NewWeather Demand Modelling is live. Forecast demand before it arrivesWeather Demand Modelling is live

The analysis that doesn't fit a template, built for you.Custom analytics, on the same rigorous engine.

Our editions answer the questions most businesses share. When yours does not, because the business is large or complex enough that an off-the-shelf report cannot serve it, we build the answer: churn and risk models, profitability and financial analysis, demand forecasting, bespoke dashboards, whatever the decision needs. Same reconciled data, the same lineage on every number, fitted to how your business actually runs. If it can be measured, we can model it.

Custom Business IntelligenceDemonstrative data
A custom forecast · base case, confidence band and scenarios
todayupsidebasedownside
A decision rarely needs one number. Whatever we model for you, the output is decision-grade: a base case, the band of uncertainty around it, and the upside and downside scenarios with the levers that move between them.
Models in the engine
26+
ready to extend
Reconciliation
tied out
every number to source
Lineage
full
drillable to the row
First read
weeks
not quarters
Connectors
any
your stack plus custom
Security
per-tenant
isolated, SSO-ready
Margin intelligence from the award-winning team behind 100+ brands
PanasonicRainCoCheapest LiquorKing CoolingAuto ComfortiHeat & CoolAACAEInsider Experience SportsInterosPeter JacksonLa TrobeToy WorldForesightOncoreMindshopSurface SpectrumNAWTeafyEZI TagPanasonicRainCoCheapest LiquorKing CoolingAuto ComfortiHeat & CoolAACAEInsider Experience SportsInterosPeter JacksonLa TrobeToy WorldForesightOncoreMindshopSurface SpectrumNAWTeafyEZI Tag
Purest SolutionsCuratedVoir VodkaSwing & ServeNewLeafNavigataFirst EnergyHaldatecSirius GreenElite ElectricalMobile SkipsBrowedAsenoMGIDCRebateM8TMJ TutoringAcademic TutorsUCSPurest SolutionsCuratedVoir VodkaSwing & ServeNewLeafNavigataFirst EnergyHaldatecSirius GreenElite ElectricalMobile SkipsBrowedAsenoMGIDCRebateM8TMJ TutoringAcademic TutorsUCS
What we build

A standing engine, pointed at your question.

These are the models and surfaces we build most. They are a starting point, not a menu. If the decision needs something else, we build that.

Churn & retention modelling
Predict who is about to leave and why, with survival curves and propensity scores.
Financial risk analysis
Stress-test profit against the drivers that move it, with downside and upside scenarios.
Profitability analysis
True contribution margin by product, channel, customer and region, reconciled to source.
Demand forecasting
Forward demand with confidence bands and scenario planning, fitted to your history.
Attribution modelling
Credit across channels and touchpoints, several models side by side with the spread shown.
Pricing & elasticity
How demand and margin respond to price, own and cross-product.
Segmentation & personas
Group customers by behaviour and value, ranked on the profit they carry.
Anomaly detection
Catch the breaks and spikes in your data before they cost you money.
Custom dashboards & BI
Your KPIs on one board, every number drillable to the record behind it.
Data engineering
Pipelines, warehousing and connectors that tie out, built to extend.
Churn & retention modelling

Know who's leaving before they go.

Retention is the cheapest growth there is, and most businesses only see it in the rear-view mirror. We model churn from your real behaviour, fit survival curves and score each customer's risk, then hand you the at-risk list with enough lead time to act. The save is a number you manage, not a surprise you absorb.

  • Modelled, not counted. Survival curves and churn propensity from your own data
  • Scored early. The at-risk list with time to intervene
  • The save quantified. The retention a proactive play actually earns
Talk through a churn build
Custom BI · churn modellingDemonstrative data
Retention over time · modelled churn and the save from acting early
0%25%50%75%100%m0m3m6m9m1155%38%
With proactive retentionLeft to churn
We model when and why customers leave from their own behaviour, then score who is at risk. The gap between the two curves is the retention a proactive save earns: 17 points of the base kept at twelve months.
Financial risk analysis

Stress-test the profit, not just report it.

