How Dyme runs its Meta ads
All 425 ads in Dyme’s library, August 2024 to November 2026: who they target, what they say, what happens in their videos, how long ads run and where the clicks go.
Market: Netherlands412 of the 425 ads reached only the Netherlands, by Meta’s EU reach data. The other 13, all launched since 27 July 2026, show no country: Meta gives no reach data for them and they did not appear when we searched the Netherlands, Belgium, Suriname, Curaçao, Aruba, the UK or the US. They are in Dutch and link to the same app store pages, so we keep them in.
Click any card row week or picture to open the ads behind it.
An ad is proven when it is among their 100 most-seen ads and ran 30 days or more. 36 ads started less than 29 days before 2 November, so they are too new to judge and left out of every proven share. Meta marks 29 ads active; 7 of those last delivered more than 2 days before 2 November, so we count 22 as live.
Their best-ranked creatives, one per creative. Hover to pause, or click one to open it. Rank is the ad’s place when Meta sorts all 425 ads by impressions, with 1 the most seen. Ads that share a creative can share a place.
A Dutch money app built on subscriptions
Founded in Amsterdam in 2018 and bought by the insurtech RISK in September 2025. The app links to your bank account, shows your subscriptions, fixed costs and spending, cancels subscriptions for you and helps you budget. 397 of its 425 ads are videos, and every ad links to an app store.
What their ads are for Their campaign types, each with one of its best-ranked ads. Click one for its ads.
Sources: Silicon Canals, 10 September 2025 for the founding, the user count and the sale to RISK; Wikipedia for the PSD2 licence; their App Store and Google Play pages and their homepage, read on 2 November 2026. The user count is their own claim, not checked by us.
Five things their ads show
Proven means among their 100 most-seen ads and ran 30 days or more: a sign of where Meta put reach, not of downloads. The library shows no spend, results or audiences.
Two in three ads are about forgotten subscriptions
291 of their 425 ads (68%) sell one job: finding the subscriptions you forgot. Spending overview and budget ads come second (87 ads) and are proven slightly more often: 24% against 22% for all ads. Cancelling, insurance and tips stay small.
Proven means in the first 100 places of Meta’s impressions order for all 425 ads on 2 November, and ran 30 days or more. With a few hundred ads we use a top 100, not a top 300. Campaign types with fewer than 10 ads old enough to judge get no share. A campaign type is what the ad sells, read from its message. New ads per week count the 117 weeks from 12 August 2024 to 2 November 2026.
Most ads target people who forget their subscriptions
233 ads (55%) target people who lose track of their subscriptions, mostly through street interviews. Budgeters stuck in spreadsheets get far fewer ads (35) but the highest proven rate: 37%. Young earners and couples get skits and scenes that are rarely proven (16% and 9%).
Each customer type Share of each type’s ads that are proven (all ads: 22%). Click a row to see its ads.
We coded each ad’s customer type from the video or picture: who is in the scene and what situation they are in. The ad text is short and the same on most ads, so it rarely names a customer type. Types with fewer than 10 ads old enough to judge get no share.
Old-way openings are proven most often
Ads that open with a spreadsheet or a list in the notes app are proven most often: 13 of 30 (43%). Street challenges, their most common opening, score 28%, and skits 11%. Across all ads it is 22%. Pick a customer type to see what they say to it.
Old-way openings are proven most often
46 of its 52 proven ads use the “street interview” format.
How their best-ranked ads open In Dutch, with our English translation. Videos with a checked spoken opener come first, then the rest, each best-ranked first. Spoken openers are quoted only where two separate transcriptions agree word for word; otherwise the text on the picture or the first sentence of the ad text. Click one to open it.
Its proven ads are spread over several formats. “salary and budget story” and “talking to camera” have 2 each. Their most common format for this type is “skit” (19 of 55 ads).
How their best-ranked ads open In Dutch, with our English translation. Videos with a checked spoken opener come first, then the rest, each best-ranked first. Spoken openers are quoted only where two separate transcriptions agree word for word; otherwise the text on the picture or the first sentence of the ad text. Click one to open it.
Its proven ads are spread over several formats. “couple or friends talking money” and “skit” have 2 each.
How their best-ranked ads open In Dutch, with our English translation. Videos with a checked spoken opener come first, then the rest, each best-ranked first. Spoken openers are quoted only where two separate transcriptions agree word for word; otherwise the text on the picture or the first sentence of the ad text. Click one to open it.
