Yep, this is a thing. I manage a similar product. It determines if ads should appear in the funnel based on the expected likelihood and value of conversion the expected likelihood and value of an ad click. I’d be surprised if google isn’t doing something similar.
Do you find that this kind of optimization hurts retention? I'd expect that a decrease in retention rates would decrease long-term revenue - but that an A/B test would show a revenue increase as measuring retention will never hit statistical significance relative to revenue on a 2 week test.
It's a good question, and definitely a dynamic we see sometimes in other parts of the marketplace - all though that's usually more driven by market dynamics than consumer behavior (e.g. publishers over-optimizing something that only works until the advertisers update their bidding strategy).
In this specific case our clients are generally insurance companies, and the traffic in their funnel is not made up of current customers. The point of the predictive model is to identify people who aren't going to convert, so retention isn't relevant. Passing those consumers on to competitors who are a better fit helps the carrier recoup money, and helps the consumer find a plan they're more likely to buy.
In terms of re-shopping down the road (customer lifetime could average anywhere from 3 months to 5 years from now, depending on the carrier), people generally don't visit insurance websites enough to build any strong opinion about what the UX was like - so as long as it's not spammy like a travel site, it's a fairly fresh game every time they shop for a policy.
Search does show non ad links for me, however I find other searches I make on Google pretty much shows me a list of ads with a couple of non ad hits at the bottom. This is the reason I switched to DuckDuckGo a few months back.
I find myself unsurprised, primarily because so much cancer "fundraising" et al feels very scammy, even (if not especially) the ones that are widely thought to be "legit?"
I think the variability of the results everyone is reporting is as damning as the original report of all ads.
Surely there is not variable truth about the topic.
It also probably exposes how G no longer needs to rely on your cooperation being logged in and not having changed the default enabled personalized results setting.
Unironically, the worst nightmare I had recently was that Kagi went bankrupt. I can’t imagine going back to a search engine that treats its users as poorly as Google.
I have actually been using ddg for a long time but lately I have some things that I know exist which I access routinely and I just use search to get there for convenience because the site (a wiki) does not have the short/pretty url option enabled so the urls are all inconvenient.
When I search for it lately (several months at least), by it's exact, correct, unique name, ddg gives me everything in the world BUT the actual site. It gives me pages and pages of links to a mail list archive, admittedly from the same domain, and other random links that are either simply unrelated and wrong, or some that include mentions so they are relevant but should appear far down the list.
It's ridiculous.
Google produces links to the site top front center, as well as the indirect mentions further down.
It's a small community so I'm afraid to slashdot the guy but really a story like this is kind of dry without the real example, and maybe an ordinary HN entry is not so bad.
Search "bitchin100 rex" or "bitchin100 tpdd" or even just the plain exact unadulterated domain "bitchin100.com" in google, then the same thing in ddg. I don't even need to provide the correct results here for reference. the google ones will be correct or at least reasonable, as in on the first screen if not the very first non-ad result.
I'd be curious if using ddg anyone else in gets anything but a never ending scroll of Narkive links. To be fair, the mail list is also from the bitchin100.com domain, so those results are not wrong in that they should be there also, but the better-matching results are not only not first, they're not even there at all no matter how many "more" you load.
Can someone seeing ads post screenshots? Maybe to imgur? I am not running an ad blocker, and I don't see any ads, just what look like relevant results.
I find it interesting that people still haven't learned when trying to post "proof" of Google's underhanded ways. Google could be changing the ads returned because it's detecting that we are hitting it at unnatural frequency. Who knows this black box anymore?
Post screenshots! Literally no one will have the same search experience as you as the dozens of replies here indicate. It's /impossible/ to replicate whatever anyone says about their search experience anymore, so the best solution for presenting it is to screenshot it
(Literally says the user who has posted no screenshots.)
Unlike you, I did post screenshots and pointed out the issue of it being impossible to reproduce/share search results to Danny Sullivan (Google’s Public Liaison of Search) on HN few weeks ago:
PS: I don’t believe Google is being “under handed” - they are a massive company with billion of users running a complex system. Unless there’s proof they’re intentionally being malicious, I would give them the benefit of doubt, since attacking them with baseless accusations won’t help anyone.
I don’t understand the purpose of your comment? The burden of proof is on people who claim something. What’s your problem? If someone says the page is full of ads and I can’t replicate it, why should I post a screenshot? Please, make sense.
Sharing and comparing screenshots is pointless if Google actually has built-in support which was URL based. It would also allow for more than simply reporting bugs; for example, comparing two different, but equivalent searches for the same topic.
I'm flagging this, not because I think the topic necessarily isn't appropriate for HN, but because the link isn't. It's well known that Google results vary from person to person, so the link won't accurately demonstrate whatever the issue is for most people. Document it with screenshots or a link archiving service and post a link to those instead.
First I got 4 consecutive ads as result, people started asking for proof. I refreshed the page to take a screenshot it showed me just one ad. I refreshed it again and its not showing me any ad.
Whats the sorcery and how do I convince HN that I am not trolling anyone here.
I tried Chrome, Firefox and Edge with default and incognito/private/inprivate sessions. My Chrome default is logged into a google account. I get varying results. None that I see match the original post ("only returns ads") but I see from 0-8 ads on the first page.
All three incognito result sets are the same: first "Scholarly articles", then 9 quality organic search results, then a "People also ask" section, one more search result, then three marked "Ad" advertising for medical conferences, and finally "Related searches" at the bottom of page 1.
The signed in Chrome has the same except omitting the "People also ask" and the three ads.
