Productivity · by Anthropic PBC
Claude by Anthropic Meta Ads & Creative Strategy
Every long-running ad Claude by Anthropic runs on Facebook & Instagram — pulled from the public Meta Ad Library, x-rayed by the same system that runs ad accounts autonomously.
Active ads
686
Winning set
65
30+ days live
Longest live
88d
App rating
4.72★
258K ratings
Brand palette
The short version
Key facts
- As of September 17, 2026, Claude by Anthropic runs 686 active ads in Meta's Ad Library across Facebook and Instagram; 65 of them have survived 30 or more days of continuous delivery.
- Its longest-running ad has been live for 88 days.
- The winning set splits 48% video, 12% static, 40% carousel.
- The most common message angle is feature reveal, carried by 31% of the winning set.
- All winning copy runs in English.
- In App Store reviews, the loudest praise goes to quality & accuracy of answers (11% of mentions) and the loudest complaint is reliability & crashes (17%).
- The App Store listing holds 4.72 stars across 257,780 ratings.
The winning set
Live Meta ads gallery
65 ads have survived 30+ days of continuous delivery, out of 686 currently active. Below: 30 creatives picked to span every angle and buyer the account runs; the ×N badge counts the ads running each one. Tap any card for its Meta Ad Library record. Sorted by longest live: the account's most proven creatives lead, podium first.
88d video ×3 ads EN
Essentially Claude writes the code.
83d video ×2 ads EN
Now Claude can collaborate right alongside your team.
83d video EN
Now Claude can collaborate right alongside your team.
75d video EN
I keep buying bedding I don't need because the colors never match what I already have.
67d dco ×2 ads EN
Your Monday mornings are for thinking again
67d video ×2 ads EN
The engineers adopt Claude Code first, and then the eng adjacent roles look over their shoulder and they're like, whoa, this thing is very powerful. Let me try
67d video ×2 ads EN
For us Claude is just at the center of everything. Whenever I have a question, I ask Claude.
67d dco EN
Here for the 3am deploy
67d video EN
The engineers adopt Claude Code first and then the eng adjacent roles look over their shoulder and they're like, whoa, this thing is very powerful.
67d dco EN
Here for the merge conflict
67d dco EN
Here for the final commit
67d dco EN
Here for the last refactor
67d dco ×2 ads EN
Keep the conversation flowing between your brief and your deck
66d carousel EN
Your Monday mornings are for thinking again
62d image EN
Here for the 3am deploy
62d image EN
60d video EN
59d video EN
59d image EN
59d image EN
59d image EN
58d video EN
57d image EN
57d image EN
41d video EN
41d video EN
41d video EN
39d video ×2 ads EN
39d carousel EN
33d video ×2 ads EN
+22 more winning creatives in the full set, with every hook and persona call — free with a Google account.
Browse the full set ↗ or claim the Claude by Anthropic workspace
The x-ray — every winner analyzed frame by frame
What the winning set is actually doing
Language distribution — where the winning set actually runs
Share of the winning set per copy language. Translation is a deliberate spend decision: a language earning survivors is a market the money already believes in.
English
100% · 65 ads
Format split — statics vs video in the winning set
What survives 30+ days by medium.
video 31 ads · 48%
static 8 ads · 12%
carousel 26 ads · 40%
Angle mix — what the money says works
Every winner tagged with the message angle it leads with, and the languages it runs in — longevity under spend is the only public vote that counts.
feature reveal
31% · 20 ads
product positioning brand
28% · 18 ads
unclassified
20% · 13 ads
time freedom workload
6% · 4 ads
ai capability showcase
5% · 3 ads
social proof
5% · 3 ads
all in one platform
5% · 3 ads
urgency scarcity
2% · 1 ads
The winning set, read from the creatives themselves — the top 15 analyzed winners, hook by hook.
