ChatGPT, Copilot, AI Overviews, X — four AI engines now sell ad inventory. How they compare to Google Ads on CPM, CPC, conversion rate, and where to send each dollar in 2026.
Generative Engine Advertising (GEA) is the paid counterpart to GEO (Generative Engine Optimization) — sponsored placements served inside AI engines rather than organic citations earned through schema and content. Four AI engines now sell GEA inventory in production as of June 2026:
| Channel | Buying interface | Format | Status |
|---|---|---|---|
| ChatGPT Ads | ads.openai.com | Sponsored text, 240-char, end of response | Self-serve, no minimum (May 2026) |
| Microsoft Copilot Ads | Microsoft Ads + PMax | Sidebar + inline + Showroom | Live, 73% higher CTR than search |
| Google AI Overviews Ads | Google Ads + PMax + Shopping | Sponsored inside AIO block | Live, 25.56% of AIO results (Oct 2025) |
| X Ads / Grok integration | business.x.com | Standard X formats + Grok dev | Standard X; Grok integration developing |
And one channel that’s officially dead: Perplexity Ads formally abandoned advertising in February 2026 citing user-trust concerns. Anthropic Claude has an official “no ads” stance. Meta AI has no ad inventory inside conversations as of June 2026.
That leaves the four above as the channels that matter.
The numbers we’re seeing across active retainer accounts in June 2026:
| Metric | Google Ads | ChatGPT Ads | Copilot Ads | AIO Ads |
|---|---|---|---|---|
| CPM range | $40–100 | $25–45 | $30–60 | Auto-allocated |
| CPC range | $1–50+ | $3–5 | 15–25% cheaper than Google | Same as Google |
| CTR typical | 3–6% | 0.8–1.4% | +73% vs Bing search | Higher than SERP |
| Cost-per-conversion | Lowest commercial intent | $45–80 DTC | Bing equivalent + 16% lift | AIO inventory-specific |
| Targeting model | Keyword + audience | Conversation topic | Conversation + LinkedIn data | Asset group + feed quality |
| Best for | Bottom-funnel intent | Pre-Google research | B2B + bundled with Google | Comparison queries |
A few things stand out:
Google Ads CPM is the highest because users are at the bottom of the funnel. They’ve already decided to buy; the question is which brand. ChatGPT Ads CPMs are lower because users are mid-research — but the cost-per-conversion math still favors Google for transactional queries.
CTR on ChatGPT is lower than Google because the placement is at the bottom of an AI answer, not above the SERP. But the CPC is also lower, so cost-per-conversion math doesn’t follow the CTR pattern.
Copilot’s 73% CTR lift over traditional Bing search is the cleanest “free uplift” available in 2026. If you’re running Microsoft Ads at all, you should be running Copilot ads.
AIO ad inventory grew 5× from March to October 2025 per Semrush. The growth rate matters more than the absolute share — this is the fastest-growing paid surface in search.
Don’t kill your Google Ads.
Commercial-intent search clicks. When a user types “buy [product]” or “best [category] for [use case]”, they’re at the bottom of the funnel. Google Search captures this; ChatGPT does not, because the user has bypassed ChatGPT and gone directly to the buying decision.
Branded search. ChatGPT does not show ads on user prompts that mention specific brand names (yet). Your Google branded-search defense remains essential.
Shopping ads. Google Shopping + Merchant Center is still the dominant paid surface for product-led commerce. Microsoft Shopping is catching up via Copilot Showroom, but Google’s product feed integration is deeper.
YouTube + Display. Visual + video formats. ChatGPT Ads remain text-only as of June 2026 (rich-media is on the roadmap for late 2026).
For most brands spending $25K+/mo on paid media, Google Ads should remain 40–60% of the budget. The discussion is what to do with the remaining 40–60%, not whether to shift Google.
Pre-funnel research reach. Users asking ChatGPT “what should I look for in a CRM” or “best ways to cook [ingredient]” are pre-Google, pre-Amazon, pre-Meta. ChatGPT Ads reach them at the moment of category formation, before they’ve decided a brand.
Sub-market CPMs. $25–45 CPM today is half what equivalent reach costs on Meta or LinkedIn. The first-mover window is real and closing.
Conversation-topic targeting. No demographic guessing, no third-party data dependency, no audience-modeling drift. The user told ChatGPT what they’re researching; OpenAI matches your ad to that topic. Targeting precision in a privacy-safe model.
For DTC + B2B SaaS specifically, this is where the incremental reach lives. We see DTC retainer clients allocating 15–25% of their paid budget to ChatGPT Ads within 90 days of testing.
Cheaper than Google for the same intent. Microsoft Ads CPCs run 15–25% lower than Google for equivalent commercial-intent queries. This has been true for years; Copilot inherits the same pricing advantage.
73% higher CTR than traditional Bing search. This is Microsoft’s published number from Nov 2024–May 2025 data. If you’re running Microsoft Ads, the Copilot uplift is the cleanest free performance gain in 2026.
LinkedIn-data audience layer. Microsoft Audience Network uses LinkedIn profile data for B2B targeting. Copilot inherits this. For B2B SaaS, this is the only channel that combines AI-conversation surface with LinkedIn-grade audience targeting.
