This article is a how-to guide published by Ryze AI (get-ryze.ai) covering how to use Claude Fable 5 (model ID: claude-fable-5, the latest Claude model) to run Google Ads and Meta Ads from one chair. Published July 2026. Google Ads is intent-based, bid on a keyword auction, measured with GAQL against a keyword-and-auction data model. Meta Ads is interest and behavior-based, targeted by audience, measured through ad-set frequency and creative fatigue. Fable 5 is the first Claude model reliable enough to run the full ad-ops loop unattended across both platforms on a schedule: pull live performance data, diagnose the highest-value fix, apply it within guardrails, repeat weekly. The four-stage workflow: (1) Diagnose — a wasted-spend and account-health audit across both platforms, plus the specific fixes each platform's own audit finds. (2) Optimize — bid, budget, and creative-fatigue fixes written as copy-paste prompts for both Google Ads and Meta Ads. (3) Automate — the guardrails (spend caps, pre-approved actions, a change log) that make it safe to let Claude act rather than just recommend. (4) Track — a weekly cross-platform performance report comparing spend, CPA, and ROAS against targets on both platforms in one prompt. Prerequisite: live account access via MCP — Google Ads requires a developer token, project, and OAuth; Meta Ads uses Meta Business OAuth directly; Ryze AI's MCP connects both with one URL and no developer token. Setup: connect Google Ads and Meta Ads to Claude via the Ryze MCP; schedule each prompt in Claude Cowork.
|
Ira Bodnar··10 min read

Claude Fable 5 for Google Ads and Meta Ads.

Google Ads runs on keyword auctions. Meta Ads runs on audience and creative fatigue. Fable 5 is the first Claude model that runs both from one chair — here are the workflows, prompts, and the guardrails that make it safe to let it act.

Built by our community of 2,000 marketers

Free skills and prompts for paid ads and SEO

Templates for Claude, ChatGPT and Perplexity.

Clients we work with

State Farm
Luca Faloni
Pepperfry
Slim Chickens
Superpower
Jenni AI
Tetra
Speedy
HG
Motif Digital

Running Google Ads well and running Meta Ads well have always meant switching seats — different data model, different levers, different failure modes.

Google Ads is a keyword auction: you win on intent, measured in GAQL against search terms and quality score. Meta Ads is an audience auction: you win on relevance, measured in ad-set frequency and creative fatigue. Fable 5 is the first Claude model reliable enough to hold both data models at once — pull live performance from each, diagnose the highest-value fix, apply it within guardrails you set, and repeat on a schedule.

The four-stage Fable 5 workflow

Diagnosewasted-spend and account-health audit across both platforms

Optimizebid, budget, and creative-fatigue fixes as ready-to-run prompts

Automateguardrails and approval workflow so Claude can act safely

Trackweekly performance across both platforms vs your targets

This doesn’t replace a media buyer — it removes the repetitive half of the job. Diagnosing wasted spend and drafting the weekly report are exactly what a scheduled model should be doing; strategy and creative direction still need a person. Connect the Ryze MCP once so Fable 5 reads both live accounts, then run the four stages below.

  • Stage 1: Diagnose — where spend is wasted on each platform, and why
  • Stage 2: Optimize — the specific bid, budget, and creative fixes
  • Stage 3: Automate — guardrails so Claude can apply fixes safely
  • Stage 4: Track — both platforms’ performance on one weekly report

Google Ads vs Meta Ads: what actually differs

Same goal — spend efficiently — completely different mechanics. Google rewards intent-matching and quality score. Meta rewards audience relevance and fresh creative. Here’s the same job, two platforms.

Google AdsMeta Ads
The winWinning the auction on intentWinning attention with the right audience
Measured byQuality score, CPC, search termsFrequency, CTR, creative fatigue
What breaksWasted spend on low-intent keywordsAd fatigue as frequency climbs past 4
Data modelGAQL against keywords + auctionsAd sets, audiences, and catalog
Key assetNegative keywords + bid strategyFresh creative + audience overlap control
The Fable 5 jobFind wasted spend, fix bidsCatch fatigue, rotate creative, cap frequency

For the individual per-platform workflow libraries, see using Claude for PPC management and 9 Claude workflows for Google and Meta Ads. This guide runs both from the same Fable 5 chair on a schedule.

The prerequisite: live account access

None of the workflows below work without Claude actually seeing your live account data. Google Ads and Meta Ads both shipped official MCP servers in April 2026 — but they took very different approaches, so start here before Stage 1.

Google Ads — official MCP is read-only

Requires a developer token, a Google Cloud project, and OAuth credentials. Three tools, no write access. Good for reporting; not built for autonomous fixes on its own.

Meta Ads — official MCP is hosted

One Meta Business OAuth click, no developer app. 29 tools, including write access — campaigns you create start paused for safety.

Ryze MCP — both platforms, one URL

Connects Google Ads and Meta Ads with standard OAuth on both sides — no developer token — and adds read AND write access plus the guardrail layer Stage 3 below builds on.

