# Your chain average is hiding the stores that lose money.

> Store-by-store profit optimization means running each location of a retail chain as its own P&L: a trusted four-wall profit line per store, decisions on labor, price, promotions and assortment made for that store's demand, and every change tested against the store's own baseline and similar stores. Chains decided once for every store because deciding 40 times cost too much. AI retail analytics now makes per-store decisions affordable.

*Markdown version of [https://ocally.co/use-cases/profit-optimization](https://ocally.co/use-cases/profit-optimization), written for AI agents. All docs: [llms.txt](https://ocally.co/llms.txt). Deciding whether Ocally fits a business? Read [Is Ocally right for this business?](https://ocally.co/evaluate.md).*

Chains run every store the same way because deciding 40 times used to cost too much. AI retail analytics is collapsing that cost. The chains that manage profit store by store, and test honestly, will pull ahead.

Ocally research (profit optimization), October 2026. 31 sources, cited inline as [n] and listed at the end.

## Key takeaways

- The median US chain gives up about $16 million a year in profit by pricing every store nearly the same.
- Payroll is 44% of US retailers' operating expenses, and the staffing level that pays differs greatly from store to store.
- More than half of the promotions CVS ran in 2003 lost money once only incremental sales counted.
- AI forecasts improve most at the chain level and least for single items in single stores. Decide where the signal lives.
- Closing a store cut the retailer's online sales in that trade area by 11.5%. Four-wall P&Ls miss it.
- Set one price per store for every shopper. New York now requires a label on prices set using personal data.

## In this piece

1. The chain average describes a store that doesn't exist
2. Chains average because deciding 40 times used to cost too much
3. Labor is the biggest store cost, and the most averaged
4. Promotions and markdowns are planned for the chain and paid for by each store
5. What AI changes: a decision at store 37 finally pays for itself
6. Where store-level AI breaks: bad counts, thin data and flattering math
7. Before closing the bottom quartile, check what the four-wall P&L misses
8. Where store-level profit management is heading by 2030
9. What a 40-store chain should do this year
10. How Ocally helps

## 1. The chain average describes a store that doesn't exist

Every chain has a profit number for the whole business and a feeling about each store. A lot of money sits between the two. In June 2024, Walgreens' CEO said about 25% of its 8,700 US stores were underperforming [1]. Earlier that year, Dollar Tree said it would close about 1,000 stores after a review of its portfolio [2]. In 2025, Starbucks closed coffeehouses that lacked "a clear path to financial performance" [3]. Stores rarely go bad overnight. Averages let them drift.

> Stores rarely go bad overnight. Averages let them drift.

Chains are built to make stores alike, and on the inputs it works. A 2025 study by BLS and Census economists finds that the shift of US retail toward large chain stores narrowed the gaps in sales per hour worked between stores, consistent with chains running many similar stores [4]. The street is the part a chain can't standardize. Stefano DellaVigna and Matthew Gentzkow found that most US food, drug and mass merchandise chains charge nearly uniform prices across stores, despite wide variation in demographics, competition and price sensitivity between them [5].

Store-by-store profit optimization treats each location as its own business: one trusted profit line per store, decisions on labor, price, promotions and assortment made for that store's demand, and every change measured against the store's own starting point. AI retail analytics is what makes it affordable below the flagship. Used well, it finds the decision that pays at each store. Used badly, it produces 40 confident dashboards and *no decisions*.

**Key figure:** $16m, yearly profit the median US chain gives up by pricing every store alike [5].

## 2. Chains average because deciding 40 times used to cost too much

Why would a chain leave $16 million a year on the table? DellaVigna and Gentzkow weighed the usual explanations and concluded that the most promising one is that uniform pricing reduces managerial overhead [5]. Put plainly, decisions are expensive. Every store-level decision needs someone to pull the data, build the case, get it approved, change the system and check that it happened. Multiply that by 40 stores and 52 weeks, and the chain does the sensible thing: it decides once.

