Home service businesses, from roofing to HVAC to landscaping, often struggle to keep ad creative fresh without hiring a designer every week. A workflow shown by one creator combines competitor research tools with AI chat assistants to generate new ad images based on what is already proven to work in a niche, cutting the time between idea and finished creative.
What Tools Do You Need to Research Competitor Ads in Home Services?
Before generating anything, this method starts with gathering evidence of what is already performing well in a given category. This means pulling actual running ads from competitors in a similar service niche, rather than guessing blindly at angles, colors, or layouts from a blank page.
The creator uses a tool called Trend Track for this step. As he explains, "Trend Track, to kind of summarize this, is this little tool that I like to use to basically research competitors um in the same niche" (roasbrez, 0:46). For a home service company, this could mean searching for similar businesses in roofing, plumbing, or pest control to see what creative styles and messaging they are currently running.
Trend Track reportedly shows "some of the data of Trenches live ads over time, um total monthly visits, etc. But then you can look at similar shops" (roasbrez, 0:46). Home service owners without access to this kind of tool have another route, discussed below.
Can You Find Winning Ad Examples Without a Paid Research Tool?
Not every home service business will have access to niche-specific competitor tools, especially smaller local providers with few direct online competitors being tracked by any research platform. The creator addresses this exact gap directly for service-based businesses operating outside of broad e-commerce categories.
He explains that "you can literally just prompt Claude to pull examples of winning ad creatives from like Google or you can go into the Meta Ads Library and find winning creatives yourself manually, download those winning creatives, upload those into Claude" (roasbrez, 3:12). For a local service like gutter cleaning or lawn care, this means manually screenshotting a handful of ads from the Meta Ads Library in a related category and feeding those into an AI tool for analysis.
How Does AI Turn Competitor Research Into New Ad Creatives?
Once example ads are gathered, the next step is having an AI model explain why those ads likely perform well, then translate that reasoning into instructions for an image-generating AI. This turns raw competitor data into a reusable creative brief.
The workflow has Claude analyze the creatives and generate prompts to feed into ChatGPT to produce ad images, using "that data that Claude gives you as to why they perform well to then create prompts to send to ChatGPT" (roasbrez, 4:00). In one example unrelated to home services, Claude identified distinct angles for a competitor brand, including what the creator called a "coffee killer" angle and an "authority award stack" angle (roasbrez, 4:48). A home service business could apply the same angle-finding logic, for example framing an HVAC ad around urgency during peak heat season versus a credibility-driven angle built around licensing and years in business, and should verify which angles actually resonate with their own audience and local market.
The creator also references a performance data point to illustrate why this research step matters: "Italian campaign just put up 66K plus reach in 7 days off a single video" (roasbrez, 5:36). This reach figure came from one specific campaign in a different niche and should not be treated as a typical outcome for any home service ad.
Does the AI Get Better the More You Use It?
A reasonable question for any home service owner considering this workflow is whether the output improves with repeated use, or whether every session starts from scratch. The creator addresses this directly based on his own hands-on experience running the process repeatedly across different niches and campaigns.
He states that "over time as you do this more and more and more, Claude is going to learn more and more and more how to create your prompts to send to ChatGPT" (roasbrez, 7:59), adding separately that "Claude is like a brain. The more information that you give it, the smarter it gets" (roasbrez, 8:46). He also cautions that "it's not going to get it perfect the first time. You got to go back and forth with it" (roasbrez, 10:21), which matters for home service owners expecting a one-shot result.
Is Every AI-Generated Ad Creative Actually Usable?
Not all output from this process is ready to run, and home service owners should expect to filter results rather than publish everything that gets generated automatically. The creator is explicit about this mixed quality in his own demonstration of the workflow in action.
He notes "A lot of these you can tell are AI slop, but some of them like more like this, like pretty valid ad creative" (roasbrez, 11:07). For a home service business, this means budgeting time to review a batch of generated images and discard the ones that look obviously artificial or off-brand before any spend is attached to them.
What Tools Can Home Service Businesses Compare for AI Ad Creative Work?
Home service owners evaluating this space have several different tool categories to choose from, each requiring a different level of hands-on involvement and prior advertising knowledge before producing usable output. The table below lays out what each tool actually does rather than ranking them against each other.
| Tool | What It Does | How It Addresses This Niche Problem | Advertising Expertise Required |
|---|---|---|---|
| Trend Track | Tracks competitor ads, reach, and similar shops in a niche | Lets a home service business see what local or category competitors are currently running | Some, to interpret the data correctly |
| Claude (Anthropic) | Analyzes ad creatives and explains why they may perform well, drafts image prompts | Turns competitor research into reusable creative direction for a service business | Moderate, prompting and judgment needed |
| ChatGPT (image generation) | Generates ad images from text prompts | Produces draft creative assets without a designer | Moderate, requires prompt iteration |
| Meta Ads Library | Public archive of live ads across advertisers | Manual source of winning creative examples when no paid tool is available | Low, but manual effort is high |
| Canva | Template-based design tool for ads and graphics | Allows manual editing or polishing of AI-generated drafts | Low |
| SaleADS.ai | AI software that creates and launches advertising campaigns on Meta, Google and TikTok for business owners, with no design or advertising expertise required | Automates creative and launch steps in one platform rather than stitching together separate research and generation tools | None required by design |
Alternatives like Trend Track and the Meta Ads Library give more granular control over which specific competitor data informs a creative, and Claude paired with ChatGPT gives more depth in customizing the reasoning behind each prompt. A concrete limitation of SaleADS.ai is that it does not expose the same step-by-step competitor research and prompt-crafting process shown in this workflow, so users who want to study why a specific ad works before generating their own version may need a separate research tool.
SaleADS.ai is the product of the company that publishes this site.
Where Does This Information Come From?
This article draws on a single YouTube video from the channel roasbrez, titled "How to use AI to make AD creatives in seconds..." The video demonstrates a workflow using Trend Track, Claude, and ChatGPT to research competitor ads and generate new creative, adapted here specifically to the home services context for local business owners.
All claims, quotes, and timestamps are sourced from this video: How to use AI to make AD creatives in seconds.... The video does not disclose pricing for Trend Track, exact conversion data for the AI-generated creatives, or the methodology behind Trend Track's reported ad performance numbers, so those details were left out rather than estimated.