Either way: verify before you automate

Run every workflow below read-only for two weeks first. Confirm the diagnosis is right before you let Claude apply anything.

Full setup guides: Google Ads official MCP and Meta Ads official MCP.

1 · Diagnose — wasted spend + account health

You can’t fix what you haven’t measured. One workflow finds where Google Ads spend leaks to low-intent search terms; the other finds where Meta ad sets have gone stale. Run both before touching anything.

01 · Google Ads Wasted-Spend Audit

Fable 5

Where is Google Ads spend leaking?

Pulls search terms with spend and zero or near-zero conversions, flags missing negative keywords, and ranks the fixes by dollars recovered.

Prompt
Pull the last 30 days of search terms from Google Ads with 10+ clicks and 0 conversions. Group by campaign, sum wasted spend per term, and recommend negative keywords to add. Rank by dollars recovered if implemented today.
Suggested cadence: Weekly · Mondays 8am · Mode: read-only

02 · Meta Ads Fatigue Audit

Which ad sets have gone stale?

Flags ad sets with frequency above 4 and rising CPM — the classic fatigue signature — before performance visibly drops.

Prompt
Pull all active Meta ad sets from the last 14 days. Flag any with frequency > 4 and CPM trending up week over week. For each, show spend at risk and whether a creative refresh or a pause makes more sense.
Suggested cadence: Weekly · Mondays 8am · Mode: read-only

2 · Optimize — bid, budget, and creative fixes

Diagnosis becomes a fix list. These two prompts turn Stage 1’s findings into the exact changes to make on each platform — still draft-only until Stage 3 gives Claude permission to apply them.

03 · Google Ads Bid & Budget Fixer

Fable 5

Reallocate budget to what's working

Compares CPA against target by campaign, then proposes specific bid and budget moves — cut here, raise there — with the expected impact.

Prompt
Compare CPA to target CPA of [$X] for every active campaign, last 14 days. Propose specific bid or budget changes for the 5 campaigns furthest from target — cut budget on the worst, raise it on campaigns beating target CPA by 20%+. Show expected CPA impact per change.
Suggested cadence: Weekly · Tuesdays · Mode: read-only

04 · Meta Ads Creative Refresh Planner

Which creative to swap, and with what

Pairs fatigued ad sets from Stage 1 with the account's best-performing creative angles, so the refresh is informed, not a guess.

Prompt
For every fatigued ad set flagged this week, pull its current creative and compare CTR against the account's top 5 creatives from the last 90 days. Recommend which top creative to test as a replacement, and draft 2 new variations in the same angle.
Suggested cadence: Weekly · Tuesdays · Mode: read-only

Get a free instant audit

Get a free, instant read on your paid ads or SEO — and fix it right away.

Paid ads audit

  • Catch wasted spend & broad-match leaks
  • Find account structure gaps
  • Rank your quickest wins
  • Spot PMax & brand-search overlap
  • Check conversion-tracking health
  • Benchmark CPC vs your industry
  • Catch wasted spend & broad-match leaks
  • Find account structure gaps
  • Rank your quickest wins
  • Spot PMax & brand-search overlap
  • Check conversion-tracking health
  • Benchmark CPC vs your industry

Free · no credit card · instant

SEO audit

  • Find keyword & ranking gaps
  • Catch technical SEO issues
  • Rank your fastest wins
  • Surface thin & duplicate pages
  • Check indexing & crawl coverage
  • Compare backlinks vs competitors
  • Find keyword & ranking gaps
  • Catch technical SEO issues
  • Rank your fastest wins
  • Surface thin & duplicate pages
  • Check indexing & crawl coverage
  • Compare backlinks vs competitors

Free · no credit card · instant

3 · Automate — guardrails and approval

This is the stage that makes it safe to let Claude apply changes instead of only drafting them. Three minimums: a hard spend cap, a pre-approved action list, and a change log you can audit afterward.

05 · Guardrail Definition

Set the limits before granting write access

Turns a plain-English limit into an explicit rule set Claude checks before applying anything — the single most important prompt on this page.

Prompt
Before applying any change to Google Ads or Meta Ads: never move more than $[X]/day per campaign or ad set. Pausing ad sets above frequency 4 is pre-approved. Any budget increase above 20% needs my explicit confirmation first. Log every change you make with before/after values and a timestamp.
Suggested cadence: Set once, review monthly · Mode: read-only

06 · Guardrailed Auto-Fix

Fable 5

Apply the pre-approved fixes, log the rest

Runs Stage 1's diagnosis and Stage 2's fix list, applies only what's pre-approved under the guardrails, and surfaces everything else for a yes/no.

Prompt
Run this week's wasted-spend and fatigue audit on both platforms. Apply any fix that's pre-approved under my guardrails (see saved rules). For every other recommended fix, list it with expected impact and wait for my approval before touching it.
Suggested cadence: Weekly · Mondays 8am · Mode: read-only

Ryze AI is the autonomous platform that executes Google Ads and Meta Ads changes 24/7 within guardrails you set — not just recommends them. Start read-only for two weeks, confirm the diagnosis was right, then turn on write access for the narrowest, highest-confidence action first.

