Services
Analytics, models and tools for retailers
Seven services that work together. Most clients start with one and add the next once the first is paying for itself.
Demand forecasting
Order the right amount, every week
Store and product level forecasts that account for seasonality, promotions, price changes, holidays and weather. A forecast gives buyers and planners a number they can act on, and an honest range around it. We run it for you, or build the model and hand it over.
When a product was out of stock, sales were low because of supply, not demand, and the model treats it that way.
What goes into it
- Your sales history by store, product and day, with stock availability
- Promotions, price changes, holidays, school terms, local events and weather
- Similar products, so new lines get a sensible forecast from week one
- Your pack sizes, minimum orders, lead times and shelf life
What you receive
- Forecasts by store, product and week (or day), with a likely range
- Order suggestions for buyers or planners to review
- An accuracy and bias report, refreshed every cycle
- A Power BI report or Streamlit review screen
- Documentation and a handover so you own the model
How we prove it
We hide the most recent weeks, forecast them, and compare the result with the method you use today. Then we run a pilot on one category or store group, and extend it only if it beats your current method.
What we need from you
- Sales history by store and product, ideally two years or more
- Product and store lists, your promotion calendar and price history
- Stock on hand, if you have it
- One person who knows how ordering works today
Pricing optimisation
A full pricing system, not a single model
A complete pricing system built on Bayesian, multi-factor models, with a separate model for each type of product. It covers lifecycle analysis and several pricing methods for regular price, promotion and clearance.
A single elasticity number assumes price is the only thing moving sales. In a real store, promotions, competitors, seasonality, stock and product age all move at once, and new products have no history.
What is in the system
- Product segmentation. A model suited to each group, such as everyday essentials, seasonal lines, fast-turn items, long-tail products and new launches.
- Bayesian multi-factor models. Price, promotions, competitor prices, seasonality, events, weather, stock and store effects estimated together, with similar products sharing information and a range on every recommendation.
- Lifecycle analysis. Launch, growth, maturity, decline and clearance each get their own price objective.
- Multiple pricing methods. Model-optimised, competitor position, cost-plus with margin targets, key value item and price-point rules.
- Promotion planning. Which products, what offer, how deep and when, with uplift, cannibalisation, halo and the post-promotion dip estimated.
- Clearance and markdown. When to start, how deep and how many steps, to reach a target sell-through while keeping margin.
- Guardrails and approval. Minimum margins, price ladders and key value items written as rules, with an approval screen and a record of changes.
What you receive
- A model for each product group, with documentation
- Recommended regular prices, promotion plans and clearance schedules, each with expected outcome and range
- A product lifecycle report in Power BI and a Streamlit app for scenarios and approvals
- Test results from selected stores against control stores
- Code and a handover so you own the system
How we build it
We group your range and choose a model for each group, fit the models on your history and check them on held-back weeks, then test prices and promotions in selected stores against control stores before wider rollout.
What we need from you
- Sales, price and cost history by product and store, ideally two years or more
- Your promotion calendar and the type of each offer
- Stock and availability data, product attributes and launch dates
- Your pricing rules and the lines you treat as key value items
Competitor monitoring
See competitor prices as they change
Web scraping and analysis: scheduled collection of publicly listed prices, ranges and promotions from competitor websites, matched to your own products and delivered as dashboards and alerts.
The example shows your price against two competitors, and an index where 100 means equal.
| Product | Ours | Comp. A | Comp. B | Index |
|---|---|---|---|---|
| Example product 1 | $4.50 | $4.30 | $4.60 | 101 |
| Example product 2 | $12.00 | $11.50 | $12.50 | 100 |
| Example product 3 | $7.20 | $6.40 | $6.90 | 108 |
| Example product 4 | $3.10 | $3.40 | $3.30 | 93 |
Illustrative example. Index = our price divided by the average competitor price, times 100.
What we collect
- Shelf and promotional prices and promotion dates, daily or weekly
- Which products competitors list, which are new, and which go out of stock
- Matches to your products: exact where the product is the same, close equivalents for private label and pack-size differences, with a person reviewing anything uncertain
What you receive
- A Power BI dashboard with price index by category and competitor
- Alerts when a competitor changes price or promotes a key line
- A competitor promotion calendar and a range gap report
- Data feeds into your pricing and forecasting models
How we collect responsibly
- Publicly visible pages only. We do not log in, bypass protections or collect personal data.
- We follow site terms and robots rules and keep request rates low.
- We can arrange a legal review of the sources before collection starts.
- The collected data and code belong to you.
More services
The work that makes the three answers above usable every day.
AI ready
We are not traditional consultants. We make your business AI ready: the data foundation, AI architecture and tools that let AI do real work. We find where AI will pay off, prove it on a pilot, then take it into production.
- A target AI architecture and a staged roadmap
- A shortlist of use cases ranked by value and effort, proved on a pilot
- Production rollout with monitoring, security and training
Power BI reporting
Sales, margin, stock and customer reports built in Power BI around how retail managers actually review the business: by store, category, product and week. Packages start from $2,000 per report.
- New report suites, or a rebuild of reports nobody trusts
- Fast, secure reports with the right access for each user
- Training for the people who own the reports
Data model building
Clean, joined-up data from your POS, e-commerce and inventory systems, organised into one model that every report and forecast can trust.
- A retail data model covering sales, stock, products, stores, calendar and promotions
- Automated refresh so the data is ready each morning
- A KPI dictionary so the numbers match across teams
Streamlit apps
Small custom web apps for the jobs a dashboard cannot do: forecast what-ifs, price scenario tools, buyer worksheets and approval screens that write results back.
- Apps written in Python and built around one team's workflow
- Hosted and secured in your environment
- Source code and documentation handed over
Tools we work with
We use what fits your business and avoid locking you in.