Key takeaways
- PPC hasn't died despite repeated predictions since 2017; instead, AI has fundamentally transformed who does the work and what skills matter most in 2026.
- While cost per click has roughly doubled over the past decade, cost per lead and conversion efficiency have improved significantly, revealing hidden inefficiencies rather than channel decline.
- AI Overviews and AI Mode have pressured top-of-funnel search dramatically, but bottom-funnel conversions remain strong for accounts with proper strategy.
- AI tools excel at setting bids, generating ad copy variations, and discovering long-tail queries; strategic decisions about conversion value remain firmly human responsibility.
- Success with modern search advertising requires proper conversion tracking, offline data integration, intent-based keyword sorting, and continuous monitoring of traffic quality.
- The PPC skillset evolved from tactical lever-pulling to strategic oversight; the industry needs fewer entry-level positions but continues to need professionals who understand fundamentals and can drive results.
PPC vs AI: what 10 years in PPC revealed
PPC has been declared dead roughly 27 times over the past decade. The first declaration came in 2017, when Google introduced close variants on exact match keywords. A client received a blog post suggesting a strategic pivot to SEO. The account was not redirected. That year, the client saw its best performance yet.
When "AI killed PPC" posts started flooding professional networks in 2025, the initial reaction was skeptical. The second reaction was more nuanced. This time did feel different. Users ask ChatGPT questions they used to search on Google, and Google answers many common search queries before anyone scrolls to an ad. Meanwhile, the advertising platforms now write the ads, select the queries, and choose the landing page.
The legitimate question became: How much has AI actually changed PPC performance?
The answer requires examining how the platforms evolved, what practitioners report, and what the structural changes actually mean for bottom-line results. What emerges is that the job as it was performed in 2016 has indeed largely disappeared. PPC itself, however, did not.
What actually changed
Significant developments occurred across eighteen months. Understanding them requires walking through the timeline.
Google's AI Max announcement and rollout
Google announced AI Max for Search in May 2025 and made it generally available in April 2026. The tool sits on top of standard Search campaigns and automatically expands keyword matching, rewrites headlines using landing page copy, and sometimes selects a different final URL than the one specified. This automation shifted strategic responsibility from campaign setup to upstream decisions about conversion value and tracking accuracy.
AI Overviews and AI Mode
Simultaneously, ads began appearing inside AI Overviews and AI Mode. These placements cannot be targeted directly. Campaigns running strict exact and phrase match keywords without AI Max typically fail to qualify for the broader questions that appear in AI Mode and AI Overviews. This appears intentional.
The practical impact on paid search is significant. Informational queries; the "how to" and "what is" questions that historically drove volume, now frequently show AI-generated answers before any paid results. This has visibly pressured click-through rates on those query types.
OpenAI enters the advertising space
OpenAI launched ads in ChatGPT in February 2026 with a $60 CPM. By April, the CPM had fallen to approximately $25, with cost-per-click options ranging from $3 to $5. By May 2026, self-serve access became available across the US with no minimum spend requirement.
This represents a new channel altogether, not a replacement for Google search. Early testing suggests it performs differently—more similar to discovery advertising than search, requiring different measurement approaches and audience expectations.
Meta's automation initiatives
In 2025, Meta indicated plans to achieve fully automated advertising by the end of 2026. The proposed system requires only a product image and budget; the platform handles all other decisions.
The pattern across all platforms is clear: human operators are being removed from tactical decisions entirely.
So what do the numbers really say?
The data tells a more complex story than either advocates or skeptics typically acknowledge.
The cost per click reality
Cost per click has increased materially over the past decade. This is real and visible in every account. However, the hidden story is equally important: cost per lead and cost per acquisition have improved in many accounts, and conversion rates have generally increased. This reveals something important: the efficiency problem existed before AI. The channels were always paying for low-intent traffic. AI automation simply exposes which traffic deserves budget and which does not.
Accounts that saw the worst performance shifts were typically those spending heavily on informational keywords—the query types now dominated by AI Overviews. These accounts weren't harmed by AI Max. They were harmed by budget allocation decisions that were always inefficient.
The informational search challenge is real
Informational queries, those beginning with "how," "what," "why", have experienced significant pressure. AI Overviews now answer many of these questions directly, reducing the incentive to click any result. For accounts relying heavily on this traffic type, the shift represents a material challenge.
However, the solution is not to abandon paid search. The solution is to recognize that informational keywords were always low-converting. Accounts that moved budget toward high-intent, transactional keywords improved performance noticeably.
The market stabilization effect
The market for paid search has stabilized around a new equilibrium. The platforms have been training AI models for years; the volatility of 2024-2025 has largely settled. Paid search has emerged as one of the more predictable channels available, particularly for accounts with clear conversion tracking and intent-based keyword strategies.
Critically, AI Overviews appear on roughly one-third of informational searches but only around 5% of transactional searches. The channel structure is not broken. It is different.
Monetization and pricing
The advertising platforms continue to test monetization approaches for new formats like AI Mode placements. Historical precedent from the mobile shift and from voice assistant advertising suggests that new inventory typically monetizes at lower rates initially, then improves as platforms optimize. This is a temporary pricing cycle, not a permanent discount.
