Use case · Revenue prediction
Forecast what a site will earn before you sign the lease
Every new location is a serious bet, and the lease is usually signed against a hunch about what the site will do. SmartMarket trains a machine-learning model on your own sales history and the location attributes around each candidate site, and returns a forecast revenue range with a confidence interval, so the capex decision rests on data science, not a guess.
Get a demo →The decision
The number the lease is signed against
Before capex is committed, one number decides everything: what will this site actually earn? The rent, the staffing, the fit-out and the payback period are all sized against that figure, and in most expansion teams it is still an estimate carried in someone's head, anchored to the last store that felt similar.
When the forecast is wrong, the cost is structural. A site that under-earns drags on the network for years, and the bet was the easy part to verify and the hardest part to get right. The question is not whether the site looks good. It is what it will earn, and how confident you can be in that range.
What the model reads
Two inputs, one forecast
Revenue prediction does not start from scratch for every site. It learns the relationship between location and revenue from the stores you already run, then applies that learned pattern to each candidate site using the attributes SmartMarket measures around it.
- 01
Your sales history · What your existing stores actually earn
The model trains on your own store-level sales data, so the forecast is calibrated to how your brand performs, not to a generic benchmark.
- 02
Demand patterns · Footfall and demand around the site
Visitor-volume signals at the 150m grid level and the demand patterns that surround each candidate location.
- 03
Competition density · How much competition the site shares the catchment with
Competition density read from the same point-of-interest data that powers the rest of SmartMarket, so the forecast accounts for who else is already chasing the same demand.
- 04
Adjacency · What sits next to the site
The mix of nearby points of interest and the adjacencies that lift or dampen footfall for your category.
- 05
Infrastructure · What is built, and what is coming
Roads, transit access and upcoming infrastructure projects from InfraNow, so a site is forecast against the market it will open into, not only the one that exists today.
Why a range beats a point estimate
A single predicted number invites false confidence. SmartMarket returns a forecast as a range with a confidence interval, so the decision is made against both the expected revenue and the uncertainty around it.
That lets you compare sites honestly. A site with a higher midpoint but a wide interval can be a riskier bet than one with a slightly lower midpoint and a tight range, and the forecast makes that trade-off visible instead of hiding it inside one number.
How it works
From sales history to a forecast in three steps
Revenue prediction runs the same way every time, whether you are forecasting one site or a whole pipeline.
- 01
Data preparation · Your sales history meets the location layer
Your store-level sales data is aligned with the SmartMarket location attributes around each existing store, building the labelled history the model learns from.
- 02
Model training · The model learns what drives your revenue
A machine-learning model is trained on that prepared history, learning how demand, competition, adjacency and infrastructure relate to the revenue your stores actually produce.
- 03
Forecasting · Each candidate site gets a predicted range
The trained model is applied to the attributes of each candidate site, returning monthly and annual revenue ranges with confidence intervals to compare and rank.
From signals to a forecast.
Footfall, competition, adjacency and infrastructure stream into the candidate site, and resolve into a predicted revenue range with a confidence interval. The forecast is read straight off the same map every other decision uses.
Where it fits
A forecast on top of a ranked shortlist
Revenue prediction is not a standalone tool. It layers your own sales history on top of the same scored grid that powers site selection, so a shortlist of strong locations becomes a shortlist with a revenue range attached to each one. Site selection tells you where the opportunity is, revenue prediction tells you what it is likely worth, and competitor analysis tells you who you will share it with.
The forecast then feeds the surfaces your decision-makers already use. Analysts and data scientists can train and inspect the model on the full platform with raw granular data and BI integration, while executives read the resulting ranges through AI Reports and the CXO dashboard, with the reasons behind each forecast made explicit.
Why the inputs are trustworthy
A forecast is only as current as its inputs
A revenue range built on stale surroundings is a guess with a decimal point. SmartMarket reads live point-of-interest quality signals through Google Places Insights, refreshed monthly, so the competition density and adjacency the model treats as features describe the catchment as it stands when you run the forecast, not when the data was bought.
- Retail
- Real estate
- BFSI
- Telecom
Forecast revenue on your own pipeline of sites.
Bring your sales history and a shortlist of candidate sites, and we will walk through how SmartMarket trains a model and returns a forecast range for each one.
Get a demo →