Google Ads Smart Bidding: The Strategy Selector

Smart Bidding is Google's automated bid management — algorithms adjusting your bids auction by auction, using signals no human could process in real time. It works. It also fails predictably when misapplied: wrong strategy for the goal, too little data to learn from, or constant human interference resetting the learning. This guide is the strategy selector: which automated bid strategy fits which situation, what each one needs to succeed, and when manual bidding still wins.

The strategies, explained plainly

• Maximise Clicks: gets the most clicks for your budget. Useful for research, new campaigns gathering data, and driving traffic to content. It does not care about quality — set a max CPC cap or it will buy expensive junk.
• Maximise Conversions: gets the most conversions for your budget, no target set. The default choice when you have conversion data but no firm CPA target yet.
• Target CPA: aims for a specific cost per conversion. The workhorse for lead generation — but the target must be realistic (within ~20% of your actual CPA) or the algorithm throttles volume chasing an impossible number.
• Maximise Conversion Value: optimises for total revenue rather than conversion count. The pick when conversion values vary — ecommerce, or lead gen with values assigned.
• Target ROAS: aims for a specific return on ad spend. Powerful for ecommerce with accurate values; punishing if your values or margins are wrong.
• Target Impression Share: bids to appear in a chosen share of auctions (usually top of page). A visibility tool, not a performance strategy — appropriate for brand defence and launches, wasteful as a default.
• Manual CPC: you set the bids. Still legitimate for tiny accounts, tight control situations, and brand campaigns where you want absolute position.

The selector: which strategy for your situation

• New campaign, no data: start Maximise Clicks (with a cap) or Manual CPC to gather 20–30 conversions, then graduate to a target-based strategy. Skipping this step is the most common Smart Bidding failure.
• Lead gen with a CPA target: Target CPA — once you have ~30 conversions in 30 days in the campaign and the target reflects reality.
• Ecommerce with solid margins: Target ROAS with accurate product-level values. Garbage values in, garbage bidding out.
• Variable-value leads: Maximise Conversion Value with values assigned per lead type (a quote request worth £200, a brochure download worth £10).
• Brand campaign: Target Impression Share (absolute top, ~90%+) or Manual CPC. Cheap, controlled, defensive.
• Tiny account (<15 conversions/month): Manual CPC or Maximise Clicks with strict caps. The algorithm can't learn from a trickle — don't ask it to.

The learning period: rules of engagement

After switching strategies or making major changes, Google needs time — typically one to two weeks, or one full conversion cycle for long sales processes. During learning: don't touch anything. No bid tweaks, no budget changes, no keyword additions. Every significant edit restarts the clock, and the number-one cause of "Smart Bidding doesn't work for us" is an account manager who fiddled with it daily. Set a calendar reminder, walk away, evaluate after.

Feeding the algorithm properly

• Conversion quality over quantity. One clean primary conversion (qualified enquiry, purchase) beats five fuzzy ones. Move secondary actions (page views, brochure downloads) to secondary status so bidding ignores them.
• Values where possible. Even rough values (£200 for a quote request, £20 for a callback) let value-based strategies make smarter trade-offs than count-based ones.
• Consistent data. Enhanced conversions and offline conversion imports (for long sales cycles) keep the algorithm honest about what actually became revenue.
• Realistic targets. Set Target CPA 10–20% below current CPA and let it settle; then tighten gradually. Demanding half your current CPA on day one just strangles delivery.

Portfolio bid strategies (managing targets across campaigns)

When several campaigns share the same goal — say, five service campaigns all targeting a £60 CPA — a portfolio bid strategy pools their conversion data under one target. The algorithm learns across the portfolio instead of in five isolated silos, which stabilises performance on smaller campaigns that couldn't sustain Smart Bidding alone. Use portfolios when campaigns share a genuine goal and similar economics; don't force wildly different campaigns (brand + generic + competitor) under one target just to tidy the account — the average will suit none of them.

Troubleshooting: when Smart Bidding misbehaves

• "Limited by budget" + target CPA: the algorithm wants more volume than the budget allows. Either raise the budget or loosen the target — constraining both starves learning.
• Sudden CPA spike: check for conversion tracking breaks first (a duplicated tag doubles reported conversions and the algorithm chases ghosts), then seasonality, then competitor activity. Don't change the strategy; diagnose the input.
• Zero conversions after switching: the target is likely unrealistic for the campaign's history. Loosen it 30–50%, let it stabilise for two weeks, then tighten gradually.
• Learning never ends: perpetual "learning" status usually means constant edits. Freeze changes for two full weeks and watch it settle.

Seasonality adjustments (telling Google about your calendar)

Smart Bidding learns from history — which means predictable demand spikes (Black Friday, January sales, your industry's peak season) initially confuse it, because this year's spike doesn't match last month's baseline. Seasonality adjustments let you tell Google a conversion-rate change is coming: set the date range and expected uplift, and the algorithm bids proactively instead of learning reactively. Use them for genuine spikes of 30%+ lasting at least a few days — not for minor blips, which the algorithm handles fine alone. And remove them promptly when the period ends; a stale adjustment bidding for Christmas demand in February is worse than none at all.

When manual still wins

Manual CPC isn't obsolete — it's a specialist tool. Tiny budgets where every click is scrutinised, brand campaigns needing absolute position control, testing phases where you want to isolate variables, and markets so thin the algorithm has nothing to learn from. The mistake isn't using manual bidding; it's using it by default on accounts big enough for automation to outperform you.

Smart Bidding done right is the closest thing PPC has to free performance — done wrong, it's an expensive way to automate bad decisions. Our PPC advertising team sets the strategy, feeds it clean data, and knows when to leave it alone. For the bigger AI picture, read our AI in PPC guide. Get a free consultation.