To minimize the big impact for one time high spending in Initial vs Optimised Result

[gucchi123]

I want to know your experience for MMM. When I executed the MMM, I found that one time high spending such as one time TV commercial or one time influencer campaign have a big optimised share in the budget allocator output. However, Share of Spend vs Share of Effect with CPA in the one page output, there is no big share for this because those occupied a small impact is two years training dataset compared with other ad channel. Do you have any devices or way to minimized this kind of misleading situation?

The hypothetical situation is that TV is spending 100, 0, 0, 0, 0, 0 and facebook is spending 50, 30, 50, 40, 60, 80. In Initial vs Optimised Result, I saw the quite big impact for TV from the following logic
#227

The hypothetical situation seems to be good even if I execute as it is, but the real case I face is more complexed because currently I am not prioritizing TV or influencer as it was but I want to include previous year data for the model training. So I want to minimize the impact of those. One possible solution is that such amount is allocated as pro rata (e.g. 100/6 = 16.66..) in some periods in the preprocessing. Is that usually doing in your project or any devices you usually do? It would be appreciated if you give me advises.