Most New Zealand businesses believe they know which marketing channels drive revenue — but the data often tells a different story. Last-click attribution, the default in most analytics tools, gives full credit to the final touchpoint before conversion. It ignores everything that came before: the brand search triggered by a billboard, the organic article that built trust over months, the social ad that introduced your brand to someone who later Googled you directly. Marketing mix modeling (MMM) solves this by measuring the true incremental contribution of every channel — and it is increasingly accessible for NZ SMBs, not just large enterprises.
What Marketing Mix Modeling Actually Measures
Marketing mix modeling is a statistical approach that analyses how different marketing channels — both online and offline — contribute to business outcomes over time. Rather than assigning credit to the last click, MMM uses historical data to isolate the causal impact of each channel. It answers the question: if we had spent nothing on this channel, how much revenue would we have lost?
According to Google's guidance on marketing mix modeling, the methodology has evolved significantly with the rise of digital channels. Modern MMM can now account for cross-channel interactions, seasonality, and external factors like competitor activity or economic conditions — making it far more actionable than older approaches that treated channels as independent silos.
Why Last-Click Attribution Misleads NZ Marketers
The uncomfortable truth about last-click attribution is that it systematically overvalues bottom-of-funnel channels — brand search, direct traffic, and retargeting — while undervaluing the awareness and consideration channels that made those conversions possible. An NZ retailer running Meta ads might see that Google branded search drives the most conversions and conclude that social media is underperforming. But when they pause their Meta campaigns, branded search volume drops within weeks — revealing that social was the engine, not the passenger.
As reported by Neil Patel's guide to incrementality testing, the most reliable way to validate attribution is through controlled experiments — running geo-holdout tests or channel-pause tests that reveal what actually happens when a channel is switched off. These experiments consistently show that channels deemed "underperforming" by last-click models are often driving significant incremental revenue that would otherwise vanish.
Three Steps to Building an Incrementality Framework
For NZ SMBs that want to move beyond last-click attribution, a practical incrementality framework does not require a data science team. It requires three things:
- Centralise your channel data: Bring spend, impressions, clicks, and conversions from every channel into a single dataset. Google Ads, Meta Ads, email platforms, and organic analytics each live in silos by default. A unified view is the prerequisite for any cross-channel analysis — and it is where most NZ businesses get stuck. Tools like Google's BigQuery with pre-built connectors or modern MMM platforms can automate this.
- Run controlled experiments: The gold standard is a geo-holdout test — select two comparable regions, pause a channel in one while keeping it active in the other, then measure the revenue difference. Even a simple two-week channel-pause test, done carefully, reveals more about incrementality than years of last-click attribution data. Start with your highest-spend channel and work down.
- Build a lightweight MMM model: Open-source libraries like Meta's Robyn or Google's LightweightMMM make it possible to run marketing mix models on a standard laptop. These tools use Bayesian methods to estimate channel contributions while accounting for saturation curves (diminishing returns) and carryover effects (the lingering impact of a campaign after it ends). The output is a curve showing how each additional dollar of spend translates into revenue — which directly informs budget allocation decisions.
What Makes This Practical for NZ SMBs
The common objection is that MMM requires years of data and massive budgets. That was true a decade ago. Today, with the tools and computing power available, an NZ business spending $10,000 or more per month across three or more channels can build a useful model with 12–18 months of historical data. The key is not perfect precision — it is directional accuracy that improves budget allocation.
For businesses in the New Zealand market, the smaller scale can actually be an advantage. With fewer confounding variables than global markets — fewer channels, less competitive saturation, more stable seasonality patterns — the signal-to-noise ratio in an NZ-focused MMM is often stronger than in larger markets. A well-built model can reveal, for example, that your SEM spend has reached its saturation point while your social budget is still in the high-return zone — insights that change where the next dollar goes.
The businesses that gain the most from incrementality measurement are not the ones with the biggest budgets — they are the ones that treat measurement as an ongoing discipline rather than a one-off project. A quarterly incrementality review, paired with a living MMM model that updates as new data arrives, turns marketing measurement from a retrospective report card into a forward-looking decision engine.
The bottom line: If your marketing measurement stops at last-click attribution, you are almost certainly misallocating budget — favouring channels that harvest demand over channels that create it. For NZ businesses ready to understand what actually drives revenue, a marketing intelligence approach that combines incrementality testing with marketing mix modeling reveals where your next dollar of spend will have the greatest impact.