Measurement and effectiveness
How to tell what worked: brand and activation, attribution, marketing mix modelling and incrementality, and how to read the numbers honestly.
- Sales
- Time
- Brand building: long, slow growth
- Activation: short spikes in sales
Marketing mix modelling: how it is built, how to trust it, and when it fits
A marketing mix model estimates what each activity added to sales from aggregate data over time, and what a different budget would do. This piece sets out how one is built and validated, where the open-source tools stand in 2026, and when a model is worth commissioning.
In short
- A mix model answers the budget question across channels: what each activity added, and where the next unit of money would work hardest.
- Trust comes from validation, not from the software: out-of-sample fit, plausible carryover and saturation, stable results between refreshes, and agreement with experiments.
- Open-source tools such as Robyn and Meridian lower the cost of entry, but the hard part is still the data, the variation in spend and the judgement of whoever reads the results.
Attribution and its limits: what the credit report can and cannot tell you
Attribution shares the credit for a conversion among the tracked touchpoints before it. This piece explains rules-based and data-driven models, what they can and cannot show, how consent and tracking changes stand in October 2026, and how to use attribution without letting it set the budget.
In short
- Attribution answers a tactical question well: which tracked ads, keywords and placements came before conversions, within a channel and this week.
- It cannot say what would have happened without the ad. Sequence is not cause, and targeting makes the people who see ads different from those who do not.
- As of 2026 cookies remain in Chrome, but consent choices, app permissions and modelled conversions mean every path is partial. Check attribution against experiments and the mix model before moving budget.
Incrementality and experiments: holdouts, geo tests and lift, designed and read honestly
An experiment is the only method that shows what an activity caused. This piece compares holdouts, geo experiments and platform lift studies, sets out how to design a test that can actually detect an effect, and how to read and use the result without overclaiming.
In short
- Incrementality is the difference between what happened and what would have happened without the activity; a randomised control group is how you see the second.
- Most failed tests fail at design: too small, too short or measuring the wrong outcome. Size the test before you run it, and decide in advance what result would change the plan.
- Report the range, not only the point estimate, and feed every result back into the mix model and the attribution rules; one test answers one question.
Brand tracking and brand health: measuring what lives in memory, and linking it to sales
Brand effects build slowly and show in sales late, so they need their own measures. This piece covers tracking studies built on mental availability, brand lift studies for single campaigns, share of search as a cheap leading indicator, and how to connect all three to sales.
In short
- Track category buyers, not only customers, and measure the brand's links to the buying situations that matter (category entry points), not only whether people have heard of it.
- Brand lift reads one campaign's short-term effect on memory against a control group; a tracker reads the brand's health over months. Use each for its own question.
- Link brand measures to sales on purpose: watch share of search against market share, put brand measures into the mix model, and review them with the sales numbers in one place.
Other tracks
- IMC foundationsWhat integrated marketing communications is, the models behind it, the customer journey, the message, objectives, budgets, planning, measurement and how teams organise for it.
- Marketing principlesThe ideas every integrated plan rests on, from the people who framed them: Kotler, Drucker, Levitt, Aaker, Keller, Ries and Trout, Porter and Sharp, in plain words and with their sources.
- Disciplines 101One primer per discipline, from advertising to internal communications: how it works, its history, what it costs, the rules, its limits and when to choose it, with every claim sourced.
- Planning and operationsFrom a business objective to an integrated plan that runs: audiences, positioning, the message house and the channel mix, then the brief, the agencies, the calendar, the budget and the governance that keep it on course.
- Martech and dataThe tools behind integrated marketing, how they fit together, and the customer data that connects them, collected with consent.
- Regulation and ethicsAdvertising codes, disclosure, consent and data protection across markets, and the ethics of persuasion. Not legal advice.
- Markets and regionsWhat changes from one market to the next: media habits, languages, culture and rules, from the US and Europe to the Gulf, Egypt and Asia-Pacific.
- ToolkitCanvases, templates, checklists, worksheets and calculators to use today: each one built on a named framework, explained with its sources, and ready to fill in, print or download.