AI Systems · Ref. OI-02

An owner shouldn't have to assemble reports to understand what is happening in the chain

Owner Intelligence is a management loop for a chain: POS, loyalty, delivery, costs and reputation come together in one pane. A digest arrives every morning, per-location anomalies find you on their own, and an advisor works on top of the data — explaining the numbers in plain words and proposing decisions on menu margin, load, control and marketing.

2 minMorning readinstead of an hour of assembling numbers
12Modules in one panefrom revenue to the fraud radar
16+Locations in a live loopmulti-currency, consolidated into euro
24/7Anomalies find younot you finding them a week later in a report

— 01 / What breaks in a chain

Management attention is the owner's scarcest resource — and it is spent assembling spreadsheets

The problem isn't missing data. The data exists — in the POS, in loyalty, at the aggregators, in the bank, in managers' chats. Nobody consolidates it by morning and nobody explains it.

To understand how things are going you ask three people and open five spreadsheets

One pane: revenue, checks, average check, locations, channels, categories — for any period, against the previous one

The report arrives a week late and no longer decides anything

Data loads itself; the digest is in your messenger before opening hours — for yesterday

The problem is visible once it is already in the P&L

A location's deviation is measured against its own median for that weekday and arrives as an alert the same day

Discounts, voids, deletions and bonus giveaways live outside your field of view

Control and fraud radar: the norm is computed on your own data, deviations are shown by operation, guest and shift

The menu is discussed by taste, not by margin

Every item plotted on demand × margin, with a concrete action: hold, raise price, promote, revise

Aggregators take their share and nobody calculates how much exactly

Commissions at your contracted rates, city by city, and the amount leaving through each channel

An analyst costs a full-time salary and still explains the numbers once a month

The advisor works on your numbers continuously, answers a question in text, and translates jargon into plain language

— 02 / Who buys this

One loop — four different answers to «why»

Management data is needed by four functions at once, each in a different cut. Here is what changes for each, and the metric that shows it.

Owner · founder

Today

To understand how things are going, they ask three people and still aren't sure.

With the loop

A morning read and signals, unprompted. A question in plain words instead of a task for an analyst.

Metric

Hours of attention per week and time from event to decision

CEO · COO

Today

Accountable for a result they see a week late, in a different format for every country.

With the loop

One picture across all locations, norms computed on your own data, and anomalies that arrive on their own.

Metric

Reaction speed to a deviation and share of locations performing at their own norm

CMO · marketing director

Today

Channels are judged by spend and turnover. What a channel yields after commissions stays unanswered.

With the loop

Revenue and margin by channel, real aggregator commissions at your contracted rates, promo effect against the weekday norm, loyalty and churn.

Metric

Channel contribution to margin rather than turnover; retention and visit frequency

Commercial director · service director

Today

Menu, prices and shifts are argued from taste and experience, because the data isn't at hand.

With the loop

Demand × margin per item with a ready action, hourly load, cost, and operations control.

Metric

Basket margin, food cost against the median, and losses to discounts and write-offs

— 03 / Who it is for

Chains where decisions cost money and data is scattered across systems

The loop is built where there are several locations, guest flow, and several data sources: POS, loyalty, delivery, costs. The more locations and jurisdictions, the larger the effect — because that is exactly the configuration in which an owner stops seeing the whole business.

Restaurant and coffee chainsQSR, street food, food courtsDelivery and dark kitchensRetail and grocery chainsBeauty salons and clinicsFitness chainsFuel stations and serviceFranchises and multi-brand holdings

Markets and currencies

Moldova, Romania, Ukraine are the priority markets. Multi-currency by default: Moldovan leu, Romanian leu, hryvnia, forint and euro are consolidated into one comparable picture when the chain operates across countries and each jurisdiction keeps its own POS separately.

— 04 / Chain of events

An owner's day with the loop

Not «a dashboard you have to log into», but a sequence that reaches you on its own. You log in only when you want to dig.

Overnight

Data is pulled

The loop collects checks, payments, channels, loyalty and platform ratings, computes weekday medians and looks for deviations. Nothing to click.

09:00

Digest

Yesterday: revenue, checks, average check, comparison with the norm for that weekday, best and weakest location, top items. One message instead of a round of managers.

09:02

Signals

«This location is 37% below its own norm», «average check up 12% at another» — each line links to the location breakdown: dynamics, hours, items, history.

Any time

A question in words

«Why did revenue drop in this city last week?» — an answer with numbers, broken down by channel and location. No SQL, no exports, no analyst.

Weekly

Structural review

What changed structurally, not for a day: channels, menu, hourly load, loyalty, reputation. Recommendations in money: «food cost 33% against a 26% median — roughly N per month overpaid».

Monthly

P&L and decisions

P&L by location, margin, a profitability ranking of locations, where the loss is structural and where it is one-off. Exactly the cut usually prepared for a board meeting.

