---
title: "3 Ads MCPs Ranked by What You Can Verify Before Launch"
description: "groas, Madgicx and AdSparky ranked by their route from AI proposal to approved change—and the platform-defaults check none gets to skip."
url: "https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/14aef7c8-92c1-4938-ba3b-eba7c9374992.png"
published: "2026-11-22T06:19:48.055Z"
modified: "2026-11-22T06:19:48.104Z"
---

November 22, 2026 · 11 min read

# 3 Ads MCPs Ranked by What You Can Verify Before Launch

[](https://groas.com/author/alexander-perelman)

[Alexander Perelman](https://groas.com/author/alexander-perelman)Personally Tested

Head Of Product @ groas

[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![Cutaway rocket on a launch pad with labelled approved compartments, plus a hidden hatch where two unrequested switches are already flipped on.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/14aef7c8-92c1-4938-ba3b-eba7c9374992.png)

In this article

1. [The comparison: execution scope is not proof of launch settings](#the-comparison-execution-scope-is-not-proof-of-launch-settings)
2. [1. groas: my pick for a plan that crosses channels](#1-groas-my-pick-for-a-plan-that-crosses-channels)
3. [2. Madgicx MCP: for confirmed edits to live Meta campaigns](#2-madgicx-mcp-for-confirmed-edits-to-live-meta-campaigns)
4. [3. AdSparky: a free-beta evaluation with explicit guardrails](#3-adsparky-a-free-beta-evaluation-with-explicit-guardrails)
5. [The Meta v27.0 check none of the three gets to skip](#the-meta-v270-check-none-of-the-three-gets-to-skip)
6. [The prompt I would run before the first creation call](#the-prompt-i-would-run-before-the-first-creation-call)
7. [Why Google’s official Ads MCP is not ranked](#why-googles-official-ads-mcp-is-not-ranked)
8. [The final ranking: choose the missing execution step](#the-final-ranking-choose-the-missing-execution-step)

**groas is my pick if your AI-drafted campaign needs to move from chat into several ad platforms.** But I would block launch with any tool on this list if it could not establish which settings would actually go live.

That is the gap most “AI ad tool” roundups skip. Your assistant drafts the structure, budgets, three ad angles and a negative list by 11am. At 6pm, the plan is still in the chat window. Generating another version is not the missing step. Executing the right version is.

I rank these ads MCPs by their documented route from proposal to approved change: **groas first, Madgicx second, AdSparky third**. I care about what each can write, where approval happens, and whether you can establish what the platform will add around the change you requested.

These are documentation-based verdicts, not hands-on test results. Unverified handling of a required setting is a launch blocker, not an invitation to assume it works. That distinction matters after Meta’s [Marketing API v27.0](https://developers.facebook.com/documentation/development/graph-api/changelog/version27.0) introduced creation defaults that can widen delivery unless the integration explicitly opts out.

## The comparison: execution scope is not proof of launch settings

Before I give an assistant write access, I want answers to six questions:

- Which account actions can it execute?
- Can it create campaigns, or only edit existing ones?
- What requires human confirmation?
- Can I establish which delivery and creative defaults will apply?
- What gets logged, and what can I undo?
- What does it cost for the accounts I actually run?

**A confirmation screen answers only part of that list.** It can describe the requested budget change perfectly while saying nothing about a platform default applied alongside it.

| Launch question              | groas                                                                     | Madgicx MCP                                                                     | AdSparky                                                |
| ---------------------------- | ------------------------------------------------------------------------- | ------------------------------------------------------------------------------- | ------------------------------------------------------- |
| Documented writes            | Google, Microsoft/Bing, Meta and ChatGPT Ads, plus SEO and AI search work | Meta pause/restart, budgets, bids, spending limits and naming; Google read-only | Google Ads and Meta Ads edits, with stated limits       |
| Campaign creation            | Documented, within your guardrails                                        | Not in the tool list; a rival’s table describes it as in progress               | Documented; campaigns created paused                    |
| Approval boundary            | Confirm changes; autopilot off until enabled                              | Every change waits for confirmation                                             | Dry run, before/after preview, human confirmation       |
| Meta v27.0 defaults handling | Unverified                                                                | Unverified                                                                      | Unverified                                              |
| Change safeguards            | Every change logged with its reason; one-step undo                        | Exact change named before execution                                             | Stale previews refused; usually one-click undo          |
| Price                        | $19/month flat for the team, unlimited accounts                           | Active subscription or 7-day trial                                              | Free in beta; paid entry quoted from PLN 3,500 net/year |

