On May 20, Google used its Google Marketing Live 2026 keynote to introduce Ask Advisor, a Gemini-powered agent that spans Google Ads, Google Analytics, Merchant Center, and DV360 from a single conversational interface. It is now the most ambitious AI advertising product Google has shipped — and the clearest signal yet of where the company wants buyers to spend their time.
Two weeks earlier we covered DSA's migration to AI Max. A month earlier, Meta's AI Connectors and TikTok's MCP server. Ask Advisor is Google's answer to the same question: what does an ad platform look like when the primary interface is a chat box?
Below: what Ask Advisor is, what it can actually do today, where the keynote oversold it, and how to put it to work without handing over the steering wheel.
What Google Announced
Ask Advisor is positioned as a unified Gemini agent with shared memory across Google's marketing stack. According to Search Engine Land's GML coverage, Ask Advisor is currently in beta for English-language accounts globally, with a broader rollout expected later this year.
The product covers four surfaces:
- Google Ads. Campaign briefs, asset suggestions, performance diagnostics, and structured next-step actions inside the Google Ads UI.
- Google Analytics (GA4). Audience discovery, segment construction, and conversion analysis using natural-language prompts grounded in your property's data.
- Merchant Center. Product feed health, attribute suggestions, and product-to-audience mapping for Shopping and PMax inventory.
- Display & Video 360. Programmatic planning, deal-level summaries, and inventory recommendations for DV360-managed buys.
The differentiator versus prior Google AI tools — Performance Max suggestions, the GA4 "Insights" panel, the Merchant Center quality scorecards — is that Ask Advisor carries context across all four products. PPC Land's writeup calls this out as the central change: ask about a GA4 audience, request a Merchant Center segment built from that audience, and launch a campaign — without switching tabs or re-explaining the question.
What's Actually New
Google has shipped chat-style assistants in Ads and Analytics for over a year. Ask Advisor is meaningfully different on three axes.
1. One model, four product graphs
Ask Advisor is the first time Google has exposed a single agent that can read across the Ads, Analytics, Merchant Center, and DV360 schemas in one conversation. The technical lift is non-trivial — those products use different ID systems, different attribution models, and different freshness windows. The user-facing result: you can pose a question that crosses two products ("show me Merchant Center products with falling impression share that match the audience GA4 calls 'high-LTV bookkeepers'") and get a coherent answer.
2. Shared memory across tasks
Earlier Gemini surfaces in Google Ads were per-prompt — every request started cold. Ask Advisor retains context. A campaign brief you wrote on Tuesday informs the headline suggestions on Thursday. This matters less for one-off questions and a lot for the multi-step planning work where most buyers actually want help.
3. Action-oriented, not advisory-only
The "Advisor" name is slightly misleading. The agent doesn't just suggest — it can stage actions for you to approve: a new campaign draft, an audience exclusion, a Merchant Center attribute fix, a DV360 inventory swap. You stay in the approval loop, but the surface area between "the agent has an idea" and "the change is live" is smaller than it's ever been on Google's stack.
What It Does Well
Based on the keynote demos and early beta reports, Ask Advisor's strongest use cases are the workflows that already crossed product boundaries — and were painful because of it.
Audience-first campaign building
The flow Google demoed on stage — define an audience in GA4, map it to Merchant Center products, launch a Search/Shopping campaign — used to require three browsers, two spreadsheets, and a calendar invite with the analytics team. Ask Advisor compresses it into a single prompt chain. For agencies that have analysts and media buyers as separate roles, this is the biggest workflow change in years.
Product feed triage
Merchant Center has always told you that a feed has problems. Ask Advisor is the first surface that explains why a SKU is underperforming in plain language and offers a feed-level fix, not just a UI link. For ecommerce accounts with thousands of SKUs, the diagnostic time savings alone justify trying the beta.
Cross-product reporting
Combining GA4 conversion paths with Google Ads spend and Merchant Center inventory data has been a months-long warehouse project for most teams. Ask Advisor doesn't replace a real BI stack, but it gives a directional answer in 30 seconds — useful for the question that comes up in a Tuesday standup, not the quarterly board deck.
Where the Keynote Oversold It
The launch coverage has been uniformly enthusiastic. Here are the four things worth knowing before you bet workflow on Ask Advisor in the next 60 days.