Continuously

Control

Fraud radar and giveaways: suspicious bonus accruals and redemptions, abnormal discounts and deletions, operations outside working hours — linked to guest and shift.

— 05 / What's inside

Twelve modules in one pane

The composition is configured per chain: we connect what has data; the rest is switched on later without rebuilding the loop.

01

Overview

Key metrics with trend, revenue by day, best and weakest day, share by location and city.

02

Countries and cities

Markets compared in one currency, with the channel structure inside each — you immediately see who lives on dine-in and who on aggregators.

03

Channels

Dine-in, pickup, own delivery, aggregators, in-app payments. Separately — how much leaves as commission at your rates.

04

Menu and margin

Demand × margin per item, an action for each group, a separate card for add-ons and sauces, category shares inside kitchen and bar.

05

Forecasts

Expected revenue and load based on the location's history and the weekday — so shifts and purchasing aren't set on gut feel.

06

Anti-fraud

Accrual and redemption norms computed on your own data, and deviations from them: not «suspicious in general» but «above your own norm».

07

Fraud radar

Operations outside working hours, checks paid entirely with bonuses, repeating patterns around one guest or one shift.

08

Operations control

Deletions, voids, discounts, operations with no stated reason — with amounts and locations. Plain language instead of POS jargon.

09

Reputation

Ratings by platform and location, dynamics, loyalty index. Docks with Reputation Loop into a single contour.

10

Load

Hour × day heat map, actual opening and closing times from the first and last check, peak hours per location.

11

Loyalty and guests

RFM segments, cohorts, churn, guest card, bonus transactions for the period. Staff cards with no real visits are visible separately.

12

Cost and stock

Food cost and margin per item, deviations by location, candidates for recipe and purchasing review.

— 06 / Alternatives

Three ways to get the management picture

Excel is honest but late. A BI contractor builds what the brief says, then leaves. The loop lives with the chain and explains itself.

Reports and ExcelBI contractorOwner Intelligence
Who consolidates datamanagers and the accountant, by handa developer, per written briefthe loop, daily and unprompted
Speeda week, sometimes a monthdays, after each change requestnext morning
Who explains the numbersnobodyan analyst, billed separatelythe advisor in the dashboard and in your messenger, in your language
SourcesPOS and spreadsheetswhatever got connected during the projectPOS, loyalty, delivery, costs and reputation in one loop
Anomaliesnoticed through revenueonly if specified in the briefcomputed against the location's own norm and delivered to you
Cost of growthnew location — new spreadsheetnew location — new sprintnew location — a line in the configuration
Multiple countriesconsolidated into euro by handrare and expensivemulti-currency and multi-jurisdiction by default
Who owns itit lives in people's headswith the contractor, along with the codeyour server, your data, documentation handed over

— 07 / Economics

Counted in the open: where the money actually sits

An illustrative calculation with visible assumptions. Every input is listed separately: substitute yours and see whether it holds.

Assumptions

Chain size
10 locations
Chain turnover
€400,000 per month
Owner's attention
1 hour a day assembling numbers
Typical deviation
one location 15% below its norm

One hour a day of the person at the top

a quarter of working time spent assembling numbers rather than deciding on them

≈ 250 h/year

Food cost 3 points above the median at three locations

visible only once cost sits next to demand, item by item

≈ €3,600/month

A location 15% below norm, caught a month later instead of the next morning

revenue lost on a single event; there are several such events in a year

≈ €6,000

Aggregator commissions at contracted rather than averaged rates

the difference that decides where a channel is worth keeping and where it eats the margin

2–7 points
Break-even1 event

A €6,500 Full implementation pays for itself on a single deviation caught in time. The first run in a real sixteen-location chain surfaced three — not a promise of results, just what the data shows once someone connects it.

Illustrative figures at roughly €40,000 monthly turnover per location. In the review we run them on your data and say plainly which sources are missing for the layers you want.

— 08 / Proof

A loop that reaches the owner every morning

Under NDA · food service chain, four countries

16 locations, four jurisdictions, one management loop

Each country ran its POS as a separate, unlinked network; loyalty lived in a third system and the aggregators in a fourth. We built the loop: overnight data collection, a 9:00 digest, anomalies against the weekday median, P&L by location, the menu plotted on demand × margin, and a control module. The very first run surfaced three real per-location deviations; the menu breakdown showed that nearly half of revenue comes from items priced below the margin norm. Technical specifics are under NDA.

Open the case
Locations in the loop16+
Modules12
Digest09:00
Anomalies on first run3

— 09 / Pricing

From €3,500 for a loop that replaces the reporting routine

Implementation is fixed and depends on the number of sources and the depth of layers. After that: operation, or full support with a human alongside.

Core

from €3,500

implementation, one-off

Up to 10 locations, one market. POS connection, single pane, morning digest, per-location anomalies, questions in words, location breakdown. Further layers by priority.

