The "Algorithm Is Learning" Excuse: A 90-Day Stall Tactic

"The algorithm needs more data." "Smart Bidding is still in learning phase." "Give it 90 days to optimise." These phrases sound technical. They sound patient. They are often excuses for doing nothing.

Machine learning does require data and time. But agencies have weaponised this truth into a shield against accountability. The learning period has become a convenient explanation for why results have not improved and why you should keep paying.

How Long Learning Actually Takes

Let us be specific about what Google's automation actually requires:

Notice that none of these are 90 days. If your agency is asking for three months of patience, they are buying time, not waiting for the algorithm.

The Warning Signs

What Good Agencies Say Instead

The learning period is real. But competent agencies communicate it differently:

• • "We expect learning to complete by [specific date]"
• • "Here are the leading indicators we're tracking during learning"
• • "If we don't see X by week 3, we'll pivot to Y"
• • "Learning is on track because [specific evidence]"

Questions to Ask

When your agency invokes learning periods, ask:

1. 1. What specific date do you expect learning to complete?
2. 2. What will be different once learning is done?
3. 3. What are we learning from the learning period?
4. 4. How do current results compare to benchmark?
5. 5. What happens if performance does not improve by X date?

Vague answers to specific questions indicate the learning excuse is covering for lack of progress.

When Learning is Legitimate

To be fair, there are genuine learning scenarios:

• • Brand new account with no historical data
• • Significant conversion tracking changes
• • New bid strategy on established campaigns
• • Major seasonal shifts requiring recalibration

In these cases, patience is warranted. But patience is not blind. You should still see directional improvement within 2 to 3 weeks.