sb/

Experience

Fifteen years of turning messy data into things people use.

From stock audits on the shop floor to leading a global analytics function. Here's the story so far.

By the numbers

  1. 5years

    of historical audit data backfilled into the cloud, then kept stable through repeated upstream changes.

  2. 5–10hours a week

    of manual dashboard updating, automated away.

  3. 400–700hours a year

    estimated team time saved by adopting Copilot Premium, from the business case I built and then led.

  4. £5M

    of stock errors surfaced ahead of period close, by earlier iterations of global health check reporting.

Data and analytics engineering

Things that keep working after I've stopped looking at them.

It started with ad-hoc BigQuery queries, managed on individual machines.

I led the move to version-controlled development in dbt, taught myself the stack, and coached a colleague to the same standard.

Bitbucket, Jenkins and Airflow now run it, with live feeds migrated without interrupting reporting.

Dashboards repointed without a break in service, and a catalogue of key datasets to decide what to build next.

Dashboards and visualisation

Numbers only help if people can use them.

  • Dashboards in Tableau and Metabase that fraud, supply chain, brand protection and security teams rely on day to day.
  • Global reporting you can view by region, country or store, and drill down from headline scores to individual questions.
  • Led the rollout of Metabase, retiring another BI tool and avoiding a six-figure annual licence cost.
  • Automated a dashboard estate, saving 5–10 hours a week, and cut a monthly report from half a day to around 30 minutes.

Detecting fraud and loss

A loss that went undetected for over a year.

Analysis of inventory leakage led to a multimillion-pound fraud investigation.

  • Risk scoring that replaced largely manual investigation processes.
  • Grey market activity quantified across distributor networks.
  • Years of stock audit and loss prevention in retail before any of it.

Leading teams and shaping strategy

Built a global analytics function from scratch.

Set how the team works, and what “production-ready” means.

Owned the data strategy and the roadmap.

Made the case for Copilot with real numbers, then led the rollout.

Worked across fraud, security, commercial, supply chain and logistics.

Coached colleagues in SQL and cloud tooling.

In more detail

Pick a card to go deeper.

Moved queries into dbt.

Dashboards teams rely on.

A loss unseen for over a year.

Built a team from scratch.

Bringing in AI tools.

Fifteen years in retail loss.

Data and analytics engineering

I like building things that keep working after I've stopped looking at them.

  • Led the move from ad-hoc, locally managed BigQuery queries to version-controlled development in dbt, with Bitbucket, Jenkins CI/CD and Airflow.
  • Taught myself the stack, migrated live feeds and repointed downstream Tableau dependencies without interrupting reporting, then coached another analyst to the same standard.
  • Built an API integration into Google Cloud Platform and backfilled five years of audit data, keeping it stable through repeated upstream changes while preserving comparability.
  • Built a Power App giving a controlled way to maintain exceptions within a dashboard.
  • Kept a catalogue of key datasets and mapped pipelines end to end, so work could be prioritised by data readiness, complexity and business risk.

Dashboards and visualisation

Numbers only help if people can use them.

  • Designed dashboards in Tableau and Metabase that fraud, supply chain, brand protection and security teams rely on day to day.
  • Built global health check reporting covering shrinkage, stock accuracy and at-risk inventory, viewable by region, country or store.
  • Overhauled audit reporting so you can drill down from headline scores to individual questions, highlighting recurring control failures.
  • Brought open and failed order reporting into one global view across several POS platforms, cutting the time to spot system-level failures.
  • Rebuilt the global retail stock count workbook, used weekly by retail teams across all markets.
  • Tested Metabase and led its rollout, retiring another BI tool and avoiding a six-figure annual licence cost.
  • Automated a dashboard estate, saving 5–10 hours a week, and cut a monthly report from half a day to around 30 minutes.

Detecting fraud and loss

This is where I started, and it still drives much of my work.

  • Analysis of inventory leakage led to a multimillion-pound fraud investigation into a loss that had gone undetected for over a year.
  • Designed a weighted risk-scoring model that replaced a largely manual process, with one dashboard for investigators and a long-term trend view for a brand protection committee. It surfaced patterns nobody had spotted before.
  • Quantified grey market activity across distributor networks, informing commercial decisions.
  • Combined connected-device data with company systems to detect illicit trade, with automated daily alerting.
  • Replaced a manual spreadsheet process with a tracker for stolen machines.
  • Earlier reporting surfaced £5M of stock errors ahead of period close.

Leading teams and strategy

Setting direction, and then helping people deliver it.

  • Established a global data and analytics function from scratch, then led a small team, recruiting one analyst and bringing in another.
  • Set how the team works: requests prioritised by risk, impact and effort; clear definitions of production-ready datasets and dashboards; documented ownership and refresh expectations.
  • Owned the data strategy and delivery roadmap, balancing quick wins against longer-term foundations.
  • Worked closely with fraud, brand protection, security, commercial policy, supply chain, logistics and data platform teams.
  • Coached colleagues in SQL and cloud tooling, which reduced repeat ad-hoc requests.

AI tools

Useful when you know what you want from them.

  • Built the business case for Microsoft Copilot Premium, identifying SQL development, reporting, documentation and meeting summaries as the biggest time savings, an estimated 400–700 hours a year across the team.
  • Led the rollout and set how the team uses it.
  • Use it day to day for query drafting and debugging, reporting, technical documentation and Power Apps development.
  • Outside work, I use Claude and ChatGPT regularly, including for learning to build things like this website.

Fifteen years in retail loss

The groundwork: seeing how stock and sales behave on the shop floor, not just in a spreadsheet.

  • Started in field-based stock audits and stocktake investigations, working with store teams and reporting to senior management.
  • Reconciled sales data across around 1,500 stores in the UK and internationally.
  • Built KPI dashboards and automated reporting for a national retail estate.
  • Supported risk management across several retail brands, covering roughly 400 sites, and led root-cause analysis on known and unknown loss.
  • Supported the global rollout of a new POS platform, including on-site delivery of the first store go-live in Paris.

Tools I use

Analytics and BI

  • SQL
  • BigQuery
  • Tableau
  • Metabase
  • Looker

Engineering

  • dbt
  • Git
  • Bitbucket
  • Jenkins
  • Airflow
  • Google Cloud Platform

Systems

  • SAP S/4HANA
  • CRM
  • Power Apps

AI

  • Microsoft Copilot Premium
  • Claude
  • ChatGPT

That's the short version.

The full timeline lives on LinkedIn. Or just say hi.

Email me LinkedIn About me