> Uniform pricing is mostly a cost decision. When the cost of deciding falls, the logic flips.

The physical cost was real too. A study of five large US supermarket chains measured what it took to change shelf prices: $105,887 per store per year, 0.70% of revenue and 35.2% of net margins [6]. The chain bound by item pricing laws changed prices two and a half times less often than the others [6]. Expensive changes mean fewer changes, and fewer changes mean one answer for every store. That study is from 1997. The habit it describes is still with us.

Both costs are now falling fast. Walmart says roughly 2,300 of its US stores already use digital shelf labels, the rest are due within a year, and price work that took multiple associates days now takes minutes [7]. Stanford's AI Index found that the inference cost of a system performing at the level of GPT-3.5 dropped over 280-fold between November 2022 and October 2024 [8]. The analyst hour that used to go only to the flagship can now go to *every store, every week*.

## 3. Labor is the biggest store cost, and the most averaged

Look at where a store's money goes after the cost of goods. In the Census Bureau's 2022 survey, payroll was 44.1% of US retailers' operating expenses, with employee benefits another 8.2%. Rent for stores and offices was 6.6% [9]. Labor is the biggest cost a chain controls, and Wharton researchers describe the usual reflex: when sales are down and the quarterly report is due, cut labor [10]. When the cut arrives as one ratio of hours to sales, every store gets the same answer.

That reflex assumes every store responds to staff the same way. Marshall Fisher, Santiago Gallino and Serguei Netessine studied a large specialty retailer and found that the impact of staffing on revenue "varies greatly by store" [11]. The stores that gained most from extra hours had high potential demand, competition close by and experienced store managers. Adjusting staff levels in 168 stores for six months raised revenue 4.5% and annual profit by nearly $7.4 million, after paying for the added labor [10][11].

How hours are scheduled matters as much as how many. In a randomized experiment at Gap, 19 of 28 stores moved staff to more stable, predictable schedules. Median sales rose 7% and labor productivity 5% [12]. A chain-wide labor ratio can't see any of this. It takes hours from the stores where hours pay and leaves them at stores where they don't. Each store needs its own answer, by daypart, and the answer changes with the season.

US retail operating expenses by type, 2022 (excluding goods bought for resale) [9]:

|  | Value (%) |
|---|---|
| Payroll | 44.1 |
| Employee benefits | 8.2 |
| Rent for stores and offices | 6.6 |
| Depreciation and amortization | 5.5 |
| Electricity | 1.6 |

## 4. Promotions and markdowns are planned for the chain and paid for by each store

Promotions look like growth because sales go up. The question is which sales. In a study of every promotion CVS ran in 2003, only about 45% of the gross lift was incremental, and more than 50% of promotions were not profitable [13]. A field test found that ending promotions in 15 of the weakest categories would cut sales by about $7.8 million and raise profit by about $52.6 million [13]. Fewer sales, more money.

> CVS estimated that ending promotions in 15 weak categories would cost $7.8 million in sales and add $52.6 million in profit.

Markdowns carry the same problem in reverse. A chain-wide 30% markdown clears stock at the store where it was stuck and gives away margin at the store where it was still selling. Zara replaced manual clearance pricing with a forecasting model feeding a price optimizer, and in a controlled field experiment across its Belgian and Irish stores, clearance revenue rose about 6% [14]. IHL Group puts the global cost of out-of-stocks and overstocks at $1.7 trillion, 6.2% of retail sales, and counts any discount above 25% as an overstock [15].

Assortment is the quiet third lever. Most retailers, Marshall Fisher and Ramnath Vaidyanathan note, used the same assortment in every store, trimmed for smaller ones. They built a method that lets assortments vary by store, with a cap on how many versions a chain has to manage. Where retailers implemented their recommendations, sales rose 5.8% in tires and 3.6% in car care products [16]. The authors call those significant next to a typical year of comparable store growth.

