Selected work

Sentiment analysis

Tracking consumer sentiment across local APAC languages

Riot Games had no dependable measure of player sentiment in Asia Pacific. This system reads local-language forums across eight markets and reports a weekly score.

Category and brand strategy

Repositioning a portfolio into category whitespace

The gum and candy portfolio competed on breath freshening, a claim every brand in the category was already making. This work diagnosed the decline in Egypt and Morocco and launched repositioned variants.

Operations and demand modelling

Reducing drive-thru service times

Drive-thru service was slow because the menu on display did not match what customers ordered at that time of day. Restructuring the assortment cut average service time by 40 seconds.

Field strategy and market sizing

Reallocating 500 medical representatives

A field force of 500 medical representatives was deployed on allocations set years earlier. Reassigning them to higher-potential therapeutic areas contributed to 2% year on year sales growth.

Decision systems and reporting

Building the reporting layer executives actually use

Leadership, brand, sales and finance teams each needed a current view of performance, but the data sat in separate systems. This work built reporting on one agreed set of metrics.

Semantic search and retrieval

Conversing with fifteen years of journal entries

Fifteen years of personal writing sat in formats that could not be searched by meaning. Tagging and embedding 3,500 entries made the archive answerable in plain language.

Growth and market entry

Evaluating entry into the trading card games segment

Riot Games was weighing entry into trading card games, with significant capital at stake. Strategic partner interviews and signals intelligence were run together to build a structured case for the decision.

Growth and market entry

Taking a local marketplace into MENA

A profitable used-car marketplace in Pakistan was looking for growth beyond its home market. Every MENA market was scored separately, and each was given its own entry model rather than one regional playbook.

Purpose-built AI

The Monday morning brief

Market monitoring tools collect news but do not judge importance. This system reviews the week and reports only the developments a general manager would act on.