| Days live | Ads | Lang | Angle | Persona | Opening hook | On-screen headline | Length |
|---|---|---|---|---|---|---|---|
| 88d | ×4 | en | product positioning brand | developer_technical_builder | Essentially Claude writes the code. | Meet your new thinking partner | 18s |
| 83d | ×3 | en | feature reveal | developer_technical_builder | Now Claude can collaborate right alongside your team. | Tag @Claude in | 56s |
| 83d | ×3 | en | feature reveal | developer_technical_builder | Now Claude can collaborate right alongside your team. | 56s | |
| 75d | ×3 | en | ai capability showcase | developer_technical_builder | I keep buying bedding I don't need because the colors never match what I already have. | 62s | |
| 67d | ×4 | en | social proof | team_manager | The engineers adopt Claude Code first, and then the eng adjacent roles look over their shoulder and they're like, whoa, this thing is very powerful. Let me try | Meet your new thinking partner | 22s |
| 67d | ×4 | en | all in one platform | developer_technical_builder | For us Claude is just at the center of everything. Whenever I have a question, I ask Claude. | Meet your new thinking partner | 25s |
| 67d | ×4 | en | social proof | team_manager | The engineers adopt Claude Code first and then the eng adjacent roles look over their shoulder and they're like, whoa, this thing is very powerful. | Meet your new thinking partner | 42s |
+37 more winners, hook by hook, with per-ad angle and persona calls — free with a Google account. Browse the full set ↗ or claim the workspace
Creative strategy
The dissection
Of 456 active Meta ads for Claude, only 21 survive past 30 days, and the longest-runner (59 days) opens on 'it doesn't work while I sleep.' This week added 6 new winners and retired 21, while the real workhorse stays a ~5-second, no-people product clip anchoring 11 of those 21 survivors.
Five Seconds, No People, Eleven Wins
Always-on support claim is the account's engine: 11 of the 21 winning ads run through just 7 x-rayed creatives, a video clocking in around 5 seconds with no presenter at all, only product and app footage on screen. It opens on the blunt frustration line 'It's cool like it's Google search, but the problem is it doesn't work while I sleep.' (59d), tying the corpus's longest survivor mark.
One Woman, Two Minute-Long Pitches
diy app build proof (4 ads, 4 creatives) runs ~62 seconds behind a female 25-34 presenter telling the Gmail Apps Script story; slow down thinking claim (3 ads, 2 creatives) runs ~58 seconds behind the same demographic delivering 'We need to rebel against time by thinking slower.' (54d, 49d). Of 9 on-camera casts logged, female 25-34 fills 4, tied with male 25-34's 4, so the long-form testimonial slot is hers by pattern, not chance.
Late-Night Engineers Fill The Feed
Late-Night Software Engineer covers 11 of 21 winning ads, the persona behind lines like 'The engineers adopt Claude Code first, and then the eng adjacent roles look over their sho' (46d, team-wide adoption proof, 3 ads). Startup Team Lead holds 3 ads, Eng-Adjacent Automator 2, DIY Non-Coder Automator 2. Creative Partner Seeker sits at 0 ads: a full persona gap inside a 21-ad set.
Ten Ads Say Nothing At All
Only 9 unique primary texts cover the entire winning set, and the top one, literally '(no body video),' runs on 10 of 21 ads. '(no body image)' adds 2 more silent entries, meaning over half the winners carry no written hook at all. The Claude Code line, 'The kind of partnership that makes you feel like you've leveled up,' trails far behind at just 3 ads.
The Sleep Claim Meets Rate Limits
Across 2,492 reviews (4.72★ lifetime), usage and rate limits draw 65 mentions and reliability/crashes 75, with vs-ChatGPT comparisons topping the voice themes at 111. Reviewers describe caps hit 'within minutes to a few messages, even on paid plans' and chats lost to crashes, the opposite of the always-on claim carrying 11 of the account's 21 winners. None of the sampled ad copy mentions limits, crashes, or price.
What we'd test next
- Produce a ~5-second, no-people product/app-footage video (the format behind 11 of the 21 winning ads) using a pattern interrupt hook for the team-wide adoption proof angle (3 ads), targeting the Startup Team Lead persona (3 ads, no dedicated signature execution yet).