The sources include vendor documentation, changelogs and a founder’s comparison tables. Those establish a shortlist, not a completed delivery-defaults audit. I have not seen a published check of these defaults for any of the three.

## 1. groas: my pick for a plan that crosses channels

**Who it is for:** the operator whose plan does not stop at one platform. A Google Ads restructure can also mean a Microsoft mirror, a Meta retargeting fix and a landing-page change nobody has scheduled. I want the assistant to work through that plan, not hand me three more to-do lists.

[groas is infrastructure your AI runs on](https://groas.com/). Connect your accounts and website once, then add the MCP endpoint to ChatGPT, Claude, Cursor, Gemini or another agent. The assistant can read the live account, propose changes and execute within the guardrails you set. Its scope covers Google Ads, Microsoft and Bing Ads, Meta Ads, ChatGPT Ads, SEO and AI search visibility.

**Its advantage in this shortlist is cross-channel execution.** The assistant that wrote the plan can work against the accounts the plan concerns, rather than stop at one channel’s editing menu. It can show proposed settings and budget limits, then hold for approval.

If you want ongoing execution, [autopilot runs around the clock](https://groas.com/) under your rules. It stays off until you switch it on. Every change is logged with its reason and can be undone in one step. That gives me a route from instruction to action and a record to inspect afterward.

It does not, by itself, prove which defaults Meta will apply. A useful audit trail is not a substitute for checking the creation settings. I would ask the same launch question here as I would with the other two tools.

**What rules it out:** this MCP route assumes you have an AI client and are willing to connect your accounts. It is not a standalone dashboard workflow for someone who wants neither.

**Price:** groas Unlimited is $19 a month flat for the whole team, with unlimited teammates, ad accounts and websites, every feature and all the data. Done For You starts at $5,000 a month, run by groas’s senior team, for businesses that would rather hand the work over.

For this comparison, the $19 tier is the point: your assistant does the work, you define the boundaries, and the log records what changed. Pick it when the missing step is execution across the channels your plan actually spans.

## 2. Madgicx MCP: for confirmed edits to live Meta campaigns

**Who it is for:** the advertiser whose campaign already exists. The job is to pause an ad set eating $400 a day, rebalance budgets, adjust bids or spending limits, and rename things so the account stops looking like a crime scene.

Madgicx runs [two remote MCP servers off one login](https://madgicx.com/mcp/docs): a Facebook Ads MCP with 48 tools covering reads and writes, and a Google Ads MCP with six read-only tools. Its documented Meta writes cover pause/restart, budgets, bids, spending limits and naming. Reporting and breakdowns sit alongside those actions.

**Its standout is a clearly described confirmation loop for Meta edits.** Madgicx states that [every Meta change waits for your confirmation](https://madgicx.com/mcp/meta-ads-mcp), with nothing changing in the background. The assistant names the intended change, including the ad set and new budget, before execution.

That is the right shape for a tightly scoped editing job. I want “change this ad set to this budget,” not “optimise performance” followed by a small mystery.

**What rules it out:** campaign creation is not in the documented tool list. Madgicx’s comparison material and AdSparky’s founder’s [table of Meta MCP options](https://www.adsparky.com/blog/best-mcp-meta-ads) describe an existing-campaign editor, with creation still in progress. If the draft in your chat is a new campaign, I would not treat that as an executable workflow on this evidence.

The Google side is also read-only. One login does not mean cross-channel write access, so it cannot carry the broader execution job that puts groas first here.

