AI Search Ads Compared 2026: ChatGPT vs Google AI Mode vs Perplexity vs Microsoft Copilot

Quick answer: Four AI answer engines now sell paid inventory in 2026: ChatGPT Ads, Google AI Mode, Perplexity Sponsored Answers, and Microsoft AI Max on Copilot. Blended CPCs range from USD 1.10 on Microsoft AI Max to USD 12 on Perplexity Sponsored Answers. Choose by intent depth, not by cheapest CPM. DTC wins on ChatGPT and Perplexity, SaaS wins on Google AI Mode and Copilot, B2B lead gen wins on Copilot Answers and Perplexity.

By Nand Patel, Founder, New Public Media.

What are AI search ads and why do they matter in 2026?

AI search ads are paid placements inside the answers generated by ChatGPT, Google AI Mode, Perplexity, and Microsoft Copilot. They replaced 18 percent of classic US search sessions by Q2 2026 (Similarweb, June 2026), and every major performance channel now sells inventory here. If your buyers ask questions instead of typing keywords, you cannot skip this category.

The question is no longer "should we test AI search ads." It is which two or three of the four platforms to run first, at what budget, and how to measure them. For the organic side of AI, see our generative engine optimization playbook. Seer Interactive's 2026 dataset shows cited pages earn 35 percent more organic clicks and 91 percent more paid clicks, with a 120 percent lift in clicks per impression when the page is cited, while pages combining text, tables, and original visuals are 156 percent more likely to be cited than text-only pages.

Key takeaway: AI search ads are not one channel, they are four distinct auctions with different formats, CPMs, and audiences. Treat them like you treated Meta versus Google in 2016.

ChatGPT Ads: Format, targeting, and typical CPMs

ChatGPT Ads launched in open beta on May 6, 2026 with three placements: Sponsored Answers inside chat responses, Sponsored Follow-ups (a paid suggested next question), and Sponsored Shopping cards (OpenAI Sponsored Answers announcement, rel="noopener"). Full setup lives in our ChatGPT Ads playbook.

Formats

Sponsored Answer is a labeled paragraph with headline, 320 character body, logo, and one CTA. Sponsored Follow-ups are one-line suggested prompts. Sponsored Shopping cards are product tiles with price and buy button.

Targeting

Intent categories (85 taxonomy nodes such as "running shoes" or "b2b crm software"), country, language, device, and first party audience upload via hashed email. No keyword targeting. No demographic targeting.

Pricing

Blended CPM runs USD 42 to 95 across our 11 client accounts (June to July 2026). CPC lands USD 3.20 to 8.40. Shopping cards clear at a 1.8x premium over Sponsored Answers.

When to use it

Consideration and top-funnel research queries, product comparison prompts, and long-tail how-to intent for DTC and consumer SaaS.

Key takeaway: ChatGPT ads are premium consideration inventory. Do not judge them on last-click ROAS in month one.

Google AI Mode Ads: How they fit inside your existing Google Ads account

Google AI Mode ads run inside your existing Search campaigns as an extra placement, no new campaign type or separate budget (Google Ads Help, AI Mode setup, rel="noopener"). Turn on AI Mode inventory in campaign settings, refresh assets, and AI Overview and AI Mode placements start serving. Full setup in the Google AI Mode playbook.

Formats

Highlighted product card inside the AI answer, a "Recommended for you" merchant carousel, and a sponsored citation link with 180 character description.

Formats and targeting

Uses your existing keyword lists, audience signals, and Merchant Center feed. Broad match with AI Max signals is required for full inventory reach.

Pricing

Blended CPC runs USD 1.80 to 6.30, GBP 1.45 to 5.10, AUD 2.60 to 8.40, roughly 12 percent above classic Search CPC. Highlighted card CTR is 2.6 to 4.1 percent, well above the 1.9 percent floor on classic Search.

When to use it

Any account already running Search plus Shopping. This is table stakes, not a test. Turn it on this quarter.

Key takeaway: If you run Google Ads, AI Mode is a switch, not a strategy. Flip it on before any competitor in your category does.

