I Connected Reflare to a 104k Channel. 216 Videos Were Already Dying
The tool that revives old YouTube videos by swapping thumbnails. I actually plugged a channel in.
YouTube is the only creative job where the catalog dies on purpose.
A musician lives on the back catalog. A filmmaker too. A YouTuber films the next one and lets the old videos rot, even when the algorithm still shows them a little.
I connected Reflare to a channel with 104,000 subscribers and 688 videos. The dashboard spat out 216 videos "to treat" in the last 7 days. That is not a widget. That is a warehouse.
The question I wanted answered is simple: is this a $49/month toy, or does changing only the thumbnail actually bring views back?
What Reflare thinks it is
Reflare.io launched on 7 August 2026. The founder is Hardisk, on YouTube since 2011, documentaries, YouTube strategy on the side. He did not ship a product for years. Then this.
The idea is not new. Big US channels already pay designers to redo old thumbnails so the video re-enters the recommendation graph. A thumbnail is a candle with a smell. Strawberry pulls one pocket of audience, vanilla pulls another. When 50,000 people have clicked, that pocket is empty. You light another candle.
Hardisk priced the manual version, conservative: 50 euros a thumbnail, often 150-200 for bigger creators. Ten variants on 80 videos. 40,000 euros a year, just to keep the catalog alive. Most French creators do not do it. The catalog dies. Reflare is the automation of that ugly job.
What it actually does when you connect a channel
You log in with Google OAuth. Official API, no password sharing, permissions limited to thumbnails and stats. Then three loops.
Scan. It ranks videos by revival potential: retention, click-through rate, daily impressions. The operational rule, confirmed by the team when I asked what to click first: sort by impressions over 7 days. Those are the ones that can still move.
Generate. It learns the art direction of the channel (face, fonts, colors, how you sell the topic) and looks at how similar videos got packaged on YouTube. Then it proposes variants. No prompt. Two clicks. You approve before anything goes live. That part I like.
A/B test. Thumbnails rotate on the real video. Duration is not a guess: 3 days per variant if the video still gets impressions, 7 days if it is quiet. Below a certain volume the app warns you if you ask for too many variants. Not enough data, the test will be too close to call. When a winner is statistically ahead, it locks. The original is kept and put back if you stop.
There is an Impact page that attributes "extra views" with an algorithm they say is the same family Google Ads uses to split ad traffic from the rest, via a Google partner. I have not audited that. Treat the number as their claim, not yours.
The numbers from the founder's own channel
These are Hardisk's, shown in a live demo, not a third-party audit.
On a Lego video: about 20 views a day, then stable around 1,000 a day after a Reflare test. Impact attributed +23,000 views to that one thumbnail change. On a Boeing video: +6,000. On a test still running ("the hidden face of MrBeast"): original thumbnail 2% CTR, 187 impressions, 9 views. A variant at 6.5% CTR, 232 impressions, 31 views. He also mentions "crazy" cases from 100 views a day to 5,000.
Take that as a demo on the founder's catalog, not your next screenshot.
He has an older story from late 2023, before the product, when Reflare was a Google Sheet. An ASMR documentary stuck at 100,000 views for 100 days. New packaging, up to 360,000. Same video, no reshoot. That one convinced him the game exists. The SaaS is the sheet, productized.
How they told me to start (this part is not on the marketing site)
I asked the team the dumb question: I cannot revive 216 videos in one sitting, in which order?
They were unusually concrete.
- Open the attention filter. Videos with the most impressions over 7 days go first.
- Launch several at once, not one per week. The first 5 to 10 on that list.
- Keep the defaults when impressions are low: 7 days, 1 variation. The app already proposes the recommended settings. If you force more variants, you get a warning.
The logic is volume over time, not a miracle in 48 hours. Hardisk's math: on 600 videos, maybe 5% of a test wave "sticks." The catalog wakes up in waves. Some customers churned after two weeks because they expected 20,000 extra views immediately, with tests still running and a Reflare thumbnail already +1 or +2 points of CTR.
The team knows this. They under-estimated how many creators do not understand the platform they upload to. They will ship "Reflare Academy" videos inside the product. That is a real product risk, not a footnote. A thumbnail that clicks more but sells the video worse (average view duration drops) can hurt. YouTube is PVP. Twelve videos on the homepage, one click. If your new packaging pulls the wrong pocket, the video gets shown, then dumped. Reflare does not fix a weak edit.
What it refuses to do
No titles. No first thumbnail for a video that just went out.
Hardisk is stubborn on this and I think he is right for a two-person company. He cites studies (including one from Seoul University) that the thumbnail carries about 70% of click intent. He also does not want to mix title + thumbnail in the same test: too much statistical noise, he cannot look a client in the eye and say "this is the view we added." You can disagree on the 70%. The product already chose.
Same for "we could generate the launch thumbnail." They could. They will not, not yet. Catalog only.
Pricing
Creator: 49 euros/month, 1 channel, 60 AI thumbnails. Pro: 99 euros, up to 3 channels, 200 thumbnails. Studio: 299 euros, up to 10 channels, 600 thumbnails. Yearly is 20% off. Extra thumbnail credits can be bought. An agency seat mode is announced, not live. No public free trial that I saw.
During beta (about 40 creators, six months) Hardisk says a majority would have at least paid back the subscription, even using it passively. Again: founder statement.
49 euros is not a toy. Against 50 euros a freelance thumbnail, the math holds if the tool revives several videos a month. On a 10-video channel, it does not.
WELCOME is 10% off the first month if you go through this link.
Vs TubeBuddy, ThumbnailTest, and YouTube native
TubeBuddy and the old thumbnail testers are manual: you bring the art, they rotate. YouTube is rolling out its own thumbnail comparison. Free. No generation.
If you already have two thumbnails and you only want to test, YouTube is enough. If you want a system that finds which old videos to touch, proposes art in your visual world, rotates, and stops, Reflare is a different product. That is the only comparison that matters.
So, for who?
Yes if you have tens, better hundreds of videos. Some still retain. YouTube still prints them a little. You can leave 49 euros in a tool for a month without expecting a spike on day two.
Also yes for a niche affiliate channel, which surprised me. Hardisk's answer: if other videos on the same topic did 72,000 views and yours did 3,000, there is still a pocket. Reflare is useful until you hit the ceiling of people who care about that subject. Past the ceiling, a new thumbnail does not invent an audience.
No if you have under 10 videos. No if the content itself is weak. No if you need a thumbnail for tomorrow's upload. No if 49 euros actually hurts. No if you want 20,000 views in two weeks.
I connected it. 216 videos waiting. One A/B test already gathering data, 13 days left. I do not have my own winner yet. I will not invent one.
The honest close
The diagnosis is correct. Almost nobody works the catalog. The product is narrow on purpose, and the details (test length, impression-based ranking, no prompt) are more mature than a three-week-old SaaS has any right to be.
On a channel with stock and retention, one month on Creator is a fair bet. On an empty channel, or if you wanted a thumbnail generator for new uploads, walk away.
The 216 number is still sitting there. That is the part that bothers me, in a good way.
(Affiliate link: I may earn a commission if you subscribe through the link above. Same price for you.)
216 videos dying in a 104k subscriber channel, and a $49/month tool that rotates thumbnails to bring them back. The question is not whether it works, it's whether you ship it like a product or a prototype. The demo vs product checklist in the welcome kit shows you the 8 criteria that separate toys from things that earn.