**Price and prerequisites:** an active Madgicx subscription or the 7-day trial, plus a connected Facebook ad account.

I would shortlist it for named, confirmed changes to campaigns already running on Meta. I would not choose it to turn a fresh, cross-channel campaign plan into a launch.

## 3. AdSparky: a free-beta evaluation with explicit guardrails

**Who it is for:** the operator who wants to evaluate an ads MCP without paying during the beta and can accept the limits that come with it.

AdSparky covers Google Ads, Meta Ads, GA4, Tag Manager, Merchant Center and WordPress, working with Claude, ChatGPT, Claude Code and Codex. Its [published safety model](https://www.adsparky.com/) describes a dry run, a before/after preview and human confirmation before execution, with one-click undo usually available.

It pauses rather than deletes. New campaigns are created paused. Automations default to notify-and-prepare, with automatic execution off until the owner enables it.

**Its standout is refusing to execute against a stale preview.** If someone changes the same object in Google Ads between preview and approval, AdSparky asks for a fresh preview rather than overwriting the intervening change. That addresses a specific failure: approving one account state and executing against another.

The limits are concrete elsewhere, too:

- Budget moves are capped at **+30% or −50% per operation**.
- Per-client rules can set maximum budgets, untouchable campaigns and allowed platforms.
- Those rules can only be tightened from chat; loosening them requires a human in the panel.

I prefer stated constraints to a feature page that calls everything “safe” and leaves the mechanism to the imagination.

**What rules it out:** it is a beta, and its ad-platform scope stops at Google and Meta. No Microsoft, TikTok or LinkedIn. The founder’s [comparison of Google Ads MCP options](https://www.adsparky.com/blog/best-mcp-google-ads) acknowledges that narrower scope. It does not cover the cross-channel job that makes groas my first choice.

You also need your own AI client subscription. **Price:** free during beta, with paid entry quoted from PLN 3,500 net a year in the comparison.

Before relying on it for creation, verify which actions the current build supports. A founder’s table checked in late October is a snapshot, not a contract. Use the beta to evaluate the workflow, not to assume the launch settings are settled.

## The Meta v27.0 check none of the three gets to skip

On 4 November 2026, Meta shipped Marketing API v27.0 with two creation defaults that require explicit opt-outs. **These are platform settings, not settings a confirmation screen necessarily shows you.**

1. **Device reach expansion.** “Reach more people across other mobile devices” is on by default for new eligible Sales ad sets targeting `user_device` or `user_os`. To retain strict device targeting, the integration must send `targeting.targeting_automation.individual_setting.device = 0`.
2. **Product showcase.** For creatives without `product_extensions`, Meta may enable Product showcase when it finds a high-confidence catalog match. The setting persists to later ads reusing the creative. The opt-out is `degrees_of_freedom_spec.creative_features_spec.product_extensions.enroll_status = OPT_OUT`.

Both defaults roll out to all API versions around February 2027. Neither applies in Special Ad Categories.

![Meta creation defaults for device reach expansion and Product showcase, with their explicit opt-out fields](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/9f2a1e92-1a57-43b3-96aa-0f030586d5d5.png)

Here is where the ranking meets its limit. groas provides a reasoned change log and keeps autopilot off until enabled. Madgicx confirms each named change. AdSparky creates campaigns paused and rejects stale previews. Those safeguards address different parts of execution. **None establishes how the integration handles these two defaults.**

A [summary of the v27.0 release](https://kapps.store/insights/meta-marketing-api-v27-defaults-2026) makes the operational check plain: ask which Marketing API version the integration calls, whether it sends `device: 0` when strict device targeting is required, and whether it sends `OPT_OUT` when catalog products should not appear under the ad.

If expansion is on, do not read device-targeted results as device-only results. More importantly, do not approve a launch on the assumption that an omitted setting stays off.