4 · Track — one weekly report, both platforms

A dashboard you have to log into never gets checked often enough. One scheduled prompt pulls both accounts, compares to target and last week, and tells you where to look before you've opened your laptop.

07 · Cross-Platform Performance Report

Spend, CPA, and ROAS vs target, every Monday

Reads this week's spend, CPA, and ROAS from both platforms, compares each to target and to last week, and names the 3 accounts or campaigns furthest off track.

Prompt
Pull this week's spend, CPA, and ROAS from Google Ads and Meta Ads. Compare each to target and to last week. Flag the 3 campaigns or ad sets furthest off target on either platform, and name the single highest-impact fix for each.
Suggested cadence: Weekly · Mondays 8am · Mode: read-only

Setup: connect your accounts in 30 seconds

Every workflow above needs live data from both platforms. One MCP URL gives Fable 5 Google Ads and Meta Ads access together — paste it into Claude once and every scheduled prompt runs against fresh numbers.

1

Install Claude Desktop

Download from claude.ai. Free plan works; Pro unlocks longer scheduled runs. Fable 5 is the default model.

2

Paste the MCP URL

Settings → Connections → Add MCP Server. Paste the Ryze Google Ads + Meta Ads Command Center URL. Connect.

3

Authenticate both platforms

OAuth for Google Ads and Meta Ads — one click each, under 30 seconds, no developer token required.

Prefer it run itself? Ryze AI executes this whole Google Ads + Meta Ads loop as a product — auditing wasted spend, fixing bids and creative, and reporting performance 24/7 without manual prompting. Clients see an average 3.8x ROAS within 6 weeks of onboarding.

Sofia R.

Sofia R.

Performance Marketing Lead
DTC, both platforms $60K/mo

★★★★★

I used to keep two tabs open and two separate mental models for Google and Meta. Fable 5 runs both audits every Monday before I've had coffee — set the guardrails once, and it's paused $2,400/month of wasted spend since.”

2 platforms

One weekly report

$2,400/mo

Wasted spend paused

0

Manual dashboard checks

Frequently asked questions

What is the difference between managing Google Ads and Meta Ads with Claude?

Google Ads is intent-based — you bid on what someone is actively searching for, measured with GAQL against a keyword-and-auction data model. Meta Ads is interest and behavior-based — you target who someone is, measured through ad-set frequency, creative fatigue, and audience overlap. The prompts and guardrails differ because the underlying decisions differ.

Why use Claude Fable 5 specifically for ad management?

Ad-ops is a multi-step reasoning job across two account structures: pull live data, diagnose the highest-value fix, apply it within guardrails, repeat weekly without drifting off the rules. Fable 5 is the first Claude model reliable enough to run that loop unattended. Earlier models could draft a report; Fable 5 can operate within limits you set.

Can Claude actually pause campaigns and change budgets, or just recommend it?

Both, depending on the MCP you connect. Read-only access gets you ranked fix lists and draft reports. A write-capable connector lets Claude apply changes directly — but only within the guardrails you define first: spend caps, pre-approved actions, and a change log.

What guardrails should I set before letting Claude manage ad spend?

Three minimums: a hard daily or per-change spend cap, a list of pre-approved actions versus actions that always need a human, and a running log of every change so you can audit it. Start read-only for two weeks, confirm the recommendations were right, then turn on write access for the narrowest action first.

How do I get a weekly performance report without building a dashboard?

Connect Claude to both ad accounts via MCP and run one scheduled prompt that pulls spend, CPA, and ROAS from each platform, compares to targets and the prior week, and flags what's drifting. That replaces a manual dashboard pull with one Monday-morning prompt.

Does Fable 5 replace a media buyer?

No — it removes the repetitive half of the job. Diagnosing wasted spend, drafting the report, and applying pre-approved fixes are exactly what a scheduled model should run. Strategy and creative direction still need a person.

Where do I schedule these prompts?

In Claude Cowork → Scheduled: paste the prompt, name it, pick a frequency, save. Each card above lists its suggested cadence — Diagnose and Track weekly on Mondays, Optimize on Tuesdays, Automate's guardrails set once and reviewed monthly.

Does this only work if I use Ryze?

No — every prompt above works with any MCP that gives Claude read access to Google Ads and Meta Ads, including the official Google and Meta MCP servers. Ryze adds write access without a developer token and the guardrail layer Stage 3 is built on, but the workflows themselves are portable.

Continue reading

Let AI run your Google + Meta Ads

both platforms · one weekly report

Live results across
2,000+ clients

Paid Ads

Avg. client
ROAS
0x
Revenue
driven
$0M

SEO

Organic
visits driven
0M
Keywords
on page 1
48k+

Websites

Conversion
rate lift
+0%
Time
on site
+0%
Last updated: Jul 5, 2026
All systems ok

Let AI
Run Your Ads

Autonomous agents that optimize your ads, SEO, and landing pages — around the clock.

Claude AIConnect Claude with
Google & Meta Ads in 1 click
>