Overview
Storebox operates locations with very different levels of occupancy, customer demand, competitive pressure and commercial performance.
A single promotional strategy therefore does not work equally well everywhere. Some locations need additional support to stimulate demand, while others can perform without discounts.
The product challenge was to create a scalable framework that could identify where promotional support was needed, adjust the level of promotion based on relevant business signals, and protect profitability at the same time.
My role was to help translate these commercial inputs into clear product logic, business rules, safeguards and operational controls that Engineering could implement and business teams could understand and manage.
How it works
- Location and market signalsOccupancy, booking performance, capacity, competitor pricing, costs
- Performance assessment
- Promotion recommendation
- Low-performance adjustment where needed
- Profitability safeguard
- Final promotion
The model combines internal performance data with market and cost inputs to determine an appropriate promotion level. Where booking performance is particularly weak, the system can recommend stronger promotional support. A profitability safeguard then checks that the resulting price remains commercially acceptable.
- NoneNo promotion
- LowPromotion tier 1
- MediumPromotion tier 2
- HighPromotion tier 3
Conceptual illustration with generic tiers. It does not show real promotion levels, thresholds or formulas.
Where business control stays
Automation handles the repeatable decision logic, while business teams retain control through manual overrides, location-level activation and voucher priority.
My role
My role was to help define the product and business logic behind the Dynamic Promotion Model and translate commercial goals into implementable rules.
I worked across prioritisation, business-rule definition, profitability safeguards, edge cases, operational controls and analytics, collaborating with Engineering and business stakeholders to ensure the model was both commercially useful and operationally manageable.
- Business-rule definition
- Commercial prioritisation
- Pricing and promotion logic
- Profitability safeguards
- Edge-case definition
- Engineering collaboration
- Operational controls
- Analytics and post-launch evaluation
How I approached it
Different locations need different treatment
Promotional decisions should respond to local performance rather than applying the same discount everywhere. Relevant signals include occupancy, booking performance, available capacity, competitive position and cost structure.
Commercial upside needs a guardrail
A stronger promotion may improve competitiveness or booking activity, but it should never undermine the underlying economics. The profitability safeguard therefore takes priority when rules conflict.
Automation still needs human control
Not every situation can be captured by a model. The system needs manual overrides, location-level activation, voucher priority and clear update timing.
Key decisions
- Profitability takes priorityWhen promotional logic and profitability conflict, the financial safeguard wins.
- Different data availability requires different logicNot every product or location has the same quality of competitor data. For a product segment where reliable competitor data was unavailable, missing inputs needed explicit fallback rules rather than being ignored.
- Automation must remain controllableBusiness teams need the ability to disable automated promotions or intervene when operational context requires it.
- Promotions need predictable behaviourChanges happen on a known cycle, so business teams understand when new logic becomes active.
- Promotional systems must work togetherVouchers, manual promotions and automated promotions need clear precedence rules, so customers do not receive conflicting offers.
Technical and operational complexity
The model draws on these signal categories:
- Location performance
- Occupancy
- Booking activity
- Available capacity
- Unit characteristics
- Competitor pricing
- Cost structure
- Historical performance
The complexity came from combining signals with different levels of reliability and applying them consistently across locations and unit groups.
New locations, established locations, low-occupancy locations and product groups with limited market data could not all be treated identically. The product therefore required explicit fallback logic and clear rule precedence rather than relying on a single formula.
Evaluating performance
I use SQL, Metabase, PostHog and data warehouse data to evaluate booking behaviour, location performance, promotion behaviour and commercial outcomes, and to find opportunities for further optimisation.
Outcome
The work helped move promotional decision-making from a largely manual, location-by-location process toward a scalable framework that uses operational, market and commercial signals to recommend where promotional support is needed.
The model combines automation with profitability safeguards and manual controls, allowing the business to respond more consistently to location performance without giving up commercial oversight.
The framework also created a clearer basis for analysing promotion performance and continuously refining the underlying rules.
What I learned
Pricing products are not only about choosing the right formula. The most important decisions often involve rule precedence, fallback behaviour, data quality and knowing when automation should give way to business control.
Missing or unreliable data needs explicit product logic. It should never silently become an assumption.