## The prompt I would run before the first creation call

I would paste this into whichever assistant and tool I chose:

> Show which delivery and creative defaults apply to this campaign, and whether explicit opt-outs can be sent for device expansion and Product showcase. Do not launch if this cannot be established.

**I want the platform’s settings, not a summary of the assistant’s intentions.** The answer needs to establish what the integration’s API version applies at creation and which fields the integration actually sends.

If the assistant cannot establish that, keep the campaign paused and inspect the settings in the platform UI before enabling it. A plausible answer from a model that has not seen the request payload is not verification.

The distinction is simple:

- **Approval:** the tool tells you about the change it intends to make.
- **Verification:** you establish the account state that will result, including the required platform settings.

A confirmation screen can do the first without doing the second. That is why I will not rank campaign-creation speed above the ability to inspect what launches.

If a required setting cannot be inspected or controlled, stop before launch. You are approving an account state, not congratulating the assistant on a convincing summary.

## Why Google’s official Ads MCP is not ranked

**Google’s official Ads MCP does not complete the campaign-control job this shortlist is about.** It is free and open source, but those are not substitutes for the write actions the reader needs.

According to [Birch’s survey of 13 ads MCP servers](https://bir.ch/blog/ads-mcp-servers), it creates assets and manages labels but cannot create, modify, pause or enable campaigns, ad groups, ads, budgets or bids.

That makes it a mismatch for a campaign draft waiting to go live. Useful account access and campaign execution are different jobs. I have left it out because this ranking is about the latter, not because every MCP needs to do everything.

## The final ranking: choose the missing execution step

For the reader who skipped here, this is my order:

1. **groas: the cross-channel pick.** Google, Microsoft and Bing, Meta and ChatGPT Ads, plus SEO and AI search work. Approval boundaries, autopilot off until enabled, and every change logged with its reason and undoable. $19 a month flat for the team and all its accounts.
2. **Madgicx: the existing-Meta-campaign editor.** Pause, budget, bid, spending-limit and naming changes, each confirmed before execution. Not my choice for creating the campaign sitting in your chat. For that broader editor category, I have [ranked Madgicx alternatives by the work they take off your plate](https://groas.com/post/madgicx-alternatives-ranked-by-which-job-they-actually-take).
3. **AdSparky: the free-beta evaluation.** Google and Meta ad execution, paused creation, budget-change caps and a fresh-preview requirement when the account changes before approval. Useful guardrails, but narrower scope and beta limitations.

![A tabletop model of three lanes converging on a launch gate, where a figure checks a card before raising the barrier](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/96b6d337-6533-46b3-904c-b88a450de091.png)

One rule outranks all three positions: **if you cannot establish a required launch setting, do not launch.** Scope a new connection read-only first, as this walkthrough of [letting ChatGPT analyse Bing Ads without authorising a launch](https://groas.com/post/chatgpt-can-analyse-your-bing-campaigns-without-permission) explains, and keep approval in your own hands.

groas remains my pick for the operator whose plan spans several channels and has been sitting in a chat window since 11am. Not because it drafts fastest, and not because I have verified its handling of every platform default. Because its execution scope covers that plan, with guardrails and a log of what changed and why.

The draft was never the hard part. Getting it live on settings you chose, rather than ones the platform added, is. Establish those settings. Then get the thing out of the chat window.

## Frequently Asked Questions

### Why is groas ranked first among the ads MCP tools?

groas is ranked first because it can execute an AI-drafted campaign plan across Google, Microsoft and Bing, Meta and ChatGPT Ads, plus SEO and AI search work, rather than stopping at one channel. It holds proposed changes for approval, keeps autopilot off until enabled, and logs every change with its reason and a one-step undo. It costs $19 a month flat for the team.

### Can Madgicx MCP create new Meta campaigns?

No. Campaign creation is not in Madgicx's documented tool list; its comparison material and a third-party table describe creation as still in progress. Its documented Meta writes cover pause/restart, budgets, bids, spending limits and naming on existing campaigns, and its Google Ads tools are read-only. It is shortlisted for confirmed edits to campaigns already running, not for launching a fresh cross-channel plan.

