How to record SKU steal effect on Promotions with engine Indirect effect is turned off

We run engine in PE mode. Few of the products, when promoted, show significant negative impact on other SKUs' sales. Statistical engine does not give satisfactory results at present. We wish to go live with indirect effect turned off and continue to work on data /engine. During this period, users will manage the indirect effect manually. Has anyone come across such a case earlier? What is the best way to achieve this?

For example, I am promoting SKUs A and B. Promotion in Demantra will therefore have two SKUs selected (A and . However, SKU X gets impacted as a result. Now if indirect effect is turned off, engine will not add X. So, X will not be available on Promotions worksheet. Planners need it to record impact on X's demand and also to evaluate on profitability of promotion.