Review of personal concerns in attempting to precisely utilize website ad serving systems, and conceptions for business policy and process changes within such systems. August 2025.
Christopher Clayton
08/22/2025
Because I have been advertising this website on Internet ad campaign systems (ad serving and bidding systems) since August 2020, I wanted to share concerns I've consistently had and how I would tweak existing systems in terms of how I would set up a similar business.
Background regarding personal concerns in precisely targeting Internet ad campaigns to specific audiences by topic via ad serve systems
Regardless of the subjective quality of any given Internet ad campaign served via a given system, the most frustration I've had is attempting to restrict what websites acting as bidders to serve ads could actually bid on this website's ad campaigns at all (aescreta.com). I've wanted to attract people seeking aid on business projects, supply chain admin work and research writing tasks, so I've always attempted to focus on 'business planning,' 'supply chain' and similar as keywords; occasionally others as well in the vein of 'MS Excel' or 'enterprise resource planning software.'
Before even attempting keyword targeting, I let my choice of ad system loose on 'recommended settings' for bidder matching while still setting a moderately affordable budget for myself relative to my never expecting to make actual money off this website. Of course I received a relatively absolute high number of clicks, but that doesn't mean it was useful having a campaign for this type of website running on relatively popular YouTube entertainment channels and the like.
However, once I did explore keyword targeting, the issue since then has consistently been my finding that the content of the majority of websites serving my ad campaigns have barely or not at all matched the keywords I'm attempting to target. This can also still include entertainment-oriented YouTube channels running the ad; decidedly not 'business'-related except in how the channel may operate as a business. I don't target the singular keyword 'business' so I can only conclude that this extremely loose conception of a YouTube channel acting as a business may result in some entertainment channels being considered by the ad system as in line with 'business operations' as a keyword.
Admittedly, my keywords are nevertheless rather broad but the majority of sites that I can see bidding on my ad campaigns tend to be blogs with few articles, and nominally business-like articles at most. Even then, the business articles are in the bent of 'how to use this software' types of simple guides, if even that much.
Eventually, probably due to my lowering the cost per click budget, the pattern started to turn into mostly Indonesian-language blogs bidding to serve my ads. Again, with the same sort of (in my opinion) minimally business topic-like articles at most as part of the blogs' content; very little to nothing that I could see about business planning or supply chain and not to any depth (such as citing sources to back up conclusions or to demonstrate where the knowledge came from or what the opinions are based on).
Largely based on these types of topic mismatches, I ended up eventually restricting Indonesia and Malaysia as countries from bidding on my ads given that they have been the largest 'offenders,' if you will, where I decided to do this even in spite of having my own Indonesian-language content on this website. Of course, because websites associated with certain countries or regions are not necessarily hosted in those countries, ad systems might categorize websites by 'country' based on the language used in the majority. This should nevertheless result in being able to apply region restrictions of a sort. Even after applying these blacklists, however, some Indonesian-language blogs occasionally are able to bid on and serve my ads anyway.
I'd rather like to support websites in those languages because of my Asian studies background and in wanting to see sustainable world economic growth generally. However, I can't do that when I find such content to be consistently thin by my subjective standards (as a reader) in terms of the topics presented; that is, a lack of insightful or unique posts in that no significant data, concepts or possible entertainment value is presented (again, lots of software recommendation-style blogs or other content I'd call repetitive and largely in less than 1,000 words each, if not that it is dry information without any sources cited and thus that much of it may have been copied from elsewhere but adjusted just enough to not be called plagiarism, but where the grammar at least meets algorithmic heuristic standards). 'Possible entertainment value' to me might entail at least an attempt at fiction, travel posting or writing about personal experiences, but that is not the case from the ones bidding on my ads and that is not the category I want bidding to run my ad anyway, even though that would at least be a step up in content. Because of this, I can't believe that any organic viewership could possibly be taking place to those particular sites, which to me makes the high numbers of clicks on my ad compared to total impressions from some of them to be improbably legitimate (suspicious appearance of 'click fraud' as I describe in the next section).
Otherwise, I fully understand why Indonesians in particular might see blogging (or making minimal frameworks of blogs) as a way to make additional money, and how this may be related to Indonesia's growing eCommerce sector. However, regardless, it overall doesn't explain why my ad campaign (entailing a website that is vast majority in English) is even able to be bid on amongst Indonesian-language websites, let alone as often as I've seen it happen. I don't see websites comprising any other non-English languages bidding on ad campaigns I've run, except very occasionally my current ad has run on Hebrew-language software apps.