Discuss implementation

Full

from €6,500

implementation, one-off

Multiple countries and currencies. Everything in Core plus P&L by location, menu and margin, channels and aggregator commissions, loyalty with RFM and cohorts, operations control and fraud radar, reputation.

Discuss implementation

Advisor + External CMO

€2,500

per month

The loop plus a human: a weekly review of the numbers, priorities for the month, decisions on margin, channels and marketing. This is the standard External CMO format — run on your data instead of slides.

Discuss implementation

Operation without support — €350 per month: server, models, source updates, adding locations. Not included: keeping your primary documents and bookkeeping. The loop computes what you actually record; what's missing, we'll show you in week one.

— 10 / Objections

What usually gets asked before signing

«We already have POS reports and BI»

The POS shows its own part: it doesn't know aggregator commissions, doesn't consolidate several countries into one currency, doesn't compute a weekday norm for a specific location, doesn't see loyalty or reputation, and explains nothing. The loop doesn't replace the POS or argue with BI — it connects what they don't, and adds the explanation layer.

«Our IT team could build this»

They could — collecting data isn't magic. The difference is what comes after: someone has to calibrate norms, fix sources when they change, hold the quality of explanations, and add layers. We hand over a finished loop with documentation: if IT wants to take over operation, that's available at any point.

«We don't want vendor dependency»

The loop lives on your server, the access and the data are yours, documentation is handed over. This is an implementation you own, not a subscription that switches off together with access to your own numbers.

«Guest data and commercial confidentiality»

What goes to the model are aggregates for explanations and recommendations, not personal data. Phone numbers and identifiers are hashed at collection. Access is separated: a manager sees their location, the owner sees the chain.

«The team won't use it»

Which is why the primary format is not a dashboard but a digest and alerts in the messenger the team already lives in. The dashboard is opened when someone wants to dig. If people have to be forced to use a tool, the tool is designed wrong.

«What if we don't have enough data?»

That surfaces in week one, not two months in. You get a source map and an honest list of what's missing for the layers you want: a full P&L needs costs and staffing, and if those don't exist we say so before you pay for a layer that cannot be built.

— 11 / Implementation

Four steps to the first digest

01

Week 1 · Sources

We look at what the POS, loyalty, aggregators and bookkeeping actually expose. Output: a source map and an honest list of what is missing for the layers you want.

02

Weeks 2–3 · The loop

Data collection, single pane, digest and anomalies — on your data, not on a demo. The first digest arrives before implementation ends.

03

Week 4 · Calibrating norms

We tune what counts as a deviation for you specifically: thresholds, minimum amounts, promo periods. Without this step a loop screams at campaigns and stays silent on real problems.

04

Then · Layers by priority

P&L, menu and margin, control, loyalty, reputation — switched on one at a time, in the order of your pain, not our architecture.

— 12 / Questions

What chain owners ask

Which POS and systems can be connected?+

One practical criterion: if the system exposes an API or a regular export, it connects. Already live: Syrve and iiko, loyalty programs with their own API, delivery platform storefronts, payment and QR services, cost and staffing spreadsheets. For retail and e-commerce: inventory systems and store platforms.

Is this a BI dashboard or an AI advisor?+

Two layers of one product. The lower layer is data: collection, consolidation, norms, anomalies, cuts. The upper layer is the advisor: it answers a question in words, explains the deviation and proposes an action with an amount. Without the lower layer the upper one hallucinates — which is why we don't sell «a chat on top of Excel».

Does our data leave our systems?+

The loop is deployed on your server or on one dedicated to you. What goes to the model are aggregates for explanations and recommendations, not guests' personal data. Phone numbers and identifiers are hashed at collection.

Why do we need this if the POS already has reports?+

The POS shows its own part: it doesn't know aggregator commissions, doesn't consolidate several countries into one currency, doesn't compute a weekday norm for a specific location, doesn't see loyalty or reputation, and explains nothing. The loop doesn't replace the POS — it assembles what the POS never connects.

How many locations make this worthwhile?+

Three to four and up. With one location the owner holds the picture in their head. From the third, a gap opens between what happens and what they learn — and that gap is closed by a loop, not by hiring another manager.

Who uses it besides the owner?+

The operations director, the finance lead, location managers — with different access rights. The owner sees the whole chain, a manager sees their location. Digests and alerts are configured per recipient.

What if our bookkeeping is partial and some data is on paper?+

We start with the layer that has data: revenue, checks, channels, locations. A full P&L needs costs and staffing — if those don't exist, we say so in week one, not two months into implementation.

How is this different from hiring an analyst?+

An analyst costs more as an employee, works on request, goes on holiday and takes the context along. The loop runs every morning and doesn't depend on one person. Where human interpretation and decisions are needed, we add them in the External CMO format — but on top of numbers that are already there.

Next step

We'll work out on your data which three signals the loop shows in week one

One meeting: what sources the chain has, what can be assembled from them in two weeks, and which decisions that changes in the first month. Output: a source map — whether or not we work together afterwards.