It's a beta, and English-only
"Available globally" in the announcement means English-language accounts only. Non-English markets — much of EMEA and APAC — are not in the beta. If your team operates multi-language campaigns, half your stack is still on the old surfaces, and the time you save in English you spend reconciling differences across languages.
The four-product integration is uneven
Google Ads and Analytics are the most mature surfaces. Merchant Center is solid for feed diagnostics, thinner for inventory strategy. DV360 — the most complex product in the stack — is the least developed integration, and most agency DV360 traders we've spoken to are skeptical the agent has the context depth needed for real programmatic planning. Treat DV360 coverage as a v1, not a v3.
Shared memory is also a shared blast radius
Ask Advisor's memory is account-scoped. A misremembered objective from a Monday prompt can show up in Friday's campaign brief. There is no audit log of "facts the agent currently believes about my account" — the closest equivalent is scrolling chat history. For agencies managing multiple clients in the same Google account, this is the gap that needs the most attention.
It runs inside Google's product, not your stack
The TikTok MCP server and Meta's AI Connectors expose programmatic APIs that external agents — Claude, ChatGPT, custom workflows — can call. Ask Advisor is a UI feature inside Google's products. There is no public API. If your team has been building a multi-platform AI ad ops stack with MCP at the center, Ask Advisor is a peer product, not a node in your graph. That may change; today it hasn't.
How It Compares to Meta and TikTok
| Google Ask Advisor | Meta AI Connectors | TikTok Ads MCP | |
|---|---|---|---|
| Interface | Chat inside Google products | External agents (Claude, ChatGPT) via connector | MCP server consumed by external agents |
| Cross-product scope | Ads + GA4 + Merchant Center + DV360 | Ads + Pages + WhatsApp Business (read-mostly) | TikTok Ads only |
| Shared memory | Yes, account-scoped | Per-session in the external agent | Per-session in the external agent |
| Programmatic access | No public API at launch | Connector-only, no MCP yet | Open MCP server |
| Best for | Teams living inside Google's UI | Buyers running Meta from existing AI workflows | Agent-native ad ops, custom builds |
The strategic split is now clear. TikTok bets on agents you bring. Meta bets on connectors into the agents buyers already use. Google bets on its own UI becoming the agent. Each path has tradeoffs; none of them is wrong. The mistake is assuming one of these patterns is going to win and committing your workflow to it before the rest of 2026 plays out.
How to Use Ask Advisor This Quarter
If you have beta access — or get it when the rollout widens — the highest-leverage way to evaluate it:
- Pick one cross-product workflow you do every month. Audience-to-campaign, feed triage, post-mortem on a launch — anything that used to live in three tabs. That's where Ask Advisor will demo best.
- Time it against your current process. Not a vibe check — actual minutes for the same output. The agent saves time in some places and burns it in others (especially when its first answer is subtly wrong). You need data before you change a workflow.
- Treat staged actions as drafts, not decisions. The approval step is your only audit trail. Read every change before clicking apply, especially in the first month. Beta agents make confident mistakes.
- Keep a "things the agent got wrong" log. Beta feedback is the only way Google fixes the gaps that matter to your account. The teams shaping the v2 are the ones writing actual bug reports now.
- Don't abandon your multi-platform plan. Ask Advisor is excellent inside Google. It does nothing for Meta, TikTok, Reddit, or LinkedIn. If your media mix crosses platforms — and almost everyone's does — the cross-platform layer still has to live somewhere else.
The Bigger Picture
Google, Meta, and TikTok have now each shipped their version of the "AI runs the ad platform" pitch within six weeks of one another. The architectures differ, but the buyer's job is converging: less time clicking through dashboards, more time defining objectives, guardrails, and measurement. The platforms supply the automation. The leverage stays with whoever wraps it in policy and a measurement story that doesn't depend on the platforms grading their own homework.
Ask Advisor is the most ambitious of the three launches because of the scope it covers in a single product. It's also the most locked-in: there is no MCP, no external API, no way to use Ask Advisor from your own agent stack. If you live inside Google's UI, that's fine. If you're building a cross-platform AI ad ops layer, Ask Advisor is a feature you evaluate, not a foundation you build on.
Either way: the version of Google Ads where humans pick keywords and write headlines is, quietly, already gone.
One agent across Google, Meta, and TikTok
Ads Agents wraps Ask Advisor, Advantage+, and Smart+ in a single workflow — with the guardrails, exclusions, and weekly reviews none of them ship by default.
Get Started Free →