The default Firefox has 5 additional ads at the top of the page but is otherwise the same as the incognito results.
The default Edge has 4 ads at the top of the page and no ads at the bottom, with the other parts being the same as the incognito results.
Pointed out the issue of it being impossible to reproduce/share search results to Danny Sullivan (Google’s Public Liaison of Search) on HN few weeks ago, specifically from my prior comment:
>> Conclusion: To me, if you cannot share a search result and get the EXACT same search results (and possibly ability to see different ones AND annotations of why they are different) — how this not a bug; how can anyone independently test Google’s search quality? [1]
Which was result of me trying to document a few Google search bugs — only to have Google repeatedly and randomly giving me different search results by just reloading the page.
Hopefully Google will finally listen and make it easy to reproduce, share, and compare search results.
45 comments
[ 1.9 ms ] story [ 92.7 ms ] threadThanks, fancy Google statisticians, quantitative analysts, data scientists, or whatever you're called these days.
In this specific case our clients are generally insurance companies, and the traffic in their funnel is not made up of current customers. The point of the predictive model is to identify people who aren't going to convert, so retention isn't relevant. Passing those consumers on to competitors who are a better fit helps the carrier recoup money, and helps the consumer find a plan they're more likely to buy.
In terms of re-shopping down the road (customer lifetime could average anywhere from 3 months to 5 years from now, depending on the carrier), people generally don't visit insurance websites enough to build any strong opinion about what the UX was like - so as long as it's not spammy like a travel site, it's a fairly fresh game every time they shop for a policy.
Edit: seems to be intermittent.
edit: have setup a VPN + Pihole dns server + ad blockers on devices themselves
"Scholarly articles for cancer genetics conferences" with several links to Google Scholar.
City of Hope Clinical Cancer Genetics | Annual Conference
CGC Annual Meeting 2022 - Cancer Genomics Consortium
Cancer Genetics and Epigenetics - Gordon Research Conferences
A few more links to the "City of Hope" conference
Genetics in Clinical Cancer Care - Controversies, Insights & Scalable ... (Uchicago.edu)
EACR Conference on Cancer Genomics
Conferences - Ambry Genetics
Oh wait, I get a single (possibly relevant?) ad at the very, very bottom - 2022 Cancer Conference - Attend the 40th Annual CFS®
Surely there is not variable truth about the topic.
It also probably exposes how G no longer needs to rely on your cooperation being logged in and not having changed the default enabled personalized results setting.
I have actually been using ddg for a long time but lately I have some things that I know exist which I access routinely and I just use search to get there for convenience because the site (a wiki) does not have the short/pretty url option enabled so the urls are all inconvenient.
When I search for it lately (several months at least), by it's exact, correct, unique name, ddg gives me everything in the world BUT the actual site. It gives me pages and pages of links to a mail list archive, admittedly from the same domain, and other random links that are either simply unrelated and wrong, or some that include mentions so they are relevant but should appear far down the list.
It's ridiculous.
Google produces links to the site top front center, as well as the indirect mentions further down.
It's a small community so I'm afraid to slashdot the guy but really a story like this is kind of dry without the real example, and maybe an ordinary HN entry is not so bad.
Search "bitchin100 rex" or "bitchin100 tpdd" or even just the plain exact unadulterated domain "bitchin100.com" in google, then the same thing in ddg. I don't even need to provide the correct results here for reference. the google ones will be correct or at least reasonable, as in on the first screen if not the very first non-ad result.
I'd be curious if using ddg anyone else in gets anything but a never ending scroll of Narkive links. To be fair, the mail list is also from the bitchin100.com domain, so those results are not wrong in that they should be there also, but the better-matching results are not only not first, they're not even there at all no matter how many "more" you load.
Could it be the word bitch in the name?
F-ing Microsoft, it's probably that.
Be better HN, don't just upvote a post that matches your view of google.
Post screenshots! Literally no one will have the same search experience as you as the dozens of replies here indicate. It's /impossible/ to replicate whatever anyone says about their search experience anymore, so the best solution for presenting it is to screenshot it
Unlike you, I did post screenshots and pointed out the issue of it being impossible to reproduce/share search results to Danny Sullivan (Google’s Public Liaison of Search) on HN few weeks ago:
https://news.ycombinator.com/item?id=32534048
PS: I don’t believe Google is being “under handed” - they are a massive company with billion of users running a complex system. Unless there’s proof they’re intentionally being malicious, I would give them the benefit of doubt, since attacking them with baseless accusations won’t help anyone.
Whats the sorcery and how do I convince HN that I am not trolling anyone here.
All three incognito result sets are the same: first "Scholarly articles", then 9 quality organic search results, then a "People also ask" section, one more search result, then three marked "Ad" advertising for medical conferences, and finally "Related searches" at the bottom of page 1.
The signed in Chrome has the same except omitting the "People also ask" and the three ads.
The default Firefox has 5 additional ads at the top of the page but is otherwise the same as the incognito results.
The default Edge has 4 ads at the top of the page and no ads at the bottom, with the other parts being the same as the incognito results.
>> Conclusion: To me, if you cannot share a search result and get the EXACT same search results (and possibly ability to see different ones AND annotations of why they are different) — how this not a bug; how can anyone independently test Google’s search quality? [1]
Which was result of me trying to document a few Google search bugs — only to have Google repeatedly and randomly giving me different search results by just reloading the page.
Hopefully Google will finally listen and make it easy to reproduce, share, and compare search results.
__________________________
[1] Related prior HN comment:
https://news.ycombinator.com/item?id=32354078