YouTube Ads Agent

This guide explains what the Youtube agent does, when to use it, what you need, how to get started, guardrails, and current limitations. It turns public channel data and video analytics into a shortlist of high quality placements and creative insights you can act on.

What this agent is

One Agent that discovers channels, scores them for fit and safety, analyzes their content and audience, and turns findings into a test plan. It is built for YouTube advertising and partnerships, not for organic channel management.

Use it to replace manual research, scattered spreadsheets, and guesswork.

When to use it

• You want to expand beyond the same channels you always buy.
• You need a defendable shortlist for sponsorships or in stream placements.
• You want to understand creative patterns before producing ads.
• You want a fast way to compare channels across markets and languages

What you need before you start

• Your primary market and language targets.
• A short list of seed channels or topics, or a set of competitors to anchor the search.
• A timeframe, for example, lasts 30 days or last 3 months.

No ad account access is required.

How to get started

1. Share seed inputs, for example “fitness coaching in India, English and Hindi, lookalikes of AbcFitness.”
2. Tell the agent the timeframe to analyze.
3. Add any constraints, for example minimum subscribers, brand safety rules, or geo focus.
4. Ask for outputs in plain language, for example “give me a tiered buy list with scores and risks.”

Capabilities

Channel discovery and expansion

• Industry research and partner discovery across verticals such as fitness, finance, and tech.
• Trend monitoring by topic and hashtag to surface channels riding real demand.
• Country and region filters to align with your media plan.
• Subscriber range filters to match your budget and testing appetite.

Example prompts

• “Find 100 finance channels in India with 50k to 500k subscribers that posted in the last 30 days.”
• “Show rising fitness creators in the UK discussing zone 2, VO2 max, and strength after 40.”

Community and conversation tracking

• Hashtag watches that reveal active conversations and engaged audiences.
• Seasonal opportunity scans around shopping periods and cultural events.
• Movement and event tracking for conferences or industry moments.

Example prompts

• “Track Diwali related tech deals content for the past 60 days, return channels and top videos.”
• “List channels covering SaaS conferences and map engagement patterns.”

Channel intelligence and performance analysis

Partnership evaluation

• Audience quality checks such as subscriber growth, engagement rate, and likely authenticity.
• Content category mapping to see what actually performs on the channel.
• Posting cadence insights and timing windows that matter for delivery.
• Brand safety review against your guidelines.

Performance benchmarking

• Benchmark engagement rates against peers in the same niche and size band.
• Format and length patterns that correlate with performance.
• Audience behavior such as when viewers are most active.
• Growth trend analysis to separate rising stars from channels in decline.

Strategic channel scoring

• Score channels from 0 to 100 for predicted fit and performance.
• Risk assessment for controversy, misinformation, and unsuitable topics.
• Quality ranking by authority and engagement to drive a tiered media plan.
• Budget allocation guidance that tiers channels for test versus scale.

Example prompts

• “Score the top 10 channels for Indian personal finance, show engagement rate, growth trend, and safety notes.”
• “Compare three tech review channels by content mix, average view duration proxy, and comment quality.”

Audience expansion and lookalike discovery

• Similar channel discovery at scale using your top performers as seeds.
• Audience pattern matching for demographics and viewing behavior where observable.
• Thematic expansion into adjacent interests that share buyers.
• Geographic expansion to clone winners into new markets.

Example prompts

• “Find 200 lookalikes of our top 10 channels in the US, then clone to the UK and Australia.”
• “Expand from fitness to recovery and sleep optimization, keep only channels with high comment quality.”

Content strategy and creative insights

• Top video identification on target channels with themes, hooks, and offers.
• Format guidance on what lengths and structures tend to win.
• Topic trend analysis to inform your next creative sprint.
• Upload strategy insights such as day and time patterns.

Example prompts

• “Extract common hooks from the 100 most engaging videos across these channels.”
• “Show which video lengths correlate with higher engagement for fintech education.”

Audience sentiment and community analysis

• Brand mention monitoring across channels for you and named competitors.
• Competitor sentiment scans to learn how audiences react to their positioning.
• Community health checks, for example toxicity flags and spam density.
• Relationship opportunities where sentiment is positive and aligned.

Example prompts

• “Summarize how audiences talk about Competitor X in the last 90 days, list top praise and top complaints.”
• “Identify channels with positive sentiment toward our category and low toxicity.”

Business impact and ROI

• Research time is cut from weeks to days, sometimes hours.
• Scale from dozens of channels to hundreds without losing safety.
• Better CPM and conversion rates from smarter selection.
• Head start on emerging channels before competitors arrive.

Output formats you can request

• Tiered buy list with scores, risks, and notes.
• Side by side channel matrix with engagement, growth, category mix, and safety flags.
• Trend brief with topics, hooks, and creative angles ready for testing.
• Country by country plan that maps channels to funnel stages.
• Executive summary one pager with recommended next steps.

Guardrails and tips

• Specify markets, languages, timeframe, and any brand safety rules.
• Set minimum activity, for example at least two uploads in the last 30 days.
• Ask for scores and the reason behind each score, not just a list.
• Keep brand names and topics unambiguous, include official domains for clarity.
• Separate discovery and deep dive. First collect breadth, then analyze the top 10 to 30.

Limitations and assumptions

• Data is based on public channel and video signals. Exact revenue, watch time, and private analytics are not exposed.
• Sentiment and authenticity checks are best effort using observable patterns. Treat them as decision aids, not final truth.
• Safety assessments follow your rules. If rules are not provided, a default set is used and may be stricter than you need.
• Predictions use historical patterns. Sudden content pivots by creators will change outcomes.

Example end to end prompt

“Analyze personal finance channels in India and the UK for the last 180 days. Return three outputs: 1) a tiered buy list of 120 channels with scores, risks, and notes, 2) a hook and offer matrix from the top 100 videos, 3) a four week test plan with budgets and expected ranges for CPM and conversions.”