## 5. What AI changes: a decision at store 37 finally pays for itself

Strip away the vendor language and AI changes one number: the cost of a good decision at a small store. The work that once paid only at the flagship (a staffing study, a promotion post-mortem, a price check against the street) can now run for every location, every week. An agent can read a store's sales, hours, stock and reviews, propose one change, check it against the chain's rules, and wait for a yes from the person who owns that decision.

> The scarce thing was never data. It was the hour someone spent deciding for one store.

Retailers are moving. In NVIDIA's 2026 survey of retail and consumer goods companies, 91% said they were using or assessing AI, and 47% were using or assessing agentic AI [17]. Read the headline results with care: 89% said AI raised their annual revenue, a self-reported answer in a survey run by a company that sells AI chips [17]. The useful question for a chain is narrower. Which decisions get better when they are made per store?

Three, mostly. First, forecasts where the signal is strong: in the M5 competition on Walmart sales data, the winning method beat the best statistical benchmark by 22.4% overall [18]. Second, small experiments, because a chain with 40 stores is a laboratory. Kohl's ran a controlled test of opening its stores an hour later and found no meaningful decline in sales [19]. Third, decisions with receipts, so every store builds a record of what was tried there and what happened.

## 6. Where store-level AI breaks: bad counts, thin data and flattering math

Start with the data. In one retailer's 37 stores, researchers checked nearly 370,000 inventory records and found 65% inaccurate [20]. Shrink cost US retailers $112.1 billion in 2022, 1.6% of sales [21]. An agent that trusts the system count will promote an item the shelf doesn't have. Before any store-level decision, the stock, sales and costs behind it need a source and a date, and gaps need to stay visible as gaps.

Then the noise. The M5 gains that look so strong at the top shrink at the bottom. The average gain of the top 50 methods over the benchmark was about 40% at the top level of Walmart's hierarchy and about 3% at the product, product-state and product-store levels [18]. A single item in a single store sells too few units to forecast well. Make store decisions where the signal lives: the store by week, the category by store.

Then the math that flatters. Pick your ten worst stores, change something, and they will likely improve, partly because extreme results drift back toward the average on their own [22]. A before-and-after at one store can't separate your change from a new competitor or a wet month. At CVS, only about 45% of promotional lift was incremental [13], and spillovers run across stores and channels. Compare changed stores with similar unchanged stores, and randomize which stores go first.

Finally, most ideas don't work. At Google and Bing, only about 10% to 20% of experiments generate positive results [23]. A store-level AI that reports every action as a win is broken. Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027, citing escalating costs, unclear business value or inadequate risk controls [24]. The tools worth keeping will be the ones that say "inconclusive" and "do nothing" out loud.

| Numbers that flatter | Numbers worth trusting |
|---|---|
| Gross lift from a promotion | Incremental profit after cannibalization |
| One store, before and after | Changed stores against similar unchanged stores |
| The worst stores improving next quarter | Results checked for regression to the mean |
| Every action reported as a win | Inconclusive and harmful results kept on record |
| Forecast accuracy for the whole chain | Accuracy at the level where you decide |

## 7. Before closing the bottom quartile, check what the four-wall P&L misses

Closures are the bluntest store-level decision, and 2024 and 2025 were full of them [1][2][3]. Four-wall profit is the right place to start: store sales minus cost of goods, store labor, rent and other store costs, before corporate overhead. It answers one question well. Does this location pay for itself? What it can't see is everything the store does outside its walls.

> Closing a store cut online sales in its trade area by 11.5%. The four-wall P&L never sees it.

ICSC studied 2,103 stores of 69 retailers from 2019 to 2022. Opening a store raised the retailer's online sales in the surrounding trade area by an average of 6.9%, and closing one cut them by 11.5% [25]. E-commerce was 17.1% of US retail sales in the second quarter of 2026 [26], so stores still carry most of the business and help carry the online part too. A store that looks unprofitable inside its walls may be holding up the brand's web sales in its town.

