Every ad your competitor is running on Facebook and Instagram is sitting in the Meta Ads Library right now. Their video scripts, their offers, their landing pages, the influencers they’re paying. All of it, publicly accessible. All of it, waiting to be used.
Most founders look at this library once, click through a few ads manually, and walk away with a vague impression. “They’re doing a lot of video.” “Their creative looks polished.” That’s not competitive intelligence. That’s a gut feeling dressed up as research.
The problem isn’t access. Its scale and structure. When a competitor is running 30 to 50 active ads across influencer collaborations, catalogue formats, and direct-response video, you can’t manually watch every creative, transcribe every hook, and extract every offer. So you don’t. You run your own ads based on what you think is working for them, and you leave a lot of signal on the table.
That’s what the /meta-ads-gap-analysis skill for Claude Code is built to fix. You can find and install the skill at github.com/smacient/marketing-skills.
Table of Contents
| What It Actually Does |
| What You Learn From One Analysis |
| The Gap Analysis Tab Is Where It Gets Strategic |
| Who This Is For |
| How to Run It |
| Frequently Asked Questions |
What It Actually Does
You provide two Facebook page URLs and a country code. The skill handles everything else.
It pulls up to 50 active ads per brand from the Meta Ads Library, downloads every video and image before the CDN links expire, then runs each creative through Google Gemini AI. For every single ad, yours and your competitor’s, it identifies the emotional angle being used, the narrator type, the hook structure, the key claims made, production quality, and any pricing or offer mentions.
For every video, it also extracts the verbatim first spoken line and any on-screen text in the opening seconds. Not a summary. The exact words.
The output is a 9-tab Excel report plus a markdown summary. The whole process runs unattended. You come back to a structured, comparative breakdown of both brands’ full ad libraries. No manual watching, guessing, and gut feelings.
What You Learn From One Analysis
Here’s where it gets concrete. These are the actual types of insights one analysis surfaces, the kind of thing that changes how you allocate budget and write briefs next week.
Influencer footprint. Does your competitor run branded content through creators? How many influencer-style ads vs. direct brand ads? Which creators are appearing repeatedly? If they’re testing 8 influencers and you’re running zero, that’s a gap. If they’re running one influencer in heavy rotation, it tells you something is working for them.
Hook strategy. Are they leading with identity and aspiration (“For moms who refuse to compromise”) or fear-of-harm (“Most kids’ products contain chemicals you’ve never heard of”) or product features (“Soft, breathable, designed for India’s climate”)? When you can read the emotional angle across 40+ ads, you stop guessing what type of hook converts in your category.
Audio hooks, verbatim. This is the piece most founders don’t think to look for. The exact first sentence spoken in every video. Not a description, the words themselves. “I used to spend 3,000 rupees a month on this.” “My daughter has been wearing this brand for two years.” When you read 20 of these in sequence, you start to see the pattern your competitor has tested their way to.
Creative velocity. How many new ads did they launch in the last four weeks? If you’re publishing two new ads a month and they’re publishing ten, creative volume alone could explain the performance gap. Or you might find you’re actually outpacing them, which tells you something different.
Format breakdown. What share of their spend is going into video vs. static images vs catalogue formats? If they’re running 70% video and you’re running 70% static image, that’s a meaningful structural difference worth understanding.
Offers and pricing. What discounts are they featuring in creatives? Are they leading with free shipping, first-order offers, percentage discounts, or bundles? What products are they linking to: a homepage, a category page, a single hero SKU? This tells you what they’re optimizing for and what’s likely converting.
Landing page strategy. Where are they sending traffic? If every ad links to the same product page, they’ve likely found a winner. If they’re testing five different destinations, they’re still in discovery mode. Useful either way.
The Gap Analysis Tab Is Where It Gets Strategic
The Excel output includes a dedicated Gap Analysis tab that doesn’t just show differences. It prioritises them. Each gap is flagged as CRITICAL, HIGH, or MEDIUM, with a description of what the gap is and why it matters.
This is the part that turns data into a brief. Instead of scrolling through competitor ads and free-associating, you get a structured list of the specific areas where the other brand is doing something you’re not, and a starting point for deciding which gaps to close first.
