Ad ideas for a new client in an unfamiliar niche
Short answer: Learn what the customers say before you look at what rivals run. Spend half a day reading reviews, forum threads, support questions and sales call notes, write the customer's own words into an "angle map", then check the free Meta Ad Library (Meta's public archive of ads on Facebook and Instagram) and Google Ads Transparency Center (Google's public list of advertisers and ads) to see which angles are crowded. Turn the gaps into about 10 hooks and test a few at a time with a capped budget. Competitor ads show what is running, not what is working.
You have a new client in a niche you know nothing about, and you need ad ideas that sound like the customers, not like a guess.
I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.
The fastest way into an unfamiliar niche is to read what buyers write when nobody is selling to them. Customer language gives you the problems, the words and the objections. Competitor ads then tell you which of those angles are already crowded. Together they turn "I have no idea what to say" into a list of 10 hooks and a plan to test two to five of them. A tool can speed up the looking. Ploy, an AI platform that builds and runs marketing websites, is one option, but its ad research tools are on a waitlist today, so this page gives you a method that works with free tools.
This page may contain affiliate links from which Blur Studio earns commissions.
The step-by-step on collecting rival ads is in competitor ad research with ad libraries. This page is about what to do before that step, when you do not yet know the niche, and about turning what you find into hooks. Platform rules and layouts change, so as of November 2026 check each platform's current screens and rules.
Why customer language comes first
If you start with competitor ads in a niche you do not understand, you copy their words, their mistakes and their blind spots. You also cannot tell which ads are good. Starting with customers gives you a yardstick. It also gives you the exact phrases people use, which is what makes an ad feel like it was written by someone who knows the trade.
The methods below are practitioner advice. No controlled study shows one research method beats another, so treat each as a way to generate ideas to test, not a proven formula.
Where to listen: five sources, cheapest first
| Source | What you get | Watch out for |
|---|---|---|
| Your client's own reviews, emails and sales call notes | The words buyers use, the objections that come up, what made them choose | The client's own bias. Ask for the 10 best and the 10 worst jobs |
| Competitors' public reviews (Google, Trustpilot, app stores) | What buyers praise and complain about in the niche | Quote the idea, not the review. Do not paste a named reviewer into an ad |
| Forums and Reddit (a large discussion site) | Unfiltered questions, fears and slang | Search by hand, and check Reddit's terms before you use its content for anything beyond reading |
| The client's support tickets and enquiry forms | Real questions in real words | Remove names, emails, phone numbers and order details before pasting into any AI tool, and check that tool's terms on how it uses what you paste |
| Five 15-minute customer calls | Why they bought, what they almost did instead | Ask the client to introduce you. People answer an introduced call |
Aim for 30 to 50 real phrases. If you can only get 10, that is still enough to start. Write them in a sheet exactly as written, with the source and date.
Build an angle map
An angle is the reason to buy that an ad leads with. The map is a table that links each customer phrase to an angle, a hook and a first test. Use these columns:
Source | Customer phrase (exact words) | Angle (the reason to buy) |
Rival uses it? (none / some / many) | Hook (first line or first 3 seconds) |
Format (image, video, carousel) | First test | Status: untested
Group the phrases. Most niches have four to six repeating angles, such as price, speed, trust, ease, quality and fear of a bad outcome. You are looking for angles that customers mention often and rivals use rarely.
Then check what rivals run
Now use the libraries to fill the "Rival uses it?" column. What each library shows, as of November 2026:
- Meta Ad Library shows active ads across Meta products. Spend, reach and who paid are shown for social issue, election and political ads, not for ordinary ads. So you cannot see what a competitor spends or who they target.
- Google Ads Transparency Center shows an advertiser's name, location and who pays, with the ads served on Google platforms over a time period. The page lists no results or spend.
- TikTok's Creative Center has a Top Ads page of example ads on TikTok that you can filter by region, objective, ad language and format. Open it for your client's country and read what it shows, because content differs by region.
Set the country filter first. Take 10 to 20 ads per rival and tag the angle in each. A tally such as "9 of 12 lead with price, 1 shows the finished result, none talks about the fear customers mentioned" is more useful than a folder of screenshots. The source pages are Meta's ad library tools page and Google's Transparency Center help page.
Two cautions. First, an ad that has run a long time is a lead, not proof. The libraries do not show results. Second, do not state competitor performance, spend or targeting in a client report. You cannot see them.