Bottom-quartile lists are also unstable. Some of this year's worst stores will rise next year without help, the same regression to the mean that flatters turnarounds [22]. Others are weak for a fixable reason: understaffed in the busy hours [11], the wrong assortment for the street [16], or missing from the searches nearby shoppers run. The order that protects profit is to fix what's fixable, test it against similar stores, and only then decide.

## 8. Where store-level profit management is heading by 2030

This section is our view, and we label it that way. Within five years, changing a price, a schedule or a display at one store will cost close to nothing, and the hard part will be deciding well and governing the change. Walmart says its prices are the same for all customers in any given store, regardless of demand, time of day or who is shopping [7]. When changes get cheap, the discipline around them becomes the product.

The line that matters will run between the store and the shopper. Localizing by store is ordinary retail. Personalizing prices by shopper is drawing regulators. Since November 2025, New York has required the label "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA" on such prices [27]. The FTC has studied how intermediaries use personal data to shape the prices people see [28]. When Wendy's mentioned dynamic pricing in 2024, the backlash was loud enough that it said it "will not implement surge pricing" [29].

Our bet: the chains that win set one price per store, for everyone who walks in, tuned to that store's street. It is also a fairer default than it sounds. DellaVigna and Gentzkow estimate that uniform pricing raises the prices poorer households pay relative to richer ones [5]. Store-level pricing done in the open, with margin floors and named approvers, can serve a neighborhood better than one national price list or one price per person.

Two more shifts follow. Agents will do most of the store-level analysis, and people will own every consequential yes. Gartner warns that governing every AI agent the same way, either locked down or fully trusted, is the root cause of failure [30], so approvals should scale with risk. And the asset that compounds will be each store's decision log: what was tried, why, and what happened. Software can be bought. *Five years of evidence about store 37 cannot.*

| When | What |
|---|---|
| Nov 2025 | New York requires a label on prices set by algorithms using personal data. |
| 2026 to 2027 | Walmart takes digital shelf labels chain-wide, with one price for every shopper in a store. |
| End of 2027 | Gartner expects over 40% of agentic AI projects to be canceled. |
| 2028, our view | Weekly store P&Ls, with forecasts and ranges, become normal at chains of 10 or more stores. |
| 2030, our view | The edge belongs to chains with years of evidence about what worked at each store. |

## 9. What a 40-store chain should do this year

Build one profit line per store that your CFO and your district managers both believe. Use the same allocation rules everywhere, close it monthly at minimum, and mark incomplete periods instead of filling them in. Then rank stores two ways: by four-wall profit, and by change against each store's own prior year. The second list is where the decisions are, because it separates a weak location from a good store that is slipping.

> Rank stores two ways: by four-wall profit, and by change against their own last year.

Pick three decisions you make chain-wide today and should make per store. Our shortlist: labor hours by daypart, which promotions each store runs, and local price or assortment exceptions inside brand rules. Write the guardrails before anything runs: a margin floor, the biggest discount allowed, minimum staffing, and who approves what. A store manager should be able to read each rule in one sentence and know whether a change passes.

Test before rollout. Take ten similar stores, pick five at random to go first, fix the measurement window in advance, and compare both groups against their own baselines. Gap's scheduling study used the same idea, with a randomized group of 19 of 28 stores [12]. Expect most ideas to fail, keep the inconclusive results, and roll out only what survives. Above all, keep the log, store by store.

When you buy AI for this, ask four questions. Who owns the data and the logic if you leave? Gartner expects 70% of enterprises to abandon agentic AI built by a vendor's forward-deployed engineers by 2028, trapped by costs and unable to evolve it on their own [31]. Do approvals scale with risk? Does it ever recommend doing nothing? And does it measure each store against its own starting point, with the uncertainty shown?