The bottom line: Bottom-funnel PPC performs better than industry commentary suggests. Accounts that placed half their budget on informational keywords experienced visible losses, but this revealed inefficiency that existed previously. AI did not create the weakness; it exposed it.
What people in the trenches are saying
Perspectives vary depending on where insights are sourced. Corporate blog posts suggest AI has made PPC effortless. PPC community forums suggest it has become impossible. The reality typically emerges from less-polished sources.
The most common complaint; by a significant margin, involves the black box problem. Every automation discussion thread converges on the same challenge: Where did the budget go, and why did performance change month to month? This issue predates AI automation and remains a persistent concern for marketing professionals managing automated systems.
Community analysis of AI automation test results reveals interesting patterns. Ecommerce accounts typically show increased conversion value but rising cost per acquisition, with only a portion achieving their ROAS targets. Lead generation accounts show more variable results, with performance depending heavily on implementation approach and tracking configuration.
An additional pattern has emerged: invalid traffic often increases following automation activation in accounts without proper traffic quality monitoring. This is not an automation problem; it is a monitoring problem. Accounts with weekly review practices corrected invalid traffic issues quickly. Accounts without that discipline experienced persistent inefficiency.
A clear pattern emerges: professionals whose value derived primarily from tactical lever-pulling express the most frustration. Professionals demonstrating strong knowledge of client margins, lead quality, and conversion value report managing the transition successfully.
Where the machine wins, and where it really doesn't
AI automation has genuine strengths and consistent limitations.
Where algorithms excel
Smart Bidding sets more effective bids than manual approaches. The system evaluates device, location, time, and audience signals in every individual auction—inputs a human operator cannot process simultaneously. The AI tools also generate credible ad copy variations in seconds and identify long-tail query opportunities that would require extensive manual research. On a well-structured account, resistance to these capabilities largely represents ego rather than strategic necessity.
The machine is genuinely better at tactical optimization when the strategic framework is sound.
The critical limitation
The algorithm pursues whatever metric is defined as success. It never questions whether the definition was correct. If junk form submissions are counted as leads, the system finds more junk. If ROAS targets ignore product margin, the algorithm happily scales the lowest-margin products. If conversion tracking includes incomplete transactions, the system optimizes for incomplete conversions.
The system cannot know what the business actually values unless someone sends that information into the platform.
Lead generation has remained the weak spot in every automation test. This occurs because lead generation success depends on qualitative judgments about lead quality that AI cannot replicate without explicit human instruction. A "lead" is only valuable if it matches specific criteria—industry, company size, decision-making authority, budget. The algorithm must receive that definition from a human.
The split in practice
The machine handles bidding, ad copy volume, and query discovery. Humans determine what constitutes a valuable conversion, integrate offline data sources, allocate budget between campaigns, optimize landing pages and offers, and explain results to financial leadership.
An account audit conducted in 2026 demonstrated this principle clearly. AI automation showed strong metrics on every dashboard. Revenue from paid search had not changed in six months. The system was functioning exactly as designed. Nobody had defined what "success" actually meant.
What a practitioner would do with an account tomorrow
If given a Google Ads account tomorrow morning, the recommended sequence would be:
First priority: Establish proper tracking
Get offline conversions and qualified-lead stages flowing back into the system. Bid on value rather than raw conversion counts. Define what constitutes a valuable conversion, not just any conversion. Every other decision depends on this foundation.
Second: Sort keywords by intent
Brand and high-intent terms require protection and investment. Informational terms must demonstrate actual conversion history; most will not qualify for continued spending. Categorize the account by query intent before making automation decisions.
Third: Test AI Max strategically
Run AI Max as a controlled experiment, not a Friday afternoon switch. Maintain the core exact and phrase match keywords. Monitor search term reports and invalid traffic weekly.
Fourth: Invest in creative quality
When targeting is automated, copy, images, and video determine differentiation. Review every headline the automation generates before publication. The machine generates variations; humans determine which deserve to run.
Fifth: Get invited to the SEO meeting
Being cited in AI Overviews comes from content authority and domain expertise. This also correlates with paid search performance. Accounts whose brand content ranks and gets cited perform better in both channels.
Sixth: Test new platforms deliberately
ChatGPT ads warrant small-budget testing for any business where customers think before purchasing. Evaluate results like discovery advertising, not search. Expect different conversion patterns and longer consideration cycles.
Seventh: Build the right skills for the next decade
For professionals early in their careers, learning fundamentals remains essential. Understanding how systems work enables proper auditing. After that, focus on analytics, CRM integration, conversion definition, and testing design. These skills are what the market will reward.
So, PPC vs AI?
The framing of this as a conflict was likely incorrect from the beginning. AI became the engine of PPC. Those who performed the engine's functions got repositioned to driver roles, whether they sought advancement or not.
The landing points tell a clear story. Clicks cost more than they did in 2016. Leads cost less. Top-of-funnel reach has shifted. Bottom-funnel performance has stabilized. Google's newest ad slots are exclusively automated. This does not sound like channel death. It sounds like a channel that shifted who does what.
Over the next decade, PPC job openings for entry-level positions will likely decrease further. The industry will still require people who understand the strategic side. It simply will not need people performing the tactical roles of 2016.