Options, from least to most effort
| Option | What you do | Cost as of November 2026 |
|---|---|---|
| No tool | Browse the free libraries once a week and save examples in your sheet | Free, your time |
| TikTok Creative Center and Meta Ad Library | Check where your client's buyers spend time | Free |
| By hand end to end | Reviews, forums and calls, then the angle map | Free, about half a day for a first pass |
| Ploy | An AI workspace for building pages and, in time, ads (see below) | Free $0, Starter $50 a month, Pro $300 a month, Enterprise by quote. Credits do not roll over |
| Freelancer or consultant | Someone who knows the niche writes the angles | Quote or hourly. Ask for a sample of past angle work |
Paid ad-intelligence tools exist and add search across platforms, alerts and history. Prices and features change, so check each vendor's page. I would not buy one until you have done a by-hand pass for two clients and know which columns you use.
Where Ploy fits, and what exists today
Ploy's homepage lists an "Ads" area that includes ad library intelligence, creative at scale, attribution and Meta campaign management. The page for that product, PloyAds, is a waitlist with the words "coming soon" and no launch date, and its sister product PloyGrow is also a waitlist. So do not plan client work around any of those features, and do not expect it to research a niche for you today. For the long version see Ploy's ad tools explained.
What does exist is the Web product: an AI site builder, analytics and forms. Ploy's docs also list a Meta Ads connection on every plan, including Free (its docs were updated October 30, 2026; the pricing page (read November 21, 2026) shows ad integrations as not included on Free or Starter, so open Integrations on a Free workspace and see whether the Meta Ads card says Connect or Upgrade), and say it can create campaigns, ads and creatives when you ask. Google Ads and Reddit Ads connections need Pro. Treat the Meta connection as something to test on a throwaway account with human approval of every budget, audience and word, not as a research tool. Where Ploy fits well is the step after the idea: building the landing page that matches the hook you chose, then reading what visitors do with its analytics. Plans per ploy.ai/pricing as of 21 November 2026 are Free $0 (500 credits a day, up to 2,000), Starter $50 a month, Pro $300 a month and Enterprise by quote. Ploy does not publish a per-action credit price, so start on Free. See Ploy pricing explained.
Turn the map into 10 hooks
Write one hook per angle first, then add variants. These are formats, not guaranteed winners. Nobody can tell you that a given format converts at a given rate, so no figure is given here.
- Customer's exact words. Open with the phrase buyers use for the problem.
- The fear. Name what they worry about, then the fix.
- The number. A real figure from the client's own records, such as "installed in 1 day".
- The proof. A real, permitted customer comment.
- The before and after. Real photos or footage.
- The myth. Correct a belief you saw repeated in forums.
- The question. The question buyers ask most.
- The comparison. A fair statement of what is different, with no unsupported claim.
- The how it works. Three simple steps.
- The offer. A free quote, a trial or a booking, if the client truly offers it.
Use the wording in the hook swipe file for starting lines and batch ad variants from a brief to produce versions fast. Every claim in an ad must be true and supportable. See claims and testimonials and check your country's advertising rules.
Worked example (made-up client and numbers)
Say your new client installs home EV chargers and you know nothing about it. You collect 40 phrases from the client's reviews, a car-owner forum and four customer calls. They group into five angles. Rival ads: 36 collected across five installers.
| Angle (customer phrase, made up for illustration) | Phrases | Rival ads using it (of 36) | Hook to test |
|---|---|---|---|
| Price ("what will this really cost") | 11 | 20 | Skip. Crowded |
| Landlord or building approval ("my landlord won't let me") | 9 | 1 | "No driveway? We handle the building paperwork" |
| Fear of a bad install ("will it trip my breaker") | 8 | 3 | The fear plus a real certified-installer fact |
| Speed ("can I charge by Friday") | 7 | 9 | Test with a fixed install window, if the client can keep it |
| Running cost ("is it cheaper than petrol") | 5 | 3 | Only with the client's own figures |
The gap is approval and install fear. You write 10 hooks, two for each angle except price, and pick the first test: landlord approval against install fear, same offer, same landing page, same image style. The point is the shape of the work, not the made-up numbers.
