Holding a new account through the learning phase instead of killing it.

The challenge

This was a brand-new account with no history and no conversion data, so Google’s Smart Bidding had nothing to learn from. The first month produced just two booked appointments at roughly R2,535 each. On paper, a disaster. In reality, that is exactly what a cold account looks like in its first few weeks: volatile, expensive and thin while the algorithm gathers signal.

There was a second complication. Online booking was a brand-new channel for this business. Their patients had always picked up the phone, so every online booking the ads produced was demand captured in a channel that did not exist for them before.

The approach

The instinct with an underperforming new account is to rip out the bid strategy, cut the budget, or change something every day. Each of those resets the learning and starts the clock again. Instead we fed the account cleaner signals and held our nerve.

• Early search-term audits cut competitor and wrong-intent terms (ophthalmologist, eye surgeon) before they drained budget.
• Conversion tracking was fixed before scaling: genuine booking calls were too short to clear the minimum call length Google was counting, so real leads were going uncounted until the threshold was corrected.
• Every path a lead can take was tracked, not just online form-fills.
• The bid strategy was held through the learning phase rather than reset.

The takeaway

A new account looks worst exactly when you are most tempted to interfere. The win here was not a clever trick. It was clean signals, correct tracking, and the patience to let Smart Bidding finish learning. If a new campaign looks broken after two weeks, make sure you are reading a learning curve, not a final verdict.