The method
How PerformHub decides.
Operations research is the use of data and maths to make better operating decisions. In a store chain that means what will sell, where stock should be, what each store should aim for and what to do today. PerformHub works through the same seven steps every time.
The seven steps
- 01
Observe
Read every receipt, stock level, online order and hour of foot flow, and make them agree.
In PerformHubA return receipt counts as a return, not a sale. A day a store never sent is rebuilt from its receipts and marked as rebuilt, and a missing day shows as missing, never as zero.
- 02
Forecast
Project sales by weekday from your own history, with sale days kept out of the norm.
In PerformHubMonth-end pace uses the last six same weekdays, weighted towards the most recent, with a damped trend. Registered promotions and unusual spikes are kept out of the normal baseline.
- 03
Allocate
Split targets and stock under your rules, with totals that add up exactly.
In PerformHubA new collection is split across stores by sales history, store grade and recent sales, with a minimum share for each grade, and every unit that arrived is placed: no more, no less.
- 04
Simulate
Try a decision on your own numbers before anything changes.
In PerformHubA sale plan is run at three strictness levels. For each discount depth you see the expected units, revenue, profit and ending stock, labelled as estimates, before a single price moves.
- 05
Explain
Every recommendation comes with its reasons, in Arabic or English.
In PerformHubAsk why a store got its target and the answer shows the forecast, its share of the total and the score behind it. Joe quotes figures calculated by the product; he does not work them out himself.
- 06
Approve
Nothing changes until a person approves it.
In PerformHubTransfer plans stay drafts until someone approves them. A published target plan is never edited, only replaced by a new version. A Shopify price change runs as a dry run first and can be reverted.
- 07
Measure
After an action is done, check whether it worked.
In PerformHubStore sales for the seven days before and after a completed action, with a verdict: worked, partly worked, did not work, or needs more data.
Methods in use today
In testing (shadow mode)
The transfer optimiser, in testing.
A mathematical optimiser for stock transfers runs in shadow mode on real data, next to the rule-based transfers PerformHub suggests today. Nothing acts on its output yet, and we publish no results from it.
Illustration with invented figures of the optimiser being tested, not of live behaviour: stock moves between four stores, and one store's size curve goes from uneven to balanced, with 48 of 48 units placed.
Units placed 48/48 exact
AI built for retail
AI that knows retail, and only speaks from your numbers.
PerformHub uses leading AI models, guided by retail rules, definitions and checks built into the product. Every figure comes from your own sales, stock, foot flow and staff data. The maths is done in code; the AI explains, drafts and answers in your language.
Every figure from your data
Sales, stock, foot flow and staff, read from your own systems and checked against each other. A missing day shows as missing, never as zero.
The maths is done in code
Totals, shares, targets, transfers and forecasts are calculated by the product, so the same question gives the same number every time.
The AI explains and drafts
Morning briefs, answers and draft plans in Arabic or English, with the numbers they are based on. A person approves what changes.
Month to date against target pace
Day 18 of 30- Mall East112%
- New Cairo104%
- Downtown99%
- Maadi93%
- Zayed86%
- Mall West78%
Joe: Mall West is behind on visitors, not on conversion. Three hours of foot flow were not entered on Thursday, so check the count before changing staff.
Retail logic built in
PerformHub does not train an AI model on your data. It runs on your company's own AI account, sends only what a question needs, and replaces phone numbers and e-mail addresses with placeholders first.
The work happens before you open.
Each step runs on its own every night. If a source is late or a day is missing, it waits, retries and tells you, instead of showing a number it can't stand behind.
Yesterday's sales arrive
Every receipt from every till, with returns counted as returns.
Gaps are repaired
A day a store never sent is rebuilt from its receipts, and marked as rebuilt.
Competitors are checked
Public prices, offers and new styles from the brands you choose to watch, with AI research that links every fact to its source.
Actions are ranked
Transfers, reorders, markdowns and staffing, sorted by what they are worth, and measured afterwards.
The brief goes out
Each manager gets their own stores, in their own language.
See it on your own stores.
Book a demoNothing changes at the till.
PerformHub connects to what you already run and only reads from it. Arabic and English across the product.
Works with what you already run
- A read-only connection to Retail Pro, encrypted end to end: nothing is ever written back to your tills.
- You connect it yourself in an online form that tests each connection as you type.
- Your history is loaded and checked against your own figures. If the link drops, loading pauses and picks up where it stopped.
- Camera reports, targets, costs and stock counts come in as sheets, with a filled-in template beside every upload.
- Attendance comes in as punch files.
Arabic, built in, not bolted on
- Arabic and English across the product, with Arabic laid out right to left.
- Arabic addresses you can share, so a link opens in Arabic for whoever you send it to.
- Each person picks their language once; it follows them to every device.
- Figures stay in Latin digits, so they match the till and the printed receipt.
Start with the systems you already use.
Setup time depends on your connections, how much history you bring and the scope you agree.
- 1
Connect or import
Agree the data sources and permissions.
- 2
Check the figures
Review history, coverage, and any missing data.
- 3
Review and act
Give the right people their views, plans, and follow-ups.
See it on your own data.
Book a demo