Powered byCongineer Business Solutions
Book a demo

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

  1. 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.

  2. 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.

  3. 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.

    Mathematical optimiser for stock transfers: in testing.

  4. 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.

  5. 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.

  6. 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.

  7. 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

Weekday-seasonal demand forecastingBaselines with promotions excludedExact-sum allocation of targets and new collectionsRule-based transfers with size-curve repairMarkdown depth curves with margin floorsStaffing from hourly foot flowBefore-and-after impact measurement

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

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%
On or ahead of paceBehind pacePace for today

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.

Illustration with invented figures

Retail logic built in

Trading weeks that start on the day you chooseSell-throughStock cover in weeksSize curves and core sizesLast piecesTransfers between storesTargets and month-end paceFoot flow and conversionReturns and their reasonsLike-for-like comparisons, sale days kept apart

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 demo

Nothing 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

Retail Pro 9ShopifyAttendance punch files
  • 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

العربيةEnglish
  • 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. 1

    Connect or import

    Agree the data sources and permissions.

  2. 2

    Check the figures

    Review history, coverage, and any missing data.

  3. 3

    Review and act

    Give the right people their views, plans, and follow-ups.

See it on your own data.

Book a demo