Bots are getting good at mimicking engagement • Hacker News Board

Re: Bots are getting good at mimicking engagement

#271

Post by nodja » Thu, Oct 16, 2025, 12:20 AM UTC

seviu wrote: post #68

simul007 wrote: post #2

Hi HN. I run a marketing agency and fell down this rabbit hole after a client's analytics made no sense (50k visitors, 47 sales). I ended up building a simple script to track user behavior and analyzed 200+ small e-commerce sites. The average was 73% bot traffic that standard analytics counts as real. The bots are getting creepily good at mimicking engagement. I wrote up my findings, including some of the bizarre pat…

I once worked for the yellow pages in Switzerland. Our paid clients had a dashboard which reported how many users visited their business entry. We at engineering decided to filter out bots. Figures fell dramatically by more than 50%. In less that a day business mandated us to remove the filter. Bots are real people after all

Couldn't you kept it in and labeled it as bots? i.e. using stacked barcharts and such.

Re: Bots are getting good at mimicking engagement

#272

Post by harshreality » Thu, Oct 16, 2025, 12:21 AM UTC

cortesoft wrote: post #230

harshreality wrote: post #215

Earlier quoted context omitted.

"Good bots" would be ones that fuel a healthy, competitive economy. Whether or not they're good from the perspective of a retailer, data about products and supply chains is important. "Bad bots" would be ones that primarily support grift. They siphon money from advertisers, or get other people to waste advertising $, or game product listing metrics so that the bots' owners products get better placement, or so that ot…

I can see many bots falling into the two categories you describe, but I can also see many other bots that would not fit easily, and whose value is debatable. For example, bots that scrape content for AI training; is that good or bad for a healthy economy? AI can be productive, but is it ‘stealing’ other’s productivity, which could hurt in the long term if it causes decrease in future human production because of dimin…

"Good" vs "bad" was not intended to be an existential question about whether some bot-powered activity like AI training will be a net good or bad 20 or 100 years from now.

It's not even a question of whether, in that particular case, AI will displace some forms of human labor, even if that causes a macroeconomic crisis where society needs to figure out a better way to operate in light of new AI-driven economic realities.

It's a simple question of whether internet bots are causing economic inefficiencies and mis-allocation of resources in the classical sense.

The bad bots, discussed in the article, are. AI scraping bots are not. All it takes to see this are all the mentions of fraud in this HN thread. Bad bots are, at their core, fundamentally fraudulent in how they interact with sites. That fraud is compounded into how companies realizing the extent of that fraud react: by hiding or lying to investors and boards, because the fraudulent bot activity has significantly altered market perception. Nothing about pure scrapers is fraudulent, regardless of whether their ultimate goals are good or bad for the economy or humanity, if that could even be known.

Re: Bots are getting good at mimicking engagement

#273

Post by lazide » Thu, Oct 16, 2025, 12:45 AM UTC

philipwhiuk wrote: post #80

lazide wrote: post #21

Earlier quoted context omitted.

This is the ‘gold’ that agentic/generative AI really produces. For most consumers, it’s entertainment, but for industrial use it’s great for this kind of fraud. And very difficult to detect at scale, since the only cost effective tools for this kind of analysis are also ML/AI, and hence can be fooled more predictably/trained against.

> This is the ‘gold’ that agentic/generative AI really produces. And here's me thinking it was producing a tonne more data scraping.