Reporting tells you what happened; risk analysis tells you what could. We model how your profit responds to each driver, rank them by how far they swing the bottom line, and run the downside and upside scenarios so the exposure is explicit. You see which levers carry the risk and plan against a range, not a single hopeful number.

  • Sensitivity ranked. The drivers that move profit most, biggest first
  • Downside and upside. Scenarios modelled in both directions
  • Exposure made explicit. Where a small slip costs the most
Talk through a risk build
Custom BI · financial riskDemonstrative data
Profit sensitivity · how far each driver swings the bottom line
base profitRepeat rate-18%+22%Average order value-12%+15%Discount depth-16%+9%COGS-11%+8%Acquisition cost-9%+7%Return rate-7%+5%
Profit is not equally exposed to everything. Here repeat rate and discount depth move it most, so a small slip in either matters more than a large move in return rate. We model the exposure so you defend the levers that actually carry the risk.
Profitability analysis

See where profit is really made and lost.

A healthy top line can hide a thin bottom one. We decompose revenue down through every cost that actually applies, COGS, fulfilment, marketing and overhead, to the profit you keep, and we build it for whatever dimension matters: product, channel, customer or region. Every figure reconciles to its source, so the number survives a CFO's scrutiny.

  • Every real cost. Revenue down to profit, deducted in order
  • Any dimension. By product, channel, customer or region
  • Reconciled to source. A profit number that ties out to the cent
Talk through a profit build
Custom BI · profitabilityDemonstrative data
Profit waterfall · revenue down to true profit, by any dimension
$0M$210M$420M$420MRevenue-$182MCOGS-$34MFulfilment-$61MMarketing-$58MOverhead$85MNet profit
Revenue of $420M becomes $85M of profit once every real cost is taken out in order. We build this for whatever cut matters to you, by division, product line, channel or region, so the leaks are visible and reconciled, not buried in a blended number.
How we build it

Bespoke, but never improvised.

Custom does not mean unreliable. Every build runs on the same warehouse, reconciliation and lineage as our products.

01

Scope the question

We define the decision and the answer it needs before any data moves.

02

Connect your data

Your stack and any custom source, joined into one reconciled model.

03

Model it

The right method for the question, built and validated, never a template forced to fit.

04

Reconcile and verify

Every number tied to its source, with the confidence shown where data is thin.

05

Deliver and iterate

A dashboard, a model or a one-off read, productionised on the same engine.

Your stack, connected

Your commerce, ad, CRM and finance systems joined into one model, plus custom connectors for anything else through our pipelines.

One reconciled warehouse

A medallion warehouse with per-tenant isolation, so raw data becomes clean, conformed marts your whole business reads from.

Trust on every number

Reconciliation checks, full lineage to the source row, and honest states where data is missing. We never fabricate a number to fill a gap.

The team behind it

Built by an award-winning analytics team.

Margin OS comes from Blufire, trusted by 100+ mid-market and enterprise brands and recognised across the APAC and Global Search Awards. The same people now model your margin.

100+
Brands served
$5M-$1B
Turnover served
4
Industry awards
100 Fast StartersAPAC Search Awards 2025 WinnerGlobal Search Awards 2025 FinalistGlobal Agency Awards 2025 Finalist
Questions

The things buyers ask.

If it can be measured, we can usually model it. Common builds are churn and retention models, financial risk and scenario analysis, profitability and unit economics, demand forecasting, attribution, pricing and elasticity, segmentation, and anomaly detection. If your question is none of those, bring it anyway.
Those are tools you have to staff and build with. We are the team that builds the model and delivers the answer, reconciled to your source data with the analytics rigour baked in. You can have the dashboard at the end, without having to become a data team to get it.
Whatever the question needs. We connect the stack you already run, and build custom connectors for anything else through our pipelines. The more history you connect, the sharper the models get.
It is scoped per project, but the first usable read usually lands in weeks, not quarters, because the warehouse, connectors and reconciliation are already built. We start narrow and deepen from there.
Yes. Anything we build can be productionised on the same engine as our editions, so a one-off read becomes a live surface your team checks whenever they need it.

Bring us the question.

Tell us the decision you are trying to make. We will tell you what we can model, what data it needs, and what the answer will look like.

Book a demo Talk to an analyst