Ad Naming Convention: How to Make Your Creatives Analyzable

You test twenty new ads every month. Some with testimonials, some with a price, some with a provocative question. After three months you want to know: which kind of ad brings customers? And then you're looking at names like "Video new final 2", "Copy of Carousel Oct" and "Test Max".

This guide shows how to name ads so you can later analyze them by hook, offer and format - in the ad account, in the dashboard and with AI assistants like Claude. It complements the UTM naming convention, which covers source, medium and campaign.

Why ad names decide your analysis

An ad account knows a lot about an ad: spend, impressions, clicks, format. What it doesn't know is the idea behind the ad. Whether a video opens with a price, a testimonial or a problem isn't recorded anywhere - unless you put it in the name.

The name is also the only thing that travels with an ad through every tool:

• Ad account: Filter and search by name parts in Ads Manager or Google Ads.
• UTM parameters: With utm_content={{ad.name}}, the name lands in every landing page URL.
• Tracking and CRM: Tools like LeadMetrics sync the names of campaigns, ad sets and ads and attribute leads, deals and revenue to them.
• AI analysis: An AI agent doesn't see images or videos, only names and numbers. If it can read the hook type from the name, it can group by hook.

Without structure in the name, all you can do is compare individual ads. With structure, you compare ideas: "Testimonial ads cost 30% less per deal than price ads." That's an insight you can build the next twenty ads on.

What belongs in the name

Only include what you'll actually compare by and what the platform doesn't provide itself. These parts have proven useful:

Audience, placement and region belong one level up in the ad set name, and the campaign objective and product in the campaign name. That way, each piece of information appears exactly once.

Fixed values instead of free text

The most important part of the convention is the value list. "proof", "testimonial", "review" and "social-proof" are four different hooks to any analysis. Define a short list of allowed values per part and only extend it deliberately.

For hook type, a list of five to eight values works well:

• price: opens with price, savings or an offer
• proof: testimonial, result, number, before and after
• problem: opens with a pain point of the audience
• curiosity: question, surprising statement, open loop
• authority: expert, award, media mention
• urgency: deadline, limited spots, season

The scheme

Combine the parts in a fixed order, separated by underscores, all lowercase:

Campaign:  leads_mortgage_us
Ad set:    homeowners_35-55_advantage
Ad:        2610_proof_consult_video_v01

A few rules that save a lot of work later:

• Always the same order: Swap format and hook once and analysis by name part breaks.
• No free text at the end: Notes like "new" or "final" belong in a separate document, not in the name.
• Number variants, don't describe them: What makes v02 different from v01 goes in the creative brief.
• Rename duplicates: When duplicating, platforms append "Copy". Fix it right away.

Renaming existing ads

Meta and Google Ads keep the ad ID when you rename, so metrics and the learning phase are preserved. Two things to keep in mind:

• UTMs with the old name: If you write the ad name into utm_content via {{ad.name}}, clicks before the rename still carry the old name. So rename on a fixed date and note it.
• Only active and relevant ads: You don't need to touch old, paused ads. It's enough if all ads from a certain date on follow the convention.

Analyzing by hook type

With clean names, you can compare ads by idea instead of one by one. The key is not to look only at click-through rate and cost per lead, but at what happens in sales.

An example analysis by hook type over three months:

By cost per lead, price and curiosity ads win. By cost per acquisition, testimonial ads are clearly ahead. Without the hook type in the name, this pattern would be nearly invisible across 60 individual ads.

In LeadMetrics

In ad analysis, search for a name part like proof and compare up to four ads with up to three metrics, across projects and with first-click or last-click attribution. In the dashboard you see leads, deals, cost per acquisition and revenue down to ad level.

With Claude

Via the LeadMetrics MCP server, Claude fetches metrics per ad including the ad name and can group them by name parts:

Get all ads from the last 90 days for project "Mortgage" with
spend, leads, deals and revenue.
The ad names follow the scheme month_hook_offer_format_variant.
Group by hook and by format, sum the values and calculate
cost per lead and cost per acquisition. Which combination of hook
and format should we expand in our next ads?

How to connect Claude to LeadMetrics is covered in the guide analyze campaigns with Claude.

Classify competitors with the same list

You can use the same hook list for competitor research. If you classify ads from the Meta Ad Library by the same hook types - by hand or with an AI assistant - you'll see which openings are overcrowded in your market and which almost nobody uses. A bigger workflow with several agents is described in the guide on the AI marketing team.

Spotting creative fatigue

With the launch month in the name, you can also see how long an idea lasts. Typical signs of creative fatigue: frequency rises, click-through rate falls and cost per lead goes up while audience and budget stay the same. Compare a creative's metrics over several weeks, and ads with the same idea from different launch months. In LeadMetrics you can set daily threshold alerts on ad metrics such as frequency or click-through rate, which appear in the app.

Common mistakes

• Too many parts: Nobody maintains ten name parts consistently. Four to six are enough.
• Unmaintained value lists: New values appear spontaneously, old ones get reinterpreted. Clean up once a quarter.
• Judging by click-through rate only: A hook that sparks curiosity doesn't automatically bring customers.
• Comparing groups that are too small: A hook type with two ads and three deals isn't a pattern. Wait until enough spend has accumulated.
• A/B tests with several changes: If hook, format and offer change at once, you don't know what made the difference.

Checklist

• Parts and order defined (e.g. month, hook, offer, format, variant)
• Value list for every part in a shared doc
• Audience and placement in the ad set name
• utm_content={{ad.name}} in the URL parameters
• Rename date noted
• Analysis by hook type with cost per acquisition, not just cost per lead