That’s the coal for the steam engine

Re: Bots are getting good at mimicking engagement

#274

Post by Terr_ » Thu, Oct 16, 2025, 1:05 AM UTC

Terr_ wrote: post #203

simul007 wrote: post #2

Hi HN. I run a marketing agency and fell down this rabbit hole after a client's analytics made no sense (50k visitors, 47 sales). I ended up building a simple script to track user behavior and analyzed 200+ small e-commerce sites. The average was 73% bot traffic that standard analytics counts as real. The bots are getting creepily good at mimicking engagement. I wrote up my findings, including some of the bizarre pat…

> It seems like a massive, open secret that nobody wants to talk about because the whole system is propped up by it. I was never really a punk-rebel kid, but a certain part of me rooted in the optimism of the early-internet kinda wants to see ad-models crash and burn. Even advertising "working normally" always had a psychic odor of exploitation and deceit. Ex: "You absolutely need this product or else your peers will…

P.S.: Imagine a world with "AI" (open-source, user-controlled) which can instantly detect ads on your screen, and pass the bounding-boxes to another program which covers them up.

You could even install them on some digital glasses for a kind of They Live sunglasses situation [0], but please never wear such things while driving...

[0] https://www.youtube.com/watch?v=wqrrh86WDG0

Re: Bots are getting good at mimicking engagement

#275

Post by renewiltord » Thu, Oct 16, 2025, 1:06 AM UTC

VBprogrammer wrote: post #265

ww520 wrote: post #223

Earlier quoted context omitted.

Way back when I work in an ad based company, click fraud handling was under my overseeing. We caught about 20 percent of clicks as fraudulent and filtered them out before billing the ad placing vendors. It was a constant battle with the sales team to relax the rules, as any clicks filtered out cut into the sales revenue. Sometimes we got the customers on our side as they ran their own analysis on the billed click rep…

Yeah, the incentives there are obviously misaligned. I wonder if there is a potential way of making advertising click through tracking following the "I cut the cake, you chose the slice" model. Some countries have property taxes where you declare the value and the government retains the right to purchase the property for that value for example. My first thought was to make the advertising cost driven by revenue on th…

People will just pull ads if the ROAS isn't there. Performance marketing teams aren't fools.

Altering data would mess with everything. Why is unverified traffic increasing? What's wrong with new marketing efforts? Marketing just requires fixed definitions. e.g. if you have 97% bots but it remains constant that's okay. I know I am spending $x to get $y conversions. I can plan over time, increase or decrease and I can plan. I won't be willing to pay as much as with 0% bots (will pay far far less) but I can set my strategy on this.

It's not that it's x% bots that is the problem. Growth team doesn't adjust strategy on percentage-bot. Growth team adjusts strategy based on return on ad spend. If 0% bots but no return, way worse than 5x ROAS with 99% bots.

Re: Bots are getting good at mimicking engagement

#276

Post by dietr1ch » Thu, Oct 16, 2025, 1:14 AM UTC

mapt wrote: post #189

seviu wrote: post #68

Earlier quoted context omitted.

I once worked for the yellow pages in Switzerland. Our paid clients had a dashboard which reported how many users visited their business entry. We at engineering decided to filter out bots. Figures fell dramatically by more than 50%. In less that a day business mandated us to remove the filter. Bots are real people after all

This is essentially fraud. Your company was made aware that it was selling a product with wildly different characteristics than advertised, and chose to cover it up. There are defensible business reasons for this, in having a contract already in place at the old CPM, so being unable to double the CPM and half the views mid-contract... but still pretty much fraud.

Sounds like even more fraud. Here you can apply the new filter to historic data or create a new metric version and keep the old version to avoid breaking continuity in what's measured.

Re: Bots are getting good at mimicking engagement

#277

Post by dwd » Thu, Oct 16, 2025, 1:24 AM UTC

dpoloncsak wrote: post #182

simul007 wrote: post #2

Hi HN. I run a marketing agency and fell down this rabbit hole after a client's analytics made no sense (50k visitors, 47 sales). I ended up building a simple script to track user behavior and analyzed 200+ small e-commerce sites. The average was 73% bot traffic that standard analytics counts as real. The bots are getting creepily good at mimicking engagement. I wrote up my findings, including some of the bizarre pat…

I work in e-commerce and completely believe this could be true. The 'Cart Abandonment Bot' is actually something we just had a meeting on today, trying to figure out what's going on here. With that said, what do you think the error is here? Could your script have had some false positives? Enough to move the dial?