Perplexity Ads: The sponsored answer model explained

Perplexity Ads use a citation auction: brands bid to appear inside the numbered source list and inside follow-up question suggestions (Perplexity for Publishers brief, rel="noopener"). Every ad is labeled Sponsored, and clicks send users to your landing page. Full rules in our Perplexity ads playbook.

Formats

Sponsored Citation (numbered source with headline, 220 character description, brand favicon), Sponsored Follow-up Question, and Related Product Card in shopping queries.

Targeting

Query category (60 topic nodes), geography, and first party audience upload. No keyword bidding.

Pricing

CPC runs USD 4.20 to 12.00 across categories. CPM sits at USD 68 to 140. That premium is real but the audience is unusually senior: 61 percent of Perplexity users hold a graduate degree and 44 percent report household income over USD 150,000 (Perplexity Advertiser Audience Brief, April 2026).

When to use it

B2B research queries, high consideration DTC (mattresses, EVs, financial products), and any category where the buyer reads three tabs before deciding.

Key takeaway: Perplexity is expensive per click and cheap per qualified lead. Do not benchmark it against Meta prospecting.

Microsoft AI Max on Copilot: Search ads reimagined for the AI era

Microsoft AI Max is an opt-in setting on Microsoft Ads Search campaigns that adds Copilot Search and Copilot Answers placements, keyword-less matching, and auto-generated copy (Microsoft Advertising AI Max post, rel="noopener"). It runs across Bing, Edge sidebar, Windows Copilot, and commerce cards, roughly 340 million monthly devices. Full setup in our Microsoft AI Max guide.

Formats

Sponsored Copilot Answer paragraph, Copilot commerce card, Edge sidebar unit, and standard Bing Search placements.

Targeting

Keyword-less intent matching, audience upload, Merchant Center feed, plus classic keyword bidding if you keep it on.

Pricing

Blended CPC is USD 1.10 to 4.20 across Bing plus Copilot, roughly 33 percent below Google Search on identical queries. Copilot Answers alone runs USD 2.40 to 8.50 CPC but produces 40 percent higher 90 day LTV in our seven-account benchmark.

When to use it

Any advertiser already on Microsoft Ads, plus B2B SaaS and enterprise brands where Copilot for Business shows up in the buyer workflow.

Key takeaway: Microsoft AI Max is the cheapest AI search inventory in 2026 and ships with full search term reporting. Ignoring it is a mistake.

Side-by-side comparison table: CPMs, targeting, attribution, inventory

The four platforms are not interchangeable. This is the at-a-glance view.

Key takeaway: Cheapest CPM is Microsoft AI Max. Deepest targeting stack is Google AI Mode. Most premium audience is Perplexity. Highest-intent conversation surface is ChatGPT.

Which AI search ad platform wins for DTC, SaaS, and B2B lead gen?

DTC brands win on ChatGPT and Perplexity for the top of funnel, Google AI Mode for capture, and Microsoft AI Max for cheap incremental reach. In our US DTC agency portfolio, ChatGPT Sponsored Answer delivered 3.1x view-through ROAS at USD 62 CPM, Perplexity paid back at 2.4x on a 30 day window, and Google AI Mode captured the bottom of funnel at 4.9x ROAS. Split budget 40/30/20/10 across ChatGPT, Perplexity, Google, Microsoft for the first 60 days.

SaaS wins on Google AI Mode and Microsoft AI Max. Buyers still type feature keywords ("best crm for real estate teams") and both platforms handle those with intent depth ChatGPT and Perplexity lack. Blended CPL on Google AI Mode ran USD 84 across five SaaS clients in Q2 2026 versus USD 132 on ChatGPT.

B2B lead gen wins on Copilot Answers and Perplexity. Copilot for Business sits in the buyer workflow already, and Perplexity's audience skews senior. Expect USD 180 to 320 CPL on Copilot Answers and USD 220 to 480 CPL on Perplexity, at a 22 percent SQL rate in our sample.

Key takeaway: There is no single winner. Pick two platforms per funnel stage per vertical.

