• ChunkMcHorkle@lemmy.world
    link
    fedilink
    arrow-up
    18
    arrow-down
    1
    ·
    1 年前

    As an advertiser

    I have a serious question for you, if you have a moment. Do advertisers have any way of knowing what percentage of the views they’re paying for are actual humans, and what are bots?

    Because it seems to me that this is an excellent scam on a corporate level: Reddit ditches users and mods in favor of bots interacting with bots, the number of accounts and views don’t dip dramatically, and Reddit, Inc. continues to pull in all that sweet advertising revenue because there’s no way for advertisers to know the difference for sure, or the ratio of bot to humans on the site or in a sub with any kind of precision.

    I’d appreciate your thoughts on this, because I’ve been pondering this for a while but do not have any knowledge of advertising metrics, or what would stop a dishonest/bad-faith board like Reddit’s from doing this to some degree just because they can.

    • bleph@lemmy.world
      link
      fedilink
      arrow-up
      13
      ·
      edit-2
      1 年前

      Not an advertiser but they generally know % of views (“impressions”) to clicks (called click through rate) and percentage of clicks that turn into sales (called conversion rate).

      For that reason, I don’t think they’re trying to get rid of human users completely, just the “troublemakers”.

      I think they want to lead the “silent majority” users into a bot advertorial content hellscape where they control all the levers of power and everything is for sale.

      • ChunkMcHorkle@lemmy.world
        link
        fedilink
        arrow-up
        16
        arrow-down
        1
        ·
        edit-2
        1 年前

        Thank you, I think you’re right. Interesting you mentioned click thru rate though, because another commenting advertiser here on Lemmy noticing weird shit with Reddit lately brought that up, saying his click through rate was good but then when he looked into there were many immediate abandons, and someone else explained that’s because people were getting tricked by the ads that look like posts and immediately backing out once clicked.

        I’d be happy to find the comment for you but I have no idea how to find shit here yet, lol. I’ll look; if I find it I’ll edit this comment with a link.

        EDITED TO ADD I think this is it: https://lemmy.world/comment/644214 (see the other posts by the same guy also if you’re interested, like this one https://lemmy.world/comment/652045 and https://lemmy.world/post/837198)

        • OsrsNeedsF2P@lemmy.ml
          link
          fedilink
          arrow-up
          4
          ·
          edit-2
          1 年前

          Yup, I can echo what that commenter said. The bounce rate (when someone clicks on your ad) is atrocious, and there is extremely high competition for getting ads to US/Canada/Australia/other high income countries, which drives up the price further.

          I only advertise on Reddit because I have a really great discount, but even then it sucks and always has.

    • festus@lemmy.ca
      link
      fedilink
      English
      arrow-up
      10
      ·
      1 年前

      Ever clicked on a link and noticed that the URL ended in something like ?campaign=twitter or something? Advertisers regularly track which advertising campaign got a user to click on a link, and they’ll also track what proportion of those users eventually lead to a sale. If Reddit eventually has no users and just bots, advertisers will quickly notice that ad spending on Reddit isn’t producing profit and kill it.

    • Buckeye@lemmy.world
      link
      fedilink
      arrow-up
      5
      ·
      1 年前

      Not really (at least at my company). You pay for the campaign (display this content in this location for these dates) and you track your outcomes (number who viewed the ad, number who clicked, number who shopped, number who purchased). If the number who shopped and purchased and is low you might not be interested in continuing that partnership.

      I always recommend based on shop and buy (heavier focus on buy) outcomes so I wouldn’t know bots but they’d need to be able to make purchases.

      • Boz (he/him)@lemmy.one
        link
        fedilink
        arrow-up
        2
        ·
        1 年前

        …so the endgame here is bots that can make purchases, but immediately return what they bought for refunds?

        • Buckeye@lemmy.world
          link
          fedilink
          arrow-up
          3
          ·
          1 年前

          Lol I guess maybe! My industry isn’t set up that way (I don’t work in retail e-commerce) but that’s obviously the bigger ad targets on social. Retail can definitely track return metrics though.

          I think it would be hard to get bots past most sophisticated purchase data tracking, but it depends on what you target for that tracking. Like I know a lot of TikTok marketing is built around understanding you aren’t going to get a lot of click throughs on the ad but it is about building brand awareness. If you are just looking at impressions it is a lot easier for bots to sneak in.

          • Boz (he/him)@lemmy.one
            link
            fedilink
            arrow-up
            2
            ·
            1 年前

            I would definitely expect purchase tracking to be stricter than ad view metrics, yeah. And I know a lot of companies that used to have customer-friendly return policies have rolled back most or all of those policies (with or without non-enhshittification reasons). So I was at least partly joking, though I am getting less and less surprised about what bots are able to do. AI technology advancement just keeps accelerating.

            Good point about brand awareness. In all seriousness, I think there’s psychological research suggesting that brand awareness is valuable in and of itself, though I think there’s a limit to how negative the publicity can be and still be valuable to brands. Otherwise, I don’t think there wouldn’t be the concept of “brand safety” for ad placement.

            I feel like bots are basically the optimal tool for cheating automated systems, since it seems reasonable to fight automation with more automation, like the cat-and-mouse development race for captcha. I don’t have the technical expertise to back that up, though, just a general feeling that Murphy’s law applies to all engineering.