1. **One trusted P&L per store.** Same allocation rules everywhere, closed monthly at minimum. Missing periods are named, never filled with guesses.
2. **Pick three store decisions.** Labor hours by daypart, promotion participation, and local price or assortment exceptions inside brand rules.
3. **Write the guardrails.** Margin floor, biggest discount, minimum staffing and who approves what, each one readable in a sentence.
4. **Test on half first.** Randomize which similar stores go first. Fix the window in advance. Compare both groups against their own baselines.
5. **Keep the decision log.** What was tried at each store, why, and what happened, including the inconclusive and harmful results.

## 10. How Ocally helps

Shreyas Kulkarni, Ocally's founder, saw this problem while supporting large retail players at Google. Every store was different and could do more, but budgets went to flagship stores or were averaged out in regional and national plans, leaving money on the table. Ocally gives every store that flagship attention. Kai, Ocally's growth agent, learns each store's street, shoppers, stock and numbers. He starts by reconciling each store's revenue and expenses by period from your POS, accounting exports or uploaded files, and names missing or conflicting periods instead of guessing.

Each store gets five goals, from being easier to find to selling more, judged on that store's net profit once three complete periods are in. A decision engine can act, collect more evidence first, or decide to do nothing. Every action gets a measurement window before it runs and a verdict after: positive, neutral, harmful or inconclusive, against the store's own starting point. Harm stops the work. Nothing changes a price, spends money, goes public or messages a customer without your yes. Each store's profit model is scored on its own forecasts.

- **[One profit line per store](https://ocally.co/platform.md):** Revenue and expenses by period, reconciled per store from your POS, accounting exports or files. Every number keeps its source, and incomplete periods are held back and named.
- **[Prices and offers, checked per store](https://ocally.co/capabilities/pricing.md):** Kai spots items priced differently across stores and shows each offer's margin before you approve it. No offer runs below the margin floor you set.
- **[Stock counts you can act on](https://ocally.co/capabilities/inventory.md):** Kai keeps a stock count per store, flags what will run out or sit unsold, and suggests what to reorder or move before it turns into a markdown.
- **[Store routines with proof](https://ocally.co/capabilities/store-operations.md):** Kai writes each checklist, gives every store its work on its own clock and checks the proof, so a plan made for a store happens in that store.
- **[Two people beside your team](https://ocally.co/consultation.md):** A growth engineer connects your systems. A growth specialist checks Kai's work and helps your store managers and staff move to AI-native ways of working.

## Questions

### What is AI retail analytics?

AI retail analytics uses machine learning and AI agents to turn a retailer's sales, costs, staffing, stock and local signals into decisions. For a chain, the useful version works per store: it forecasts demand where the data supports it, proposes a change for that store, checks it against guardrails such as margin floors, and measures the result against the store's own baseline and similar stores.

### What is four-wall (4-wall) EBITDA?

Four-wall EBITDA is a store's earnings before interest, taxes, depreciation and amortization, counting only what happens inside its walls: store sales minus cost of goods, store labor, rent and other store operating costs, before corporate overhead. It shows whether a location pays for itself. It misses spillovers, such as the online sales a store drives in its trade area.

### What are same store sales, and what do they hide?

Same store sales compare sales at stores open in both periods, usually a year apart, so openings and closings don't distort growth. They are the right number for the chain and a weak one for a store: a 2% chain gain can hide a quarter of stores shrinking, and sales alone say nothing about whether each store made money.

### What are the most important retail KPIs for a chain?

Start with four-wall profit per store, by period. Under it, track what moves profit at that store: sales and gross margin, labor cost against sales by daypart, stock accuracy and out-of-stocks, markdown and promotion cost, and how easily nearby shoppers find the store. Judge each store against its own baseline and against similar stores, rather than the chain average.

