MTA

MTA stands for Multi-Touch Attribution, an approach to attribution that distributes the value of a conversion across all marketing touches along the customer journey rather than assigning it to one channel. Unlike Last Click (all credit to the last source) or First Click (all credit to the first), the MTA attempts to fairly evaluate the contribution of each channel.

The main MTA models: Linear (equally between all touches), Time Decay (more for the last touches), Position-Based / U-Shaped (40% for the first, 40% for the last, the remaining 20% ​​are divided between the middle ones), Data-Driven (the algorithm itself calculates the weights based on your conversions - requires a sufficient amount of data, usually 3000+ conversions per month).

The main practical problem of MTA is cross-device and cross-browser blind spots. The user saw an ad on Instagram on his phone, then a week later he found you through a search on his laptop and bought it. Last Click will say: “SEO delivered a conversion.” The MTA in GA4 will say the same thing because it didn't link the devices. The solution is probabilistic matching and logged-in user tracking, but this is already a complex infrastructure level.

My working approach: for operational decisions (where we cut the budget) I use Last Non-Direct Click as a “fair enough” model; for strategic decisions (we evaluate the contribution of brand campaigns) - Data-Driven Attribution in GA4 plus MMM for offline channels and long-term effects.