Competitive intelligence for physical stores

Monitor your stores and your competitors’, side by side.

Built from public data, with research method behind every number: hourly busyness, reputation and what customers write, store by store.

Illustrative data. The examples play on their own: hover to read at your pace.

The job

Know how each physical store is doing. And how the store next door is doing.

Every chain asks this. The answer tends to arrive late, in scattered files, and the comparison with competitors rests on the impression of whoever visited the store.

Subject

Physical store performance

Traffic through the day, Google rating, what customers praise and complain about, and how it all compares with competitors in the same market.

How it is done today

Visits, spreadsheets and manual reading

It still relies on field visits, mystery shoppers and reviews read one by one. Each market sends a file, and someone merges everything into a spreadsheet at month end.

The pain

Data arrives stale and without a common yardstick

The checkout queue complaint reaches the report weeks after it became a 1-star review. Saturday’s peak is known by hearsay. And nobody can say, with numbers, where the store loses to the one next door.

Where Cassi.ai comes in

A dashboard read with method

Rival Pulse reads public store listings, turns charts and text into numbers, classifies what customers write and delivers it all in one dashboard with the brands side by side.

Decisions first. Charts second. Every block in the dashboard exists because it answers an operational question.
What the dashboard answers

Operational questions, with the competition on the same screen.

Every filter applies to every block: weekday, location, stars, topic and review period. And every chart always shows the brands compared.

On which day and hour do I need more checkout staff? Busyness 7 days × 18 hours

Hourly busyness for each store, each day’s peak hour, the three busiest hours and the quietest one. Heatmap by brand and an hour-by-hour comparison curve.

Is my rating really dropping? Reputation by period

Weighted rating per brand, star distribution, trend from the oldest period to the latest, and how often each chain replies to negative reviews.

What do customers complain about, and at which store? Voice of the customer 48 topics

How much people talk about service, queues, price, product, exchanges, out of stock items and store layout. When a topic comes up, how much of it is a complaint. And the reviews side by side, with key terms highlighted.

Where do I lose to the competitor in the same neighborhood or mall? Stores and locations location matrix

Each store’s rating, location by location, with the winner marked. Brand presence by location, store map, today’s hours and contacts.

What if I want to ask in my own words? AI analyst beta

A free-text question, answered only with the numbers and excerpts in the filtered selection. The answer compares the brands and flags small samples.

Methodology

Six steps between the public listing and the number on the dashboard.

Collection runs without a language model, on explicit and verifiable rules. AI only comes in at the end, to converse with data that has already been checked.

01

Discovery

Searches by brand and city with several phrasings, so no store is left out.

each store identified by its listing’s unique code, no duplicates
02

Structured reading

Address, hours, phone, stars, topics and the popular times chart. The chart becomes a number: busyness percentage by hour and day.

the value published with each bar is read directly, with no image interpretation
03

Validation

Municipality checked against the postal code, since listings sometimes name the neighboring city. Out-of-scope stores are dropped before any collection is spent.

source access limit: collection pauses and the store is marked partial
04

Standardization

Brand rating computed from the star counts of all reviews, without the listing’s rounding. Each store is weighted by its volume.

average busyness only for hours when half the stores are open
05

Text classification

Every review with text gets one or more topics. The dashboard shows how many mentions back each percentage.

topics with fewer than 10 mentions get no percentage
06

Comparison and analysis

Filters recompute everything with the brands side by side. The AI analyst receives only the selection on screen.

ties never become rankings · sample always visible

About step 06. The analyst works with retrieval from your own data: before answering, it receives the numbers and review excerpts that pass the chosen filters, and answers using only those. So every answer can be checked against the dashboard and never depends on what a generic model thinks it knows about your chain.

Principles

What the dashboard never does.

Research rules applied to public data. They appear on screen next to each number, so readers know where it came from.

Sample

Never hides the base. Every percentage shows how many reviews or stores back it, and the dashboard warns when a filter swaps the source total for the collected sample.

Privacy

Never shows reviewer names. The review text serves the analysis. The writer’s identity stays out of the published data.

Comparison

Never compares stores with different audiences without saying so. The location matrix puts stores in the same mall or neighborhood side by side.

Source

Never works around access limits. When the source restricts reading, collection stops, the store is marked partial and resumes later.