### How can a franchise brand improve profitability at each location?

Give every location a four-wall P&L its owner trusts, then change one decision at a time per location: staffing by daypart, local promotions, and assortment or price exceptions inside brand rules. Test changes on some locations before rolling them out, and keep a shared record of what was tried where. Brand standards stay central. Local decisions follow each location's evidence.

### How should a chain decide markdowns and promotions store by store?

Count only incremental profit: the extra units a promotion sells at that store, minus the margin given away, minus sales pulled from other products or later weeks. Run the promotion where that math works and skip it where the item still sells at full price. Test it on some stores first, and compare them with similar stores that did not run it.

## Sources

1. [Supermarket News: Walgreens plans to close 'significant number' of stores (June 2024)](https://www.supermarketnews.com/nonfood-pharmacy/walgreens-plans-close-significant-number-stores)
2. [Supermarket News: Dollar Tree plans 1,000 store closures (March 2024)](https://www.supermarketnews.com/retail-financial/dollar-tree-plans-1000-store-closures)
3. [Starbucks Corporation: Form 8-K on store closures and restructuring, SEC filing (September 2025)](https://www.sec.gov/Archives/edgar/data/829224/000082922425000067/sbux-20250923.htm)
4. [US Bureau of Labor Statistics: National Chains and Trends in Retail Productivity Dispersion, Working Paper 570 (September 2025)](https://www.bls.gov/osmr/research-papers/2023/pdf/ec230100.pdf)
5. [Quarterly Journal of Economics / NBER: DellaVigna and Gentzkow, Uniform Pricing in US Retail Chains (November 2019)](https://www.nber.org/papers/w23996)
6. [Quarterly Journal of Economics: Levy, Bergen, Dutta and Venable, The Magnitude of Menu Costs: Direct Evidence from Large US Supermarket Chains (August 1997)](https://ideas.repec.org/a/oup/qjecon/v112y1997i3p791-824..html)
7. [Walmart: How the shelf got smarter and our jobs got easier (March 2026)](https://corporate.walmart.com/news/2026/03/02/how-the-shelf-got-smarter-and-our-jobs-got-easier)
8. [Stanford HAI: The 2025 AI Index Report (April 2025)](https://hai.stanford.edu/ai-index/2025-ai-index-report)
9. [US Census Bureau: Annual Retail Trade Survey, Detailed Operating Expenses 2022 (January 2024)](https://www.census.gov/data/tables/2022/econ/arts/operating-expenses.html)
10. [Knowledge at Wharton: Retail's big mistake: slashing payroll cuts into profits (February 2019)](https://knowledge.wharton.upenn.edu/article/retail-staffing-model/)
11. [Manufacturing and Service Operations Management: Fisher, Gallino and Netessine, Setting Retail Staffing Levels: A Methodology Validated with Implementation (November 2021)](https://ideas.repec.org/a/inm/ormsom/v23y2021i6p1562-1579.html)
12. [University of Chicago News: Stable scheduling increases sales and employee productivity, study finds (March 2018)](https://news.uchicago.edu/story/stable-scheduling-increases-sales-and-employee-productivity-study-finds)
13. [Marketing Science: Ailawadi, Harlam, Cesar and Trounce, Quantifying and Improving Promotion Effectiveness at CVS (July 2007)](https://ideas.repec.org/a/inm/ormksc/v26y2007i4p566-575.html)
14. [Operations Research: Caro and Gallien, Clearance Pricing Optimization for a Fast-Fashion Retailer (December 2012)](https://ideas.repec.org/a/inm/oropre/v60y2012i6p1404-1422.html)
15. [IHL Group: Fixing Inventory Distortion, 2026 study (2026)](https://www.ihlservices.com/product/inventory-distortion/)
16. [Management Science: Fisher and Vaidyanathan, A Demand Estimation Procedure for Retail Assortment Optimization with Results from Implementations (2014)](https://doi.org/10.1287/mnsc.2014.1904)