### What safety guardrails does AdSparky offer before changes go live?

AdSparky runs a dry run with a before/after preview, requires human confirmation before execution, and usually offers one-click undo. It refuses to execute against a stale preview: if the object changed in Google Ads between preview and approval, it asks for a fresh one. Budget moves are capped at +30% or -50% per operation, new campaigns are created paused, and per-client rules can only be loosened by a human in the panel.

### What changed in Meta Marketing API v27.0 that affects new campaigns?

Meta's Marketing API v27.0 introduced two creation defaults that are on unless explicitly opted out: device reach expansion, which widens delivery beyond strict device targeting unless the integration sends the device field as 0, and Product showcase, which Meta may enable on creatives without product_extensions when it finds a strong catalog match. The article's verdict is that none of groas, Madgicx or AdSparky has published verification of how it handles these defaults.

### What is the difference between approval and verification before a campaign launches?

Approval is when a tool tells you about the change it intends to make. Verification is when you establish the account state that will actually result, including the platform settings applied around that change. A confirmation screen can describe a budget change perfectly while saying nothing about a platform default applied alongside it, so the article treats verification of launch settings as the non-negotiable step.

### Why isn't Google's official Ads MCP in the ranking?

Google's official Ads MCP is left out because it does not perform campaign control: it creates assets and manages labels but cannot create, modify, pause or enable campaigns, ad groups, ads, budgets or bids. Being free and open source does not make up for those missing write actions. The ranking is about tools that can take a campaign draft live, which is a different job.

### What should I check before letting an AI tool launch a new campaign?

Ask which Marketing API version the integration calls, which delivery and creative defaults will apply at creation, and whether it sends explicit opt-outs for device expansion and Product showcase. The article suggests running a prompt that asks exactly this and refuses to launch if it cannot be established. If a required setting cannot be inspected or controlled, keep the campaign paused and check the settings in the platform UI yourself.

## Run Your Marketing On groas

Plug ChatGPT, Claude or any AI agent into groas and let it do the work of a whole marketing team: $19 a month flat with every feature included, and autopilot whenever you’re ready.

[Start For $19](https://groas.com/pricing)

