Tom Rowley. Digital marketing, analytics and practical AI, from Christchurch, New Zealand.
Every demo here runs on public New Zealand data, pulled on a schedule. Nothing is typed in by hand.
New Zealand earthquakes from GeoNet, pulled every 15 minutes into a raw data lake, modelled in a warehouse and charted live.
See it live
NZ earthquakes, live
GeoNet quakes every 15 minutes: map, magnitudes and depths.
Earthquakes & the Moon
Ten years of data, declustering and proper statistics.
NZ power vs weather
Half-hourly electricity prices against temperature and wind.
NZ flights fuel map
Every aircraft in the air now, with estimated fuel, cost and CO₂.
Crime & the Moon
Three years of daily Police data, restocked from CSV files.
Built from the Companies Office bulk data extract, as at 1 October 2026. Cleaned, de-identified and rolled up into counts by sector, region and town before anything is shown.
Net new companies: about 27,000 a year in 2020-21, 7,000-15,000 a year since 2022.
Removals from the register: 30,600 in 2020, 50,100 in 2025.
of the largest sector is property investment and rental companies.
more insolvencies in 2026 to date than at this point in 2025, itself a 15-year high.
Data and dashboards, practical AI, paid media and search. Explore the work behind each skill, from the first question to a useful decision.
Make the numbers useful enough to act on.
Define meaningful events, conversions and business outcomes. Map the customer journey and identify which signals are suitable for reporting, optimisation and decision-making.
Investigate duplicate events, missing coverage, attribution differences and reporting delay. Compare like-for-like periods and distinguish observed movement from a plausible explanation.
Bring ad platform, analytics, CRM and sales data together in one place. Choose the right home for it: a data lake for raw files, a warehouse for clean modelled tables, or a lakehouse that does both. Join it up with consistent IDs and definitions so every report starts from the same source of truth. See a live data lake ↗
Build dashboards people actually use: a few decision-ready measures, the right level of detail for each audience, filters that answer real questions and notes on what each number can and cannot tell you. See one built on NZ company data ↗ Or NZ power prices against the weather ↗ Or every aircraft over NZ, with its fuel burn ↗
A measurement plan, an event and conversion audit, a joined-up data model, dashboards with agreed definitions and a clear account of what the data can, and cannot, support.
Turn repeatable work into something easier to run.
Break a recurring task into inputs, decisions, checks and outputs. Identify where AI can help with research, drafting or organisation and where human review remains necessary.
Structure reference material so it can be found and reused. Build prompts, templates and small prototypes around a defined job, with explicit assumptions and useful boundaries.
A documented workflow, a reusable template or a working prototype. The Goldmine calculator on this site is one example of making an idea tangible.
From search intent to a useful business outcome.
Review campaign structure, conversion goals, search intent and delivery constraints. Separate a measurement problem from a demand, quality or commercial problem before choosing a change.
Connect keywords, ads and landing pages around the offer. Explore search terms, exclusions, budget allocation, bidding goals and asset coverage with clear reasons for each change.
An account audit, prioritised actions, a measurement checklist and a test plan with a baseline, owner and review criteria.
Connect the offer, the creative and the quality of demand.
Translate a business goal into a campaign brief: offer, audience, geography, desired action and budget. Consider prospecting and remarketing in the context of available data and the buying journey.
Explore distinct creative concepts, hooks, proof and calls to action. Review the path from ad to enquiry or sale, including form friction, qualification and follow-up.
A campaign brief, a creative testing matrix and a reporting approach that distinguishes platform conversions from useful business outcomes.
Make the promise clear, and give people a reason to believe it.
Connect a customer need with an appropriate offer, specific benefits, credible proof and a clear next step. Adapt the message to the channel and stage of the buying journey.
Organise meaningful variations in angle, hook, format and call to action. Keep track of what each variation is intended to test, rather than changing words without a hypothesis.
Ad copy options, creative briefs and an organised testing plan. Claims and proof need to be supported by the business before they go into an ad.
Put the next test in the context of the whole business.
Start with objectives, audience, offer, seasonality and operational capacity. Give each channel a clear role and define how progress will be assessed.
Compare opportunities using compatible outcome definitions. Consider marginal returns, confidence, constraints and test costs rather than allocating budget solely on headline platform metrics.
A media plan, a budget rationale and a small set of priorities with dependencies, owners and decision dates.
Find the questions people ask and the pages that should answer.
Review how pages are discovered, understood and surfaced in search. Use Search Console to explore queries, landing pages, clicks and visibility changes.
Connect search intent to useful page content, clear titles, internal links and a logical site structure. Prioritise opportunities according to relevance, effort and business value.
A practical SEO audit, an opportunity list and content briefs that explain the audience, intent and purpose of each proposed page.
Make each change leave the team a little wiser.
State the problem, proposed mechanism and expected outcome. Establish a baseline and identify dependencies, competing changes and the evidence needed for a decision.
Allow for outcome delay and uncertainty. Separate implementation success from performance impact, and avoid turning a small or noisy result into a sweeping conclusion.
A change register, experiment brief and a short decision note: retain, revise, stop or investigate, with the reasoning recorded.
Small browser games I built for fun, starring a slightly cartoon me. No ads were harmed.
That’s a little about me. You can find more about my work, or say hello, on LinkedIn.
Find Tom on LinkedIn