17. [NVIDIA: State of AI in Retail and CPG survey (January 2026)](https://blogs.nvidia.com/blog/ai-in-retail-cpg-survey-2026)
18. [International Journal of Forecasting: Makridakis, Spiliotis and Assimakopoulos, M5 accuracy competition: Results, findings, and conclusions (October 2022)](https://ideas.repec.org/a/eee/intfor/v38y2022i4p1346-1364.html)
19. [Fortune: Big ideas from companies that experiment (November 2019)](https://fortune.com/2019/11/17/big-ideas-companies-experimentation/)
20. [Management Science: DeHoratius and Raman, Inventory Record Inaccuracy: An Empirical Analysis (April 2008)](https://ideas.repec.org/a/inm/ormnsc/v54y2008i4p627-641.html)
21. [National Retail Federation: Shrink accounted for over $112 billion in industry losses in 2022 (September 2023)](https://nrf.com/media-center/press-releases/shrink-accounted-over-112-billion-industry-losses-2022-according-nrf)
22. [International Journal of Epidemiology: Barnett, van der Pols and Dobson, Regression to the mean: what it is and how to deal with it (February 2005)](https://academic.oup.com/ije/article/34/1/215/638499)
23. [Harvard Business Review: Kohavi and Thomke, The Surprising Power of Online Experiments (September 2017)](https://hbr.org/2017/09/the-surprising-power-of-online-experiments)
24. [Gartner: Gartner predicts over 40% of agentic AI projects will be canceled by end of 2027 (June 2025)](https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027)
25. [ICSC: The Halo Effect III: Where the Halo Shines (December 2023)](https://icsc.com/news-and-views/icsc-exchange/new-icsc-study-the-halo-effect-iii-where-the-halo-shines-quantifies-how-physical-stores-drive-online-sales)
26. [US Census Bureau: Quarterly Retail E-Commerce Sales, 2nd Quarter 2026 (August 2026)](https://www.census.gov/retail/ecommerce.html)
27. [Jones Day: New York's novel algorithmic pricing disclosure law takes effect (November 2025)](https://www.jonesday.com/en/insights/2025/11/new-yorks-novel-algorithmic-pricing-disclosure-law-takes-effect)
28. [Federal Trade Commission: Surveillance pricing update and the work ahead (January 2025)](https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2025/01/surveillance-pricing-update-work-ahead)
29. [FOX 9: Wendy's faces backlash after 'surge pricing' reports, denies claims (February 2024)](https://www.fox9.com/news/wendys-backlash-surge-pricing-reports-denies-claims)
30. [Gartner: Applying uniform governance across AI agents will lead to enterprise AI agent failure (May 2026)](https://www.gartner.com/en/newsroom/press-releases/2026-05-26-gartner-says-applying-uniform-governance-across-ai-agents-will-lead-to-enterprise-ai-agent-failure)
31. [Gartner: 70% of enterprises will abandon agentic AI built by vendor forward-deployed engineering by 2028 (September 2026)](https://www.gartner.com/en/newsroom/press-releases/2026-09-29-gartner-predicts-70-percent-of-enterprises-will-abandon-agentic-ai-built-by-vendor-forward-deployed-engineering-by-2028)

**See where each of your stores stands.** Bring your store list and a recent period of results to a free consultation. We'll show you where Ocally would start, store by store. [Book a free consultation](https://ocally.co/consultation.md).

## Next steps

- [Is Ocally right for this business?](https://ocally.co/evaluate.md): who it fits and who it does not, the cost, the risks, and how to start.
- [Pricing](https://ocally.co/pricing.md): customized, with a minimum of $500 per location per month.
- [Book a free consultation](https://ocally.co/consultation.md): 30 minutes on one store. Calendar: https://cal.com/ocally/30min.
- [See if we're a fit](https://ocally.co/fit.md): a two-minute check; an agent can submit it for its human, with consent.
- [Every doc](https://ocally.co/llms.txt): the index.

Questions a doc doesn't answer: onboard@ocally.co.