This is best explained with the marketoonist.com comic below. In this comic, the user sees shoe ads through 5 different media (leaflet, OOH, display ad, radio, TV commercial), and then buys the shoe after clicking a mobile ad.

In the last-click attribution model, all sales are contributed to users' last action, in this comic mostly to the mobile ad. The main assumption in last-click attribution is that all other channels before the purchase were irrelevant in driving the user to the purchase. This assumption makes last-click attribution highly biased towards giving high ROI to end-of-the-funnel media.

Marketing Mix Modeling, on the other hand, attempts to give a fair estimate of the sales impact of each activated medium. While last-click attribution is analytically very simple, Marketing Mix Modeling requires a more sophisticated analytics approach. Check out this article if you're interested how Marketing Mix Models work.