Showroom format. Interactive AI-powered ad surface where users explore your product range conversationally. Most agencies skip Showroom because it requires interactive creative briefs. Brands that ship Showroom get a meaningful CTR advantage over text + image competitors.
Recovering lost organic clicks. If AI Overviews are answering queries you used to rank for, AIO ads buy back the slot. Pair with AEO recovery for organic + paid coverage.
Comparison-heavy verticals. Insurance, B2B SaaS, finance, healthcare, education. “Best [X] for [Y]” queries trigger AIO at the highest rate, and AIO ads compete for placement inside the AI answer.
PMax-native. AIO inventory is auto-allocated from existing Google Ads + PMax + Shopping campaigns. No separate platform to learn. The optimisation work is asset-group structure + feed quality + audience signals — work most accounts haven’t done.
Fastest-growing paid surface in search. 5× inventory growth from March to October 2025 per Semrush. If you’re not actively optimising for AIO eligibility, you’re losing share to competitors who are.
The 2026 allocation we’d suggest for a typical $50K/mo media budget:
| Channel | Allocation | Why |
|---|---|---|
| Google Ads (Search + PMax + Shopping) | 40% | Bottom-funnel commercial intent + branded defense |
| Google AIO ads (within Google) | 10% (PMax overlap) | Recover AIO-lost organic clicks |
| Microsoft + Copilot Ads | 15% | 15–25% CPC savings + 73% Copilot CTR lift |
| ChatGPT Ads | 15% | Pre-funnel research reach, sub-market CPMs |
| Meta Ads (paid social, retargeting) | 15% | Visual + retargeting + community |
| Reserve / experimental | 5% | X, TikTok, new channels |
This is a baseline, not a prescription. Industry, funnel position, and creative production capacity shift the allocation. For B2B SaaS, shift Meta down and AIO + Copilot up. For DTC, shift Meta up and AIO down. For early-stage brands, shift Google down and ChatGPT up.
If you’re testing AI engine ads incrementally, here’s the plan we’d run:
Days 1–14 — Setup. ChatGPT Ads + Copilot Ads accounts provisioned. AIO eligibility audit on existing PMax. Creative pack for all three channels. Server-side tracking deployed.
Days 15–30 — First campaigns live. Daily monitoring, weekly creative iteration. Establish baselines for CTR, CPC, cost-per-conversion per channel.
Days 31–60 — Incrementality testing. Geo-holdout test on ChatGPT Ads (do test cells convert more than holdout?). Compare Copilot vs traditional Bing search lift. AIO share-of-voice baseline.
Days 61–90 — Allocation rebalance. Based on incrementality lift, rebalance budget across channels. Cancel underperforming sub-channels. Increase budget on channels showing true lift.
Day 91+ — Steady-state retainer. Weekly creative iteration, monthly business review, quarterly cross-channel allocation review.
This is what we run for retainer clients. The Sprint version compresses days 1–30 into the first week.
Google Ads is not going away. ChatGPT Ads is not replacing it. Both will coexist in 2027 paid media stacks, and brands running both with proper attribution will outperform brands running only one.
The question isn’t “should I switch to ChatGPT Ads?” The question is “where do I add ChatGPT, Copilot, and AIO to my existing Google + Meta stack — and what cross-channel attribution do I need to measure true lift?”
If your team can build this operational muscle in-house, great. If you’d rather work with someone who already has 8 active retainers across the four channels — that’s what we do. The first-mover window is closing. Start before December.
Raj founded Digital Marketing Agency For after 12 years running SEO, AEO, paid media, and lifecycle email programmes for B2B SaaS, DTC, and FinTech brands across the US, UK, and India. Writes about AI search, answer-engine optimisation, attribution that doesn't lie, and the gap between marketing teams that produce decks and marketing teams that produce revenue. Based remote-first; embedded in client pods across six time zones.
Not yet — and probably not at all for most brands. Google Ads still wins commercial-intent search clicks at the bottom of the funnel (transactional queries like 'buy [product]', 'best [category] for [use case]'). ChatGPT Ads reach users earlier in research — pre-Google. The right strategy is layering ChatGPT Ads incrementally on top of Google Ads, not shifting budget.
Per impression, yes — for now. ChatGPT CPMs are $25–45 vs Google equivalent $40–100; ChatGPT CPCs are $3–5 vs Google range $1–50+; Copilot CPCs run 15–25% cheaper than Google for equivalent intent. But CTRs and conversion rates differ by channel: Google's bottom-funnel click conversion rates are still highest because the user is actively searching to buy. AI engine ads are cheaper-per-click but reach users earlier — the cost-per-conversion math varies by category.
AIO ads first (sponsored placements inside Google AI Overviews on comparison queries — 'best [X] for [Y]'), then Copilot (LinkedIn-data audience overlay), then ChatGPT (category research reach). Skip X for B2B SaaS unless your buyer is technical and active on X.
Server-side tracking via GTM Server-Side + each platform's conversion pixel + Google Enhanced Conversions cross-reference. Then geo-holdout incrementality testing monthly to measure true lift vs rebalanced spend. Don't trust last-touch dashboards across AI engine ads — they're systematically wrong for upper-funnel channels.
No. Perplexity formally abandoned advertising in February 2026 citing user-trust concerns. The program is dead. Don't waste cycles.