7 of its 13 proven ads use the “picture: the old way” format.
How their best-ranked ads open In Dutch, with our English translation. Videos with a checked spoken opener come first, then the rest, each best-ranked first. Spoken openers are quoted only where two separate transcriptions agree word for word; otherwise the text on the picture or the first sentence of the ad text. Click one to open it.
All 3 of its proven ads use the “skit: asking mum or a partner to cancel” format.
How their best-ranked ads open In Dutch, with our English translation. Videos with a checked spoken opener come first, then the rest, each best-ranked first. Spoken openers are quoted only where two separate transcriptions agree word for word; otherwise the text on the picture or the first sentence of the ad text. Click one to open it.
Message and opening are read from the video or picture for every ad: 227 video creatives from four frames each (0, 1.5, 3 and 6 seconds) and a machine transcript of what is said, 28 pictures from the picture. Proven means in their top 100 by impressions on 2 November and ran 30 days or more. Openers are cut to 18 words; translations are ours.
Street interviews carry their video ads
Almost half their ads (205, 48%) are street interviews, including 12 of the 22 ads live on 2 November. Ads that put a money question to a stranger are proven 27% of the time; skits, couple scenes and talking heads 11%.
One creative is one video, or a group of near-identical cuts of the same video, or one picture: 255 creatives across 425 ads. We read the format from four frames of each video and from what is said. Quotes are spoken openers where two separate transcriptions agree word for word. Formats with fewer than 10 ads old enough to judge get no share.
Always on since autumn 2025, and growing
Before July 2025 the library holds just 7 ads, among them their three most-seen pictures. From the week of 1 October 2025 they launched new ads in 52 of 58 weeks, with launches peaking in April and June 2026, and 12 health insurance ads before the yearly switch in autumn 2025.
Their testing calendar For each month from July 2025, how many new ads they launched and whether those ads started a new concept or joined one already running. Before July 2025 the library holds 7 ads, which started 3 concepts.
Orange bars are subscription ads, green bars spending overview ads; grey is everything else. The EU library keeps stopped ads for about a year, so the few ads before October 2025 are the ones that kept running. An ad counts as running if it was shown on at least one day that week. The last week holds one day, 2 November.
One concept holds over half their ads
Almost every ad offers the free app, so a concept is really a customer type with a message. Their biggest, people who lose track of subscriptions told they pay for more than they think, holds 231 of 425 ads. A typical concept lasts about ten months, a typical ad 10 days: they keep feeding a concept new creatives.
Inside a concept Their 18 concepts, biggest first. Pick one to see how it is built.
1Pick a concept Filter by campaign type, then click a card.
2See how it is built Click the concept, a picture or a page to open its ads.
The library shows no ad sets, so we use concepts instead. A concept is one customer type with one message and one offer, the closest thing to an ad set the library shows. The growth playbooks we checked against define a concept the same way (persona × angle × offer). 371 of their 425 ads are flexible ads, which Meta can show with several images or videos under one ad text; we code the first creative of each.
Three in four ads ran under 30 days
315 of their 425 ads (74%) ran under 30 days; the median ad ran 10 days. Their most-seen ads ran longer: 84 of their top 100 ran 30 days or more. Longer runs collect more impressions, so this does not prove they did better.
Their longest-running ads one per creative
12 ads were short bursts: old enough to judge, among the 100 most-seen, but stopped within 30 days. . Read with care. The library shows no spend. Run days count every calendar day from the first to the last delivery date.
Nine in ten ads open to grab attention
386 ads (91%) open to grab attention, mostly with a street challenge or a skit. 33 build trust by showing the old way and are proven most often (39%). Only 6 open with what the app does, and none is proven.
We read the stage from how each ad opens, in the video or picture: a challenge, a scene, tips or a salary story grab attention; the old way and customer tweets build trust; a claim about what the app does, even with a star rating, asks for the download. This is the ad’s role, not who saw it: the library does not show audiences or retargeting.
Every ad sends people to an app store
Two in three ads link to the App Store, one in three to Google Play. None links to their website, not even for the health insurance switch their homepage promoted on 2 November.
Their landing pages The two pages their ads link to, and their homepage for comparison. Screenshots on a laptop screen, 2 November 2026.
Screenshots taken on 2 November 2026 at 1280 px wide, in the Dutch App Store, Google Play and on dyme.app, with non-essential cookies declined. Ratings as shown on those pages that day.