Google Ads in 2026 is a different product to Google Ads in 2022. Not different in the way that platform updates usually work — tweaked interfaces, new features, changed defaults. Different in that the underlying architecture of how campaigns work has fundamentally shifted toward AI-driven automation, and the question of what a human should actually be doing in an account is genuinely different to what it was three years ago.
Some of that shift is good for advertisers. Some of it is good for Google. Understanding which is which is the job.
#What you'll learn in this post
- What's genuinely new and why it matters
- The Performance Max question — honestly answered
- Smart Bidding in 2026 — where it works and where it doesn't
- What manual control is still worth keeping
- How to audit your own account against the 2026 reality
#What's genuinely new
AI Max
AI Max is Google's newest campaign type, launched in 2025, which extends Smart Bidding by using AI to automatically expand keyword matching, rewrite ad copy in real-time to match search intent, and optimise placement across the full Google inventory. It's positioned as the evolution of broad match Smart Bidding.
The honest assessment: AI Max performs well in accounts with strong conversion data and sufficient budget for the learning phase. It performs poorly in accounts with thin data, niche products, or categories where search intent varies significantly from query to query. The "rewrite ad copy" feature is the one to watch most closely — verify what's actually being shown to searchers via the search terms and ad variation reports.
Performance Max evolution
PMax has matured since its 2022 rollout. The most significant change is improved transparency — you can now see channel-level spend breakdowns and, in some accounts, asset group performance. This matters because it was previously impossible to know whether your PMax budget was going to Shopping, Search, Display, or YouTube.
It's now also possible to add search term exclusions to PMax campaigns, which was a major gap in earlier versions. If you're running PMax and haven't reviewed your search term exclusions recently, that's the first thing to do.
Smart Bidding with value-based bidding
Target ROAS has become significantly more sophisticated. Google now factors in predicted customer lifetime value — not just session conversion value — where that data is available. For e-commerce businesses feeding their CRM data back to Google via enhanced conversions, this changes the optimisation calculus meaningfully.
#The Performance Max question — honestly answered
Should you be running Performance Max? This question deserves a real answer rather than a hedge.
Run PMax when:
- You have consistent conversion volume (30+ conversions per month minimum, 100+ to be comfortable)
- Your conversion tracking is accurate and attributing to purchase or lead events
- You have a well-populated asset group (multiple headlines, descriptions, images, and video)
- You've set up audience signals with your existing customer list
- You have brand exclusions preventing PMax from bidding on your own brand keywords
Don't rely on PMax when:
- Conversion volume is low (the algorithm doesn't have enough data to learn)
- You haven't set brand exclusions (PMax will cannibilise your brand Search traffic and claim credit for it)
- Your asset group is thin (limited assets means limited ability to test combinations)
- You need precise search term control (a professional services firm that only wants to appear for specific high-intent queries)
The nuance: PMax running alongside a well-structured Search campaign is often more effective than either alone. PMax for incremental reach and top-of-funnel, Search for high-intent non-brand and brand defence.
#Smart Bidding in 2026 — where it works and where it doesn't
Smart Bidding outperforms manual bidding in most accounts with adequate conversion data. This is no longer a controversial statement — the evidence across enough accounts over enough time has settled the question.
The remaining cases where manual bidding is worth considering:
New campaigns in the learning phase
Maximise Clicks is still a valid starting point for new campaigns with no conversion history. It gets impressions, generates data, and builds the conversion signal that Smart Bidding needs. Switch to Target CPA or Target ROAS once you have 30+ conversions.
Very low-volume, high-value conversion accounts
Professional services firms where a single client is worth $50,000+ may have one or two conversions a month. Smart Bidding can't learn meaningfully from that data. In these accounts, Manual CPC with bid adjustments for device, location, and time remains a viable strategy.
Brand campaigns
Some account managers prefer manual bidding on brand keywords to maintain cost control. Brand clicks are high-intent and brand CPCs are typically low — Smart Bidding's incremental value is limited compared to non-brand, and the risk of Smart Bidding inflating brand CPCs is real.
#What manual control is still worth keeping
Even in a fully automated account, a human needs to be actively managing:
Search term monitoring and negative keywords — the algorithm expands match types. Someone needs to review what searches are triggering ads and add negatives for irrelevant or off-brand terms. This is the single most valuable manual task in a Google Ads account.
Budget allocation across campaigns — let the algorithm manage bids within a campaign. Don't let it decide how much each campaign gets. That's a strategic decision.
Asset quality in RSAs — Google will test your headlines and descriptions automatically, but it can only test what you give it. If all 15 headlines are saying similar things, the algorithm's testing is limited to variations on one theme. Deliberately provide headlines with different hooks, different benefits, and different audiences in mind.
Audience exclusions — set up and maintain exclusions for converted customers, irrelevant industries, and low-quality traffic sources. The algorithm doesn't know which segments are strategically irrelevant to you.
#How to audit your account against the 2026 reality
Five checks that take under an hour:
- Search terms report — when did you last add negative keywords? If the answer is "more than a month ago," spend 20 minutes in the search terms report. Filter for spend above $10 with zero conversions. Add anything irrelevant as a negative.
- PMax asset group quality — if you're running PMax, check your asset group quality rating. Anything below "Good" is leaving performance on the table. Add missing asset types (video is the most commonly missing).
- Conversion tracking verification — go to Tools > Conversions and confirm your primary conversion action is recording at expected volume. Compare to your actual sales or leads. A significant discrepancy means tracking is broken somewhere.
- Brand exclusions in PMax — check that your brand keywords are excluded from PMax so the campaign isn't cannibalising brand Search traffic.
- Bid strategy data requirements — for each campaign running Smart Bidding, confirm you have enough monthly conversions for the strategy to learn. Target CPA needs 30+/month. Target ROAS needs 50+/month. Below those thresholds, consider Maximise Conversions without a target.
#Frequently Asked Questions
Should I try AI Max? It's worth testing in accounts with strong conversion history and at least $50/day budget. Run it as a new campaign alongside your existing structure rather than replacing what's working. Give it 6-8 weeks before drawing conclusions.
My Google Ads rep keeps recommending I expand to Performance Max. Should I listen? Assess the recommendation against the criteria above — conversion volume, asset quality, brand exclusions. Your rep's incentive is account growth, not necessarily your ROAS. Make the decision based on whether your account meets the data conditions for PMax to succeed.
Is broad match worth using? Broad match plus Smart Bidding is a valid strategy in accounts with strong conversion signals. It finds queries your exact and phrase match campaigns miss. The condition: you must be actively monitoring the search terms report and adding negatives, otherwise broad match will expand into irrelevant territory quickly.
How often should I be in the account? Weekly is the minimum for any account spending above $1,000/month. The weekly tasks: search terms review, budget pacing check, conversion volume check, and creative performance scan. Monthly: broader structural review, bid strategy assessment, and audience updates.
#The accounts that perform in 2026 vs the ones that don't
The ones that don't: fully automated, rarely checked, conversion tracking assumed to be correct, no search term reviews in months. The algorithm is working hard and spending efficiently in exactly the wrong direction.
The ones that do: Smart Bidding handling real-time bids, a human handling search terms weekly, asset groups with genuinely varied headlines, brand exclusions in PMax, and a budget allocation that reflects strategic priorities rather than default settings.
The division of labour is clearer than it's ever been. The algorithm executes. The human sets the strategy, maintains the data quality, and checks the work. Neither does the other's job well.
