Resolving Conflicting Algorithm Decisions

We have a great feature to report on, relating to how AdPartner Suite handles simultaneous algorithm runs and the decisions they produce.

When a single algorithm runs against your campaigns, each keyword is assessed against your chosen rules. The algorithm may then produce a decision: raise the bid, lower the bid, pause the keyword or leave it unchanged.

That works cleanly when only one algorithm is assessing a keyword. But what happens when two algorithms inspect the same keyword and reach different conclusions?

For example, you may have a daily algorithm designed to identify underperforming keywords, using a rule such as:

More than 10 clicks and 0 sales
Example algorithm graph for underperforming keywords.
Example algorithm for pausing underperforming keywords.

In that case, the keyword is spending budget without generating orders, so the algorithm may decide that it should be paused.

At the same time, another algorithm may be focused on profitable growth, using a rule such as:

ACOS below target ACOS: raise bid by 5%
Example algorithm that increases bid due to Low ACOS

That second algorithm may conclude that the same keyword deserves a bid increase, not a pause. Conflict resolution considers the type of action being proposed, the severity of the condition, and whether the decision is defensive, such as pausing wasted spend, or growth-oriented, such as increasing bids on profitable traffic.

Because AdPartner Suite allows advertisers to build flexible optimisation logic, this kind of competing decision is possible. A simple “first decision wins” approach would be too crude for serious campaign management. Instead, we have built a conflict resolution pass that compares competing decisions, selects the most appropriate action, and marks the other decisions as conflicts.

The result is fully automated, but still transparent. Conflicting decisions are visible in the dashboard, so advertisers can understand what happened, which action was selected, and which competing recommendations were not applied. Every competing decision is retained, so the dashboard does not simply show the final action. It also shows the rejected alternatives and why they were not applied.

As with the rest of AdPartner Suite, the aim is to let advertisers build powerful optimisation workflows without needing to manually police every edge case. The system handles the conflict resolution in the background, while keeping the reasoning visible when you need to inspect it.

Scroll to Top