Ranking

Never invents a winner. Best and worst badges only appear when there is a real gap between brands.

Dates

Never presents an approximate date as exact. The period comes from the age the review itself reports, and the dashboard says so.

Verticals

Ready for any vertical with physical stores.

Collection, topics and dashboard adapt to the sector. Fashion and wholesale already run as pilots. For any other vertical, a project starts from the list of brands and cities and the topics that matter to that client.

Live pilot

Fashion and apparel

fitting rooms, exchanges, store credit

Live pilot

Wholesale clubs and supermarkets

checkout queues, shelf prices, produce

Pharmacies and drugstores

out of stock drugs, counter service, late hours

Restaurants and fast food

wait times, wrong orders, drive-thru

Coffee shops and bakeries

morning rush, queues, freshness

Gyms

busyness by hour, equipment, cleanliness

Banks and branches

waiting, service, ATMs

Car dealers and repair shops

service time, quotes, after-sales

Gas stations and convenience

price, convenience store, restrooms

Clinics and labs

waiting, scheduling, test results

Home improvement

stock, delivery, technical advice

Electronics and appliances

warranty, installation, after purchase

Pet stores and vets

grooming, scheduling, care

Shopping malls

daily traffic, parking, tenant mix

Hotels

check-in, cleanliness, breakfast

Carrier stores

queues, device swaps, number porting

Your vertical

another sector with physical stores? we set up brands, cities and topics with your team

Who it is for

People who make physical store decisions with data from outside their own company.

The same collection works for other chains, other sectors and other cities. The dashboard design starts from each team’s decisions.

Retail chains and franchises Shopping malls Expansion and site selection teams Operations and staff scheduling Trade marketing and market intelligence Research agencies and consultancies
Questions and answers

What people ask before opening the dashboard.

What is Rival Pulse?

Rival Pulse is a competitive intelligence dashboard for physical stores. It reads public store listings, turns charts and text into numbers, and shows hourly foot traffic, reputation, and review topics with the brands side by side on the same screen.

Where does the data come from?

From public store listings, collected with the date and method on record. Client databases never enter it, personal data from reviewers never enters it, and reviewer names stay out of the published collection.

How is hourly foot traffic measured?

Through structured reading of each store's public popular-times chart, across seven days by eighteen hours. The value published with each bar is read directly, without interpreting an image, and the hourly average only uses hours when at least half of the stores are open.

Does the dashboard measure sales or revenue?

No. The dashboard measures foot traffic, reputation, and what customers write. It does not estimate sales, does not estimate revenue, and does not turn traffic into money.

Can it compare a competitor in the same location?

Yes. The location matrix puts stores in the same mall or neighborhood side by side, and the dashboard says so when the compared stores serve different audiences.

Which sectors does the dashboard serve?

Any vertical with physical stores. The active pilots are fashion and apparel and cash-and-carry and supermarkets, and the same method is already designed for pharmacy, restaurant, gym, bank, fuel station, clinic, pet shop, shopping mall, hotel, and carrier store, among others.

What method guarantees does the dashboard give?

It shows the base behind every number, says when a filter swaps the source total for the sample it read, does not crown a winner when the difference is not real, does not present an approximate date as exact, and stops collecting when the source restricts access, marking that store as partial.

Where does AI come in?

Only at the end. Collection runs without a language model, on explicit and verifiable rules. The AI analyst receives only the numbers and review excerpts that pass the chosen filters and answers from those, which keeps the answer checkable against the dashboard.

How do I get access?

Through the access request form on this page. The dashboard is restricted to clients with an active project, and the vertical, brands, cities, and topics are set up with the client's team.

Physical stores leave a public trail every day. Those who read it with method decide first.

If your chain still merges field visits in spreadsheets, it is worth measuring how many days pass between the first Google complaint and the in-store action plan.

Products available as SaaS on credits, per project, or as a platform purchase. Talk to us, we have a format that fits your budget.

Talk to Cassi.ai

Cassi.ai is a software engineering company focused on the pain points of research, innovation and insights. Twenty years in market research, former ESOMAR Brazil, with talks at IIEX, ESOMAR and ABEP.

Pedir acesso ao Rival Pulse

Conte quem você é e qual vertical quer acompanhar. A Cassi.ai responde por e-mail.