- Cut the slow down thinking claim's ~58-second testimonial (3 ads, 2 creatives) into a <15s version with the same female 25-34 presenter and a pattern interrupt hook, run against the Reflective Deep Thinker persona (3 ads), who currently only sees the long-form cut.
- Shoot a screen-recording, day-in-the-life app walkthrough video, a mechanic absent from the six hook types logged (pattern interrupt, testimonial, relatable developer pain point, text-based problem statement, problem-solution, philosophical statement), for the diy app build proof angle, targeting the Eng-Adjacent Automator persona (2 ads).
- Test a testimonial-format video built around a creative-collaboration angle (Claude as brainstorming partner, absent from the current 4-angle set) aimed squarely at the Creative Partner Seeker persona, sitting at 0 of 21 ads despite the winning set's size.
The HyperScale batch
New ad concepts, generated for Claude by Anthropic
HyperScale agents are producing an original round of ad concepts for Claude by Anthropic, built from the winning angles above, the brand's real App Store assets, and what reviewers actually say. This page updates when the batch drops.
This analysis lives in a ready-made HyperScale workspace. If you work on Claude by Anthropic, it's yours: we hand it over personally, usually within a day.
Customer voice — 2361 recent reviews analyzed
What reviewers actually say
Objections the ads must pre-empt
- Rate/usage limits often hit within minutes, even on paid Pro/Max plans
- Server outages and crashes make the app unusable for hours at a time
- Accounts can be suspended or age-verified/banned without warning or support response
- Pricing is seen as high relative to the limited number of messages allowed
- Support is slow or automated, with users waiting days for a human reply
Review-theme mix
Share of analyzed review mentions per theme — ＋ praise, − complaint, ~ mixed. Themes, not individual reviews.
− usage limits & rate limits
15% · 65
− pricing value
11% · 49
− reliability & crashes
17% · 75
~ vs ChatGPT comparison
26% · 111
+ quality & accuracy of answers
11% · 47
− account bans & support
8% · 36
+ productivity & capability
6% · 24
− app performance issues
6% · 28
“Claude changed my career by multiplying my hourly rate by 20x and empowering me to accomplish previously impossible tasks”
“I've been using ChatGPT non stop these last 2 years and I thought it was the best, but then I started using Claude Ai”
“I'm able to outsource 80% of my projects to an intelligent thought partner”
“Switched to a paid subscription on Claude instead of ChatGPT. Claude is vastly better and more intelligent.”
“realizar el 80% del trabajo pesado u operativo y tú dedicarte a la estrategia”
“Claude is better than ChatGPT in some ways and overall if I had to pick one, I'd say Claude is the best at the moment”
“won me one tax review and a 34k deductible, and a small court case”
“Claude mogs chatgpt and every other ai to another dimension”
“I basically have a super intelligent PA 24 hours of the day!”
“infinitely better than the insufferable personality of ChatGPT 5.2”
“Claude is better than ChatGPT in some ways and overall if I had to pick one, I'd say Claude is the best at the moment”
“Ich finde Claude toll und 200% besser als ChatGPT”
“Switched to a paid subscription on Claude instead of ChatGPT. Claude is vastly better and more intelligent.”
“Claude changed my career by multiplying my hourly rate by 20x and empowering me to accomplish previously impossible tasks”
“Claude mogs chatgpt and every other ai to another dimension”
“I'm able to outsource 80% of my projects to an intelligent thought partner”
“infinitely better than the insufferable personality of ChatGPT 5.2”
“realizar el 80% del trabajo pesado u operativo y tú dedicarte a la estrategia”
“If you were to say something wrong it corrects you unlike ChatGPT”
“I basically have a super intelligent PA 24 hours of the day!”
“Far more competent than gpt, far less parasitic than gpt.”