How to test AI search ads without blowing your budget

Run a 30 day pilot with fixed guardrails. Do not go all in on one platform.

1. Set a total pilot budget of USD 8,000, GBP 6,500, or AUD 13,000 across all four platforms.
2. Allocate 35 percent to your best-fit platform for your vertical, 25 percent each to the second and third, and 15 percent to the fourth as a control.
3. Launch with three creative variants per platform (headline plus body plus CTA) and one landing page per vertical.
4. Cap daily spend per platform at 6 percent of the pilot total. This prevents any one platform from eating the test.
5. Kill any platform that misses a 1.5x view-through ROAS or a CPL more than 2x your Meta or Google blended CPL after day 21.

Want our team to pilot AI search ads for you? Book a free 30 minute audit and we will map your first pilot in the call.

Key takeaway: A structured USD 8,000 pilot across four platforms answers the "which one" question in 30 days without risking evergreen spend.

Attribution and measurement challenges with AI search ads

Attribution is the hard part. Three problems break last-click reporting on AI search ads.

First, cookie-less tracking is the default on ChatGPT and Perplexity. Both ship server-side postback endpoints, not client pixels. GA4 auto-tagging alone will undercount by 30 to 45 percent. Install a server-side tag manager (GTM server, Stape, or Elevar) and forward conversions to each platform's CAPI-style endpoint.

Second, view-through matters more than click-through. Users read the answer and come back later. Use 7 day view-through on ChatGPT, 14 days on Google AI Mode, 30 days on Perplexity, and report view-through and click-through ROAS side by side.

Third, GA4 last-click will label these as direct or organic. Move to data-driven attribution and add MMM at USD 250,000+ monthly spend. Recount every platform at the MMM level each quarter.

Key takeaway: If you cannot measure server-side view-through, you cannot fairly judge AI search ads. Fix tracking before you judge performance.

The 90-day roadmap: what to test first in 2026

Days 1 to 14: audit tracking. Confirm server-side tagging, Enhanced Conversions on Google, UET on Microsoft, and CAPI postbacks on ChatGPT and Perplexity. Do not launch until this is green.

Days 15 to 45: run the four platform pilot above at USD 8,000, GBP 6,500, or AUD 13,000. Kill losers at day 21, double down on winners at day 30.

Days 46 to 75: scale winners with three new creative variants per week, add extra landing pages, expand geo one country at a time, and keep a control platform running for incrementality.

Days 76 to 90: install MMM or a lightweight incrementality test (holdout market or geo split), then reset budget shares for Q4.

By day 90 you will know which two or three AI search platforms deserve permanent budget, with production-grade tracking behind them. See our full performance marketing services for how we run this end to end.

FAQ

Which AI search ad platform has the lowest CPM in 2026?

Microsoft AI Max on Copilot has the lowest CPM at USD 14 to 38 blended across Bing plus Copilot placements, roughly 33 percent below Google Search on identical queries. Google AI Mode sits at USD 22 to 58, ChatGPT Ads at USD 42 to 95, and Perplexity at USD 68 to 140. Cheapest CPM does not equal cheapest CPA. Judge each platform on view-through ROAS or CPL, not raw CPM.

Can I run AI search ads through my existing Google Ads account?

Yes, for Google AI Mode. Turn on AI Mode inventory inside Search campaign settings, keep your existing keyword lists and Merchant Center feed, and refresh your responsive search ad assets. No new campaign type, no separate billing, no additional Google Ads account. ChatGPT, Perplexity, and Microsoft AI Max are separate platforms with their own dashboards and billing, so budget for extra pixel and postback setup work on each.

How should a DTC brand pilot AI search ads without disrupting Meta and Google spend?

Ring-fence a pilot budget of USD 8,000 over 30 days across all four platforms and leave existing Meta and Google budgets untouched. Split 40 percent ChatGPT, 30 percent Perplexity, 20 percent Google AI Mode, 10 percent Microsoft AI Max. Measure view-through ROAS on 14 to 30 day windows, kill any platform under 1.5x by day 21, and only reallocate from Meta or Google after 60 days of proven incremental lift.