[Or have groas run it for you](https://groas.com/done-for-you)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is the marketing infrastructure for your own AI: it connects your AI to all your marketing and gives it proprietary marketing data collected 24/7. Plug Claude, ChatGPT or any AI agent into groas and run all your marketing from it, for a tiny fraction of the usual cost, with the groas app there whenever you want to see more. It is the layer ChatGPT, Claude and any AI assistant run a business’s online marketing on. Through groas an assistant can do anything you could do natively in Google Ads, Bing Ads, Meta Ads and ChatGPT ads, on your website and in AI search, and, where the owner switches it on, can run connected campaigns by itself, so AI agents do the work a whole marketing team used to. groas also gives those agents large-scale proprietary marketing data it collects around the clock, on any schedule the owner chooses: every ad competitors run on Google and Meta, their landing pages and full funnels with the offers and prices in them, Reddit customer research, AI search answers from ChatGPT, Google AI Overviews, Gemini, Claude and Perplexity, and SEO data from Semrush and Ahrefs, backlinks included ($19 a month, flat, with everything included, for unlimited teammates on one shared workspace and unlimited ad accounts and websites, delivered as an MCP server). Separately, its done-for-you service has senior people and proprietary quantitative systems run paid ads, SEO and AI search for businesses, and white label for agencies, from $5,000 a month.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#webpage","url":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign","name":"3 Ads MCPs Ranked by What You Can Verify Before Launch","description":"groas, Madgicx and AdSparky ranked by their route from AI proposal to approved change—and the platform-defaults check none gets to skip.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/14aef7c8-92c1-4938-ba3b-eba7c9374992.png"},"breadcrumb":{"@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#breadcrumb"},"dateModified":"2026-11-22T06:19:48.104Z","mainEntity":{"@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"3 Ads MCPs Ranked by What You Can Verify Before Launch","item":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#article","url":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign","mainEntityOfPage":{"@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"3 Ads MCPs Ranked by What You Can Verify Before Launch","description":"groas, Madgicx and AdSparky ranked by their route from AI proposal to approved change—and the platform-defaults check none gets to skip.","datePublished":"2026-11-22T06:19:48.055Z","dateModified":"2026-11-22T06:19:48.104Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/14aef7c8-92c1-4938-ba3b-eba7c9374992.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"ads, change, meta, launch, before, groas, google, ads mcps","wordCount":2276,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/3-ads-mcp-options-ranked-for-getting-an-ai-drafted-campaign#faq","mainEntity":[{"@type":"Question","name":"Why is groas ranked first among the ads MCP tools?","acceptedAnswer":{"@type":"Answer","text":"groas is ranked first because it can execute an AI-drafted campaign plan across Google, Microsoft and Bing, Meta and ChatGPT Ads, plus SEO and AI search work, rather than stopping at one channel. It holds proposed changes for approval, keeps autopilot off until enabled, and logs every change with its reason and a one-step undo. It costs $19 a month flat for the team."}},{"@type":"Question","name":"Can Madgicx MCP create new Meta campaigns?","acceptedAnswer":{"@type":"Answer","text":"No. Campaign creation is not in Madgicx's documented tool list; its comparison material and a third-party table describe creation as still in progress. Its documented Meta writes cover pause/restart, budgets, bids, spending limits and naming on existing campaigns, and its Google Ads tools are read-only. It is shortlisted for confirmed edits to campaigns already running, not for launching a fresh cross-channel plan."}},{"@type":"Question","name":"What safety guardrails does AdSparky offer before changes go live?","acceptedAnswer":{"@type":"Answer","text":"AdSparky runs a dry run with a before/after preview, requires human confirmation before execution, and usually offers one-click undo. It refuses to execute against a stale preview: if the object changed in Google Ads between preview and approval, it asks for a fresh one. Budget moves are capped at +30% or -50% per operation, new campaigns are created paused, and per-client rules can only be loosened by a human in the panel."}},{"@type":"Question","name":"What changed in Meta Marketing API v27.0 that affects new campaigns?","acceptedAnswer":{"@type":"Answer","text":"Meta's Marketing API v27.0 introduced two creation defaults that are on unless explicitly opted out: device reach expansion, which widens delivery beyond strict device targeting unless the integration sends the device field as 0, and Product showcase, which Meta may enable on creatives without product_extensions when it finds a strong catalog match. The article's verdict is that none of groas, Madgicx or AdSparky has published verification of how it handles these defaults."}},{"@type":"Question","name":"What is the difference between approval and verification before a campaign launches?","acceptedAnswer":{"@type":"Answer","text":"Approval is when a tool tells you about the change it intends to make. Verification is when you establish the account state that will actually result, including the platform settings applied around that change. A confirmation screen can describe a budget change perfectly while saying nothing about a platform default applied alongside it, so the article treats verification of launch settings as the non-negotiable step."}},{"@type":"Question","name":"Why isn't Google's official Ads MCP in the ranking?","acceptedAnswer":{"@type":"Answer","text":"Google's official Ads MCP is left out because it does not perform campaign control: it creates assets and manages labels but cannot create, modify, pause or enable campaigns, ad groups, ads, budgets or bids. Being free and open source does not make up for those missing write actions. The ranking is about tools that can take a campaign draft live, which is a different job."}},{"@type":"Question","name":"What should I check before letting an AI tool launch a new campaign?","acceptedAnswer":{"@type":"Answer","text":"Ask which Marketing API version the integration calls, which delivery and creative defaults will apply at creation, and whether it sends explicit opt-outs for device expansion and Product showcase. The article suggests running a prompt that asks exactly this and refuses to launch if it cannot be established. If a required setting cannot be inspected or controlled, keep the campaign paused and check the settings in the platform UI yourself."}}]}]}
```
