The commercial engine for retailers
Decisions arrive ready to approve inside your guardrails, not reports you still have to turn into action. Built inside a real $300M+ multi-country retailer to run its own P&L, now productized for yours. Live in your own cloud in 30 to 90 days.
Sample decision · Allocate Built inside a $300M+ multi-country, multi-category retailer to run its own P&L. It ran real OTB, allocation, and markdown cycles there for years, then we productized it.
One engine. Five verticals on one spine. Breadth by design, not bolt-on.
The gap
Your OTB recalculates on bad inputs. Allocation is a spreadsheet someone keeps alive by hand. Out-of-stocks quietly bleed margin while the enterprise suites quote a six to twenty-four month, SI-led rollout and a dollar of services for every dollar of license. The specialty tools each fix one job, in one vertical, on their cloud. Nobody runs the whole commercial function for a multi-category retailer. That gap is the tax you pay every season.
The week goes to pulling and reconciling reports across systems. The deciding happens last, on tired eyes, against a closing window.
Open-to-buy gets recalculated on stale inputs and transposed by hand between tools. Everyone references it. No one trusts it.
Most planning pilots die in integration and data, not modeling. A widely cited 2024 MIT study put the AI-pilot failure rate high, and integration is usually why.
What MEGA is
Plan, Buy, Allocate, Sell, Analyze, Decide on one spine. It ran inside a live $300M+ multi-country, multi-category retailer for years, then productized for retailers $50M to $500M and up. The AI lives in the core, not stapled on as a chat box.
What breaks if you remove the AI?
For most tools, productivity dips and the business keeps running. For MEGA the business model breaks. The agents take typed actions inside your workflow, on your governed data, and learn from the feedback. That is the test. MEGA passes it.
Plan · Buy · Allocate · Sell · Analyze · Decide
The spine is the product, the navigation, and the mental model. Start with the one workflow that hurts most, usually Allocate or OTB, then let the rest expand after the first win.
Plan
Plan sets the financial frame and the open-to-buy. Top-down targets meet bottom-up reality, and the OTB recalculates as actuals land, not three weeks later in a spreadsheet someone forgot to refresh.
See Plan in the platform →| Category | Plan | Actual | OTB open |
|---|---|---|---|
| Womens | 4.20M | 3.94M | +0.31M |
| Mens | 2.85M | 2.91M | −0.06M |
| Accessories | 1.10M | 1.18M | −0.12M |
OTB recalculated on last night’s sell-through. Illustrative.
Buy
Buy builds the order, not a hint about it. Range architecture, drops, MOQ and lead time resolve into a store-level buy plan you can edit inline, with the reasoning on every line.
See Buy in the platform →| Attr group | Drop 1 | Drop 2 | Options |
|---|---|---|---|
| Denim / slim | 200 | 80 | 10 → 5 |
| Knit / crew | 120 | 60 | 6 → 4 |
| Tee / graphic | 210 | 70 | 7 → 7 |
Buy plan across drops, store-style level. Illustrative.
Allocate
Allocate places stock at store-SKU level and rebalances as demand moves. It kills the grunt-work of pulling reports. It is the most automatable workflow and the quickest to show value day to day, which is why we usually start a pilot here.
See Allocate in the platform →| Store | On hand | Cover | Move |
|---|---|---|---|
| DXB-014 | 38 | 6d | +12 |
| AUH-006 | 71 | 21d | −16 |
| KWI-002 | 12 | 3d | +20 |
Store-SKU rebalance within your ±guardrail. Illustrative.
Sell
Sell manages in-season replenishment and markdown timing. It watches true rate-of-sale, not broken-size noise, and proposes the reorder or the markdown at the moment that protects the most margin.
See Sell in the platform →| Style | ROS/day | Sell-thru | Call |
|---|---|---|---|
| SW-2241 | 4.1 | 62% | Reorder |
| SW-1108 | 0.6 | 28% | Mark −20% |
| SW-3390 | 2.7 | 54% | Hold |
Replenish / markdown timing on true ROS. Illustrative.
Analyze
Analyze is the metric tree your merchants already think in. Sell-through, GMROI, turn, markdown, OTB, on-shelf availability. No export to a BI tool, no chart to decode. The numbers come to the decision.
See Analyze in the platform →| Metric | Now | Plan | Δ |
|---|---|---|---|
| GMROI | 3.1 | 2.8 | +0.3 |
| Sell-through | 58% | 61% | −3pt |
| Stock turn | 4.6 | 4.2 | +0.4 |
The metric tree, live. Illustrative.
Decide
Decide is where MEGA acts within your guardrails, traces every action, and graduates from suggest to execute as the calls prove out. Every decision is explainable, approvable, overridable, and logged.
See Decide in the platform →| Action | Guardrail | Status | Projected |
|---|---|---|---|
| Cut WMN receipts 8% | ±10% | Needs you | +1.4pt mgn |
| Rebalance AUH→KWI | ±20% | Approved | +0.6pt avail |
| Markdown SW-1108 | timing | Sent back | pending |
Decision trace. Projected deltas, on your data. Illustrative.
A morning inside MEGA
No dashboard to interpret, no export to a BI tool. The decisions come to you: the OTB re-forecast, the replenishment moves, the allocation rebalance, the markdown call. Each shows the signal it read, the action it wants to take, and the projected P&L delta. You approve two, override one, send one back.
Autonomy on this desk
Autonomy is earned. It starts at suggest and graduates as the calls prove out. You set the ceiling.
Illustrative surface. Every number here is sample data. In the pilot, it runs on yours.
Projected outcomes
A modeled stock-availability gain maps to a modeled revenue and margin lift, validated against your own numbers in the pilot. We reference industry benchmarks as other companies’ numbers, never as MEGA’s own results, and we do not borrow someone else’s case study.
Open the projection model →Projected, illustrative. Your factors get fit in the pilot. Numbers here are not a MEGA result.
Enterprise suites take a year and an SI army. Specialty tools fix one job in one vertical. MEGA does neither.
See the full comparison →Phase 1, not a platform bet
Pick the workflow that hurts most, usually Allocate or OTB. We run MEGA alongside your current process, on your data, in your cloud, on pre-negotiated terms. Day 30 you are configured. Day 90 you see modeled lift on the one KPI you chose, measured against your own baseline.
Day 30 configured · Day 90 modeled lift on your chosen KPI · measured on your own data
