Guide

5 common myths about AI auto-bidding

"Just set Target ROAS and forget it" — wrong, and an easy way to burn cash. What human oversight AI bidding actually needs across lifecycle stages, and where DEXUN AdWhiz can help.

·7 min de leitura·Equipe de engenharia do DEXUN AdWhiz

Este conteúdo está disponível apenas em inglês e chinês simplificado, por isso é exibido em inglês.

AI auto-bidding (Google Smart Bidding, Meta CBO, TikTok Smart+ Bidding) is convenient — but never "set and forget". Below are 5 common — and expensive — mistakes.

Myth 1: Switch to Target ROAS on day one

New accounts don't have enough conversion data for the algorithm to learn from. Google's own guidance: at least 50 conversions per campaign in the last 30 days before switching to Target ROAS. Use Max Conversions first to let the algorithm collect signal.

Myth 2: Set Target ROAS at your goal

If your goal is 4x ROAS, set Target ROAS to 400%? Wrong — this constrains Google from exploring new traffic. Field practice: set Target ROAS 20-30% below your goal (e.g., 280-320% target), giving the algorithm room to explore. The actual delivery usually lands close to your true target anyway.

Myth 3: Account structure doesn't matter, AI handles it

Wrong. AI bidding ceiling is bounded by account structure — keyword grouping, conversion tracking accuracy, negative keyword completeness. Bad structure + advanced AI = still bad.

Myth 4: Keep Target ROAS through major sales

Black Friday, Singles' Day — short, intense promos. The algorithm can't recalibrate fast enough — it predicts based on normal conversion patterns, but user behavior is completely different during peak. Field practice: 3-5 days before the promo, switch to Max Conversion Value or Max Clicks (if you're in awareness mode), then switch back after.

Myth 5: Same ROAS target across platforms

Google Search at 4x ROAS may be reasonable, but the same 4x on TikTok Spark Ads is near-impossible — TikTok has shorter attribution windows and higher first-touch costs. Right approach: set per-platform based on each platform's CPM / CVR baseline.

Where DEXUN AdWhiz helps in the bidding layer

  • No forced advice on thin data: when history is too short, the budget optimizer returns "insufficient data" instead of an unreliable suggestion
  • Built on your own data: AI recommendations read the recent spend, conversions and ROAS of each campaign in your connected accounts and point out the laggards
  • You decide: AdWhiz recommends and never changes a bid or budget on its own; whether and how to change one is your call
  • Three platforms side by side: Google / Meta / TikTok performance compared in one dashboard instead of three ad managers
A deeper bidding strategy article is in the works. If you have a specific bidding question, email [email protected] — we'll look at your account directly.

Quer testar o que este artigo descreve?

Teste grátis de 7 dias. Cancele quando quiser.

Começar teste grátis

Artigos parecidos