mukeshsharma1106
Member
I have been messing around with different targeting setups for a while, and one thing I kept circling back to was lookalike audiences. Everyone talks about them like they are some magic shortcut, but nobody really explains why they sometimes work great and other times do nothing. So I got curious and started paying attention to whether the actual quality of the lookalike audience was doing more for my gambling advertisements than the campaigns themselves.
At first, I honestly thought I was overthinking it. I figured a 1 percent lookalike is a 1 percent lookalike. The platform builds it. You set it. Done. But when I looked closer at the results across a few of my campaigns, the ROI swings were just too big to ignore. Some sets were giving solid sign ups and first deposits. Others were sending me people who clicked and disappeared. Same ads. Same budget. Same everything except the seed audience I used to generate the lookalike.
That is when the pain point kicked in. I realized I did not really know what made a “good” seed audience for gambling advertisements. Most of my early seed lists were basically any users who converted at some point over a long timeline. It felt logical, but in hindsight, those lists were all over the place. Different deposit sizes, random behaviors, some people who clicked by mistake or never came back after signing up. When I built new lookalikes from those messy lists, the targeting always felt kind of diluted. Lots of impressions, but the wrong kind of people landing on the page.
So I started playing around with different seed lists. The first time I trimmed the seed audience to only my highest value players, the difference was almost immediate. The cost per result did jump a little, but the deposits from the traffic were way more stable. It helped me see that going for cheap conversions was not the same as going for useful conversions. The more I narrowed the seed list to people with steady deposit behavior, the more predictable my ROI became. It did not suddenly turn every campaign into a winner, but the traffic at least acted like it came from people who were interested in gambling, not random scroll-clickers.
Another thing I noticed was recency. Seed lists built from recent activity worked way better than lists pulled from six months of users. I think older lists get stale, especially in gambling niches where a lot of users burn out or switch platforms fast. When I used only the most active recent converters, the lookalikes behaved more consistently. I still kept a larger seed audience, but it was made of people who actually behaved the way I wanted new users to behave.
I also tested different lookalike percentages. That surprised me too. I always assumed 1 percent was the “best” by default. But when the seed audience was strong and consistent, widening to 2–3 percent did not hurt performance the way I expected. In fact, those wider ranges sometimes worked better because they gave the system more room to find similar behaviors without overfitting. On the other hand, when the seed audience was low quality, widening the percentage made everything fall apart. So the lookalike size only worked well when the base audience was already clean and useful.
All of this led me to a softer conclusion: the strength of your lookalike audience depends almost entirely on the effort you put into cleaning and selecting your seed audience. It is not a switch you flip. It is more like tuning a guitar. If the seed is sloppy, the whole thing plays off-key. When the seed is tight, the campaigns feel easier to manage and adjust. It is not a magic trick, but it does reduce guesswork.
If anyone else is trying to figure out why their gambling advertisements sometimes hit or miss, I would say start by trimming your seed lists instead of tweaking your creatives over and over. The ads matter, sure, but the people you aim them at matter even more. I found this explanation helpful when I was trying to understand the mechanics behind it, and it breaks down the reasoning pretty clearly. Sharing it here in case someone else is stuck in the same loop: improve ROI with quality lookalike audiences.
After all this testing, I do not think lookalike audiences are the perfect answer for every gambling ad setup. But I do think the quality of the seed audience has a bigger impact on ROI than most people give it credit for. Once you tighten that part, everything else becomes a lot easier to evaluate. It gives you a cleaner baseline to compare your creatives, your landing pages, and even your bids. If your seed audience is random, everything that comes after it feels random too. That is the part I wish I had understood earlier.
At first, I honestly thought I was overthinking it. I figured a 1 percent lookalike is a 1 percent lookalike. The platform builds it. You set it. Done. But when I looked closer at the results across a few of my campaigns, the ROI swings were just too big to ignore. Some sets were giving solid sign ups and first deposits. Others were sending me people who clicked and disappeared. Same ads. Same budget. Same everything except the seed audience I used to generate the lookalike.
That is when the pain point kicked in. I realized I did not really know what made a “good” seed audience for gambling advertisements. Most of my early seed lists were basically any users who converted at some point over a long timeline. It felt logical, but in hindsight, those lists were all over the place. Different deposit sizes, random behaviors, some people who clicked by mistake or never came back after signing up. When I built new lookalikes from those messy lists, the targeting always felt kind of diluted. Lots of impressions, but the wrong kind of people landing on the page.
So I started playing around with different seed lists. The first time I trimmed the seed audience to only my highest value players, the difference was almost immediate. The cost per result did jump a little, but the deposits from the traffic were way more stable. It helped me see that going for cheap conversions was not the same as going for useful conversions. The more I narrowed the seed list to people with steady deposit behavior, the more predictable my ROI became. It did not suddenly turn every campaign into a winner, but the traffic at least acted like it came from people who were interested in gambling, not random scroll-clickers.
Another thing I noticed was recency. Seed lists built from recent activity worked way better than lists pulled from six months of users. I think older lists get stale, especially in gambling niches where a lot of users burn out or switch platforms fast. When I used only the most active recent converters, the lookalikes behaved more consistently. I still kept a larger seed audience, but it was made of people who actually behaved the way I wanted new users to behave.
I also tested different lookalike percentages. That surprised me too. I always assumed 1 percent was the “best” by default. But when the seed audience was strong and consistent, widening to 2–3 percent did not hurt performance the way I expected. In fact, those wider ranges sometimes worked better because they gave the system more room to find similar behaviors without overfitting. On the other hand, when the seed audience was low quality, widening the percentage made everything fall apart. So the lookalike size only worked well when the base audience was already clean and useful.
All of this led me to a softer conclusion: the strength of your lookalike audience depends almost entirely on the effort you put into cleaning and selecting your seed audience. It is not a switch you flip. It is more like tuning a guitar. If the seed is sloppy, the whole thing plays off-key. When the seed is tight, the campaigns feel easier to manage and adjust. It is not a magic trick, but it does reduce guesswork.
If anyone else is trying to figure out why their gambling advertisements sometimes hit or miss, I would say start by trimming your seed lists instead of tweaking your creatives over and over. The ads matter, sure, but the people you aim them at matter even more. I found this explanation helpful when I was trying to understand the mechanics behind it, and it breaks down the reasoning pretty clearly. Sharing it here in case someone else is stuck in the same loop: improve ROI with quality lookalike audiences.
After all this testing, I do not think lookalike audiences are the perfect answer for every gambling ad setup. But I do think the quality of the seed audience has a bigger impact on ROI than most people give it credit for. Once you tighten that part, everything else becomes a lot easier to evaluate. It gives you a cleaner baseline to compare your creatives, your landing pages, and even your bids. If your seed audience is random, everything that comes after it feels random too. That is the part I wish I had understood earlier.
231 Nguyễn Văn Cừ, Phường Chợ Quán, TP. Hồ Chí Minh
