. Nothing here required access to your systems.
What took the work was the chain: noticing that five Carrollton complaints in a week are one problem, that a five-star review in the same week contains the fix, that the fix belongs to a specific GM, that the star rating is mathematically incapable of telling you whether it worked, and that something therefore has to remember what was tried so next Monday can answer for it.
Guest signal to portfolio pattern to location to owner to action to measured outcome to memory. That chain is the thing PILF does not have, and it is why Fort Worth has eleven unanswered reviews and nobody noticed, and why Carrollton's own answer to its own problem has been sitting in its own reviews for four days.