The Recommendations tab goes further: actionable steps mapped to what your specific competitor is doing. Not generic advice about “testing more video.” Specific direction based on the actual patterns in their active library.
Who This Is For
If you’re spending money on Meta ads and you have at least one meaningful competitor also running on Meta, this analysis will surface something useful. Guaranteed.
It’s most valuable when you’re preparing a creative strategy for the next quarter, your performance has plateaued, and you’re trying to understand why, you’re entering a category with established players and want to know what you’re walking into, or you’re a consultant or agency onboarding a new D2C client and need to fast-track competitive context.
It’s not a tool for casual curiosity. The output is detailed, structured, and designed to be acted on. Come ready to do something with what you learn.
How to Run It
The whole setup takes about 10 minutes the first time. After that, running a new analysis is a single command.
Step 1: Create a free Smacient account
Go to smacient.com and sign up. Smacient is the data connector that pulls live ad data directly from the Meta Ads Library into Claude. Each analysis costs 12 credits (6 per brand). New accounts come with free credits to get started.
Step 2: Install Claude Code
Download Claude Code from claude.ai/code. It runs as a CLI, a desktop app on Mac and Windows, or directly in your browser at claude.ai/code. All three work identically for this skill.
Step 3: Connect the Smacient MCP connector
Inside Claude Code, open Settings and navigate to the Integrations or MCP section. Add the Smacient connector and sign in with your Smacient account. Once connected, Claude can pull live Meta Ads Library data on your behalf with a single tool call. No scraping, no CSV exports, no manual downloads.
Step 4: Install the skill
Clone or download the skill from github.com/smacient/marketing-skills. Copy the meta-ads-gap-analysis folder into your project’s .claude/skills/ directory:
cp -r skills/meta-ads-gap-analysis path/to/your/project/.claude/skills/meta-ads-gap-analysis
Then set up a Python virtual environment in your project root and install the dependencies:
python -m venv venv
venv\Scripts\pip install -r .claude\skills\meta-ads-gap-analysis\requirements.txt
You’ll also need a Google Gemini API key (free tier is sufficient for most analyses). Add it to your Claude Code settings under the GEMINI_API_KEY environment variable.
Step 5: Run the analysis
In Claude Code, type:
/meta-ads-gap-analysis https://www.facebook.com/yourbrand/ https://www.facebook.com/competitorbrand/ IN
Replace the URLs with the Facebook page URLs of your brand and a competitor, and swap IN for your country code (US, GB, AE, etc.).
Claude takes it from there. It calls Smacient to fetch up to 50 active ads per brand, downloads every video and image, runs Gemini analysis on each creative, extracts audio hooks, and builds the full Excel report. The whole pipeline runs unattended, typically 20 to 35 minutes, depending on how many videos each brand is running.
Step 6: Open the report
The output lands in outputs/ as a 9-tab Excel file and a markdown summary. Start with the Gap Analysis tab, then work through Recommendations, Audio Hooks, and Influencer Analysis in that order. Those four tabs give you everything you need to brief your next creative sprint.
Cost per analysis: 12 Smacient credits plus Gemini API usage billed to your Google account. Gemini costs on a typical 100-video analysis run a few cents to a couple of dollars.
Frequently Asked Questions
No. The skill pulls data from the Meta Ads Library, which is publicly accessible. You do not need any special Facebook account permissions or a Meta Business account to run the analysis.
The skill fetches up to 50 active ads per brand. This covers most active advertisers. If a competitor is running fewer than 50 ads, it will pull their full active library.
Google Gemini is the AI model that analyzes each video and image creative, identifying emotional angles, hook structures, offers, and on-screen text. The free tier is sufficient for most analyses. A typical 100-video run costs anywhere from a few cents to a couple of dollars on a paid plan.
Yes. When running the command, you specify a country code such as US, GB, or AE. This filters the Meta Ads Library to ads running in that market. You can analyze any brand running ads in any supported country, regardless of where you are based.
The skill is available at github.com/smacient/marketing-skills. Before you start, you need a free Smacient account (smacient.com), Claude Code installed, the Smacient MCP connector set up, and a Google Gemini API key added to your environment variables.
Related Blogs
How to Reverse-Engineer Any Creator’s Content Strategy
How to Analyze Amazon Reviews and Get Voice of Customer Insights with Claude