“won me one tax review and a 34k deductible, and a small court case”
“Claude is the adult in the room while ChatGPT is your frenemy”
“Much more capable, quality work, no hallucinations”
“Ich finde Claude toll und 200% besser als ChatGPT”
“Claude is the best at following instructions”
“Much more capable, quality work, no hallucinations”
“Claude is the adult in the room while ChatGPT is your frenemy”
“Excellent user help for Powershell, 97% accurate in suggested code.”
“Far more competent than gpt, far less parasitic than gpt.”
“How is this still acceptable to be paying over $200 a month and I can only get in 5 - 7 questions at best every 5 hours!!!”
“Excellent user help for Powershell, 97% accurate in suggested code.”
“Got THREE prompts out of the basic model and ran out of messages and need to wait 5 hours. This is ridiculous”
“After paying for a Pro membership the first question about analyse 120 page pdf and I got limit!!!”
“Got max 2 prompts out of Claude before it told me I had to wait over 4 hours”
“How is this still acceptable to be paying over $200 a month and I can only get in 5 - 7 questions at best every 5 hours!!!”
“Claude has so many limits its so bad on the free version you get like 30 messages which sounds good but if your like me you will run out fast”
“Got THREE prompts out of the basic model and ran out of messages and need to wait 5 hours. This is ridiculous”
“Paying 35 Aud a month for 2-3 questions every 3 hrs”
“Claude has so many limits its so bad on the free version you get like 30 messages which sounds good but if your like me you will run out fast”
“Rip off. Overpriced. 14 sessions to make a simple website”
“Paying 35 Aud a month for 2-3 questions every 3 hrs”
“Even with max $175/month, can't get files off, no MS word capacity, have to copy & paste in small paragraphs”
“server outages lasting several hours occur frequently, and despite charging at least $34 per month, there are many times when the service is completely unusable”
“Never have seen a ai program with so many bugs or being down so often”
“Terrible, unreliable and continually stops works”
“Now it's like other platforms. Hallucinating. Losing a days work. Just forgot the chat. Crashing.”
“Now it's like other platforms. Hallucinating. Losing a days work. Just forgot the chat. Crashing.”
“Claude has gone downhill rapidly, guessing servers can't handle the popularity. Constant glitches and no longer works in my phone”
“after 10 minutes of spinning wheels I can't even get a basic answer”
“1 day into using claude the account got banned, how ridiculous”
“i got banned for being "underage". im 18”
“it said i was 17, banned me, and told me to id verify. i cant”
“1 day into using claude the account got banned, how ridiculous”
“the input bar completely disappears with my 5minute voice transcription is actually insane and beyond frustrating”
“I have been trying to change my Premium seat to Standard for 2 months. I cannot speak to anyone”
The market around it
Brands competing for the same feed
Keep exploring the radar
More Meta ads teardowns
More Productivity advertisers on the radar
Also winning with the feature reveal angle
Browse all 87 apps on the radar · All Productivity teardowns
Quick answers
FAQ
How many Meta ads is Claude running, and how many are winning?
Claude's Meta Ad Library shows 456 active ads, but only 21 have survived past the 30-day mark HyperScale tracks as 'winning.' The longest-running ad has held for 59 days, and this week added 6 new winners while retiring 21 older ones.
What angles and languages do Claude's winning ads use?
All 21 winning ads run in English. Always-on support claim leads with 11 ads, diy app build proof follows with 4, and team-wide adoption proof plus slow down thinking claim each hold 3 ads.
What creative pattern is actually working for Claude?
Video dominates at 81% of the 16 x-rayed creatives, with a presenter on camera in 56% of them. The winning always-on support claim angle breaks that norm: its ~5-second clips run with no people at all, just product and app footage, across 7 of the 21 x-rayed creatives.
What do Claude reviewers complain about that the ads don't mention?
Among 2,492 reviews (4.72★ lifetime), vs-ChatGPT comparisons draw 111 mentions, reliability/crashes 75, and usage/rate limits 65, with users reporting caps hit 'within minutes to a few messages, even on paid plans.' None of the winning ad copy addresses limits, crashes, or pricing.