Test plan with pass and fail rules
Meta has a built-in creative testing feature. A third-party expert write-up by Jon Loomer (published October 2025, updated January 2026) describes it as testing 2 to 5 duplicated ads, with Meta recommending no more than 20 percent of budget, a default of 7 days, and results compared on cost per result. Loomer suggests about 50 optimised events a week for meaningful results. Meta's own help pages are the authority on this feature, so check them in Ads Manager before you rely on any of this. Practitioners also advise changing one thing at a time, so you know what moved.
- Set before launch: the single measure (cost per lead, say), a spend cap and an end date, and the pass line from the client's own account history, for example "at or below the account's current average cost per lead".
- Pass: the test hook meets the pass line by the end date, and the lead quality (calls answered, quotes sent) holds up.
- Fail: two rounds with no hook meeting the pass line. Go back to the angle map, not to more variants of the same angle.
- Stop rule: stop a hook early only if it breaks a rule (a wrong claim, a policy problem), not because of two bad days.
The landing page must repeat the hook's promise on the first screen. See ad landing pages that match the ad and your first month of ads. For the format limits, Meta's feed image spec recommends 4:5 at 1440 by 1800 pixels, 50 to 150 characters of primary text and a headline near 27 characters; LinkedIn's single image ad help page suggests about 150 characters of intro text. Both are recommendations, so check the current spec for the placement you use.
Rules for using rival ads and AI
- Use the pattern, never the asset. Do not copy a rival's script, footage, logo or look. See the rules in the ad library article.
- If you use ChatGPT or Claude to cluster phrases, paste only de-identified text and never client secrets.
- A person approves every ad, claim and budget.
- Do not present AI-generated people or results as real customers.
Who should skip this
- A niche with no searchable customer talk and no ad history. Call five customers and start small.
- A client who will not let you test. Without a test, the angle map is a set of opinions.
- Anyone expecting Ploy to find the ads for them. Its ad library is a homepage bullet and a waitlist today.
- A client whose edge is a person or a place. Shoot the real thing, then write.
How I'd approach this
Day one: get the client's reviews, 10 best and 10 worst jobs and call notes, and read two forums. Day two: build the angle map, then check the free libraries for the "Rival uses it?" column. Day three: write the 10 hooks, pick two angles, and build one landing page per angle that repeats the hook (a page builder such as Ploy's AI site builder is one way, and any page builder works). Week two: run the capped test, then read the result against the pass line you set on day three. Decision rule: if two rounds fail, go back to listening, not to more rival ads.
If you want something custom, Blur Studio builds research tools and reports for a niche, with AI that reads your saved reviews and ads and drafts the angle map for your team to approve. See what a custom build includes.
Related guide: How Distributors Can Compare Competitor Offers and Write Sharper Ads.
Related guide: How to Make a Share Image for Every New Page.
Related guide: A Brand Presentation Deck Made Quickly: Options.
Related guide: Which Landing Page Makes the Sale? Page by Page Tracking.
Questions people ask
What is the fastest way to learn a niche for ads?
Read what buyers write before anyone sells to them: the client's reviews, rival reviews, forums and five short customer calls. Collect 30 to 50 exact phrases and group them into angles. Then use the free ad libraries to see which angles rivals already use.
Does the Meta Ad Library show how well a competitor's ad works?
No. For ordinary ads it shows that an ad is running, not spend or results. Spend and reach appear for social issue, election and political ads. See competitor ad research for what each library shows and hides.
How many ads should I test at once?
Jon Loomer's write-up of Meta's creative testing feature describes testing 2 to 5 ads with no more than 20 percent of budget. Check Meta's current rules in Ads Manager. Change one thing at a time so you know what caused the difference.
Can I use Ploy to find competitor ads?
Not reliably today. Ploy's homepage lists ad library intelligence, but the PloyAds page is a waitlist with no launch date as of November 2026. Use the free libraries and see Ploy's ad tools explained.
Is it legal to copy a competitor's ad idea?
Reusing a general structure, such as a hook followed by proof and an offer, is normal. Copying their footage, script, logo or distinctive look can raise intellectual property and trust problems. Check local law. This is not legal advice.
How do I choose the first hook to test?
Pick the angle customers mention often and rivals use rarely, and set the pass line before launch from the client's own numbers. A crowded angle such as price is a poor first test unless the client has a real price edge.
Can I paste customer emails into ChatGPT to find angles?
Only after removing names, contact details and order details, and only if the tool's terms allow it. When in doubt, group the phrases by hand in a sheet.
More in this topic: Websites: ad ideas, share images and brand decks (4 guides) →