I would also be looking for card washing and see if they attempted to purchase and the card failed.

Repeated purchase attempts, particularly on the cheapest product in the catalogue is a bad sign.

Re: Bots are getting good at mimicking engagement

#278

Post by x0x0 » Thu, Oct 16, 2025, 3:49 AM UTC

VBprogrammer wrote: post #265

ww520 wrote: post #223

Earlier quoted context omitted.

Way back when I work in an ad based company, click fraud handling was under my overseeing. We caught about 20 percent of clicks as fraudulent and filtered them out before billing the ad placing vendors. It was a constant battle with the sales team to relax the rules, as any clicks filtered out cut into the sales revenue. Sometimes we got the customers on our side as they ran their own analysis on the billed click rep…

Yeah, the incentives there are obviously misaligned. I wonder if there is a potential way of making advertising click through tracking following the "I cut the cake, you chose the slice" model. Some countries have property taxes where you declare the value and the government retains the right to purchase the property for that value for example. My first thought was to make the advertising cost driven by revenue on th…

It's not that misaligned.

Basic ad ops has ad buyers buy ads from different vendors, track which converts (attribution, which has flaws, but generally is decent signal), and allocate spend via return on ad spend. So it hurts the vendor at least as much as the buyer by inflating the cost per action / damaging roas.

Re: Bots are getting good at mimicking engagement

#279

Post by mlrtime » Thu, Oct 16, 2025, 4:39 AM UTC

arccy wrote: post #269

LeifCarrotson wrote: post #224

Earlier quoted context omitted.

That's splitting hairs, but in a way that's important to the conversation. If YouTube served gigabytes of the video file for 40 munutes and human watched it for that time, but they didn't send a request to `youtube.com/api/stats/atr/` and periodically to `/stats/qoe?`, did the video actually get viewed? I think a reasonable person would say that the person viewed that video. Only a programmer would suggest that wasn'…

if a bot (or whatever latest of download tool) was served gigabytes of video, does that mean a human eventually viewed it? no.

So now we have to pass the touring test before the video is served? :)

Re: Bots are getting good at mimicking engagement

#280

Post by vmaurin » Thu, Oct 16, 2025, 5:16 AM UTC

jklinger410 wrote: post #95

vmaurin wrote: post #69

I did work in the ad tech industry for almost 15y and big corp like Google/FB scam their user: - they don't allow double tracking, so you have to trust their numbers - if you look at IP from their "clicks", you see often a FB/Google datacenter IP range - and for most of the traffic they might send you, they did just clever algorithm and heavy profiling to stole your organic traffic. So they get this "amazing" perform…

> they don't allow double tracking, so you have to trust their numbers Facebook Ads and SA360 both allow 3rd party impression and click tracking. Not to mention the myriad of 3rd party analytics tools you can use to track the website. > if you look at IP from their "clicks", you see often a FB/Google datacenter IP range I've never heard this before and seems hilariously simplistic. These megacorps don't have VPNs? >…

I left the industry 4y ago, but at the time, it was impossible to track impressions on FB ads. So when FB tells you "We did display 1000 time your ads", you just have to believe them. For the IP of the click, I have seen it from my own eyes in 2013, from Facebook. About analytic tools, I let you check who is the leader in this market, and about their independence in the click market ... If you have a website that pointed by a FB ad campaign, just look at the IP of the incoming clicks.

> Not sure if you had a stroke while writing this, but this makes no sense.

Well, that what is happening in practice. It is an attribution game, ad companies try to have the best conversion numbers more than increasing the overall sales of a store.

I have been working in a drive to (physical) store company, we were measuring performances with an expose group + a control group and we were showing 3% visit in store increase during a campaign, while our competitors were claiming things around 10%. We knew that their tracking technology was just about "we shown an ad, they visited the store, it is because of us"