I don't see why websites (or YouTube channels) that rank relatively low (or not at all) regarding specific keywords on search engines, with a probable resulting disproportionate number of direct visits (bots, etc.) versus organic visits due to their in no way having any sign of having relatively large viewership, could possibly be allowed to serve ads related to the keywords I have set up. Part of it may be the relatively low-cost bids they are willing to make compared to other websites, but it still doesn't seem fair or sensible if a given website bidding on serving ads has little or nothing in actuality to do with a keyword that an ad campaign is targeting, and where search engine rankings should bear that out. It does not seem to me that current ad serving systems necessarily tie keyword relevance to actual search engine rankings when allowing a website to bid on ads associated with certain keywords, or where the ad serving system may be making broad algorithmic categorization allowances in spite of search engine data regarding actual visits to or impressions for that website regarding given keywords.
Conceptions for systemically dealing with click fraud if I were running an ad serving network business
Besides the difficulty in getting an ad to truly become served in the majority by websites matching only certain types of content categories to attempt attracting certain audiences to click on the ad, the other persistent but ultimately related problem I still experience is when a campaign on an ad serving system reports a serving website source as having yielded a relatively high number of clicks compared to impressions. E.g., one click and one impression or even multiple clicks versus one impression; or perhaps a high proportion of clicks within a few hours compared to total impressions (i.e. beyond single-digit click percentages via that one source). That is, activity which suggests the serving website is a click farm in practice; a website acting as an ad bidder where friends, family or outside hire keep deliberately refreshing the website and clicking on served ads to artificially generate ad revenue for the site but where the served ad campaigns paying for the clicks have no chance to benefit in receiving viewers who may have actual interest in viewing and/or giving business to the websites running those campaigns.
All of this finally led to me thinking about ways in which I'd deal with 'click fraud' of this kind if I were creating ad serving network policies, at least in terms of a fair way to detect such fraud while allowing for a fair cross-section of websites to become ad bidders.
'Quality indices' at an algorithmic level for admitting specific websites into a given ad serving network as a valid bidder are understandably broad because the exact content of a given website itself is not necessarily subjectively judged by conscious readers as part of the admissions process. I don't necessarily agree that this means algorithmically looking for the professional qualifications of an author should be a significant quality weight in regards to the content, or at least not for all forms of content, but it makes sense that ensuring articles on a given site actually are comprised of complete sentences that are in theory consciously understandable and that they are not copies or outright plagiarisms of content established previously on other websites can go into such an index's quality factors. This is exactly one area perfectly suited for large language model heuristics within a search engine indexing context (comparison of the target website to existing indexed Internet content, and comparing the website to a database of language heuristics such as English words and grammar patterns). Still further would be tests to determine if any content itself was generated in whole or in part by large language models trawling existing data amalgamations.
However, as I've noted, this doesn't mean the content might not be 'thin' by a subjective reader's standards. The algorithmic way around this that I can see is to then measure candidate websites for participation in an ad serving system by their organic views versus direct views, with mitigating factors based on previous organic search engine performance to take into account websites that have transitioned into a state of receiving more direct (repeat or word of mouth) viewers. Search engines already rank content based on the effectiveness in how well given websites match users' search term needs, as well as by the number of backlinks to given websites by other websites relative to the weight of the backlink sources' own search rankings.
In practice, this means to me that a website with a majority of direct visits (including bots and probable friends/family members/outside hires) in proportion to organic visits from search engines and backlinks, and relative to previous organic visitor rates, should not be allowed to serve ads on a given ad serving system in my conception of such a system. Some other proportionality standard can be used, but I suggest a simple majority of direct visits versus organic visits as the starting basis. This also prevents needing to attempt tracking IP addresses making rapid repeat visits, as well as utilizing methods to attempt determining the probability of an IP address changer being utilized as part of such activity. However, this may be necessary to some extent to attempt gauging truly unique direct views (those outside of IP address swapping) versus unique organic views.
I don't see how a proportionality standard regarding types of visits would be discriminatory towards up-and-coming blogs or other relatively new websites in terms of their ability to participate in ad serve systems because how are ad campaigns supposed to achieve useful ad traffic if the ads are served on relatively low-traffic websites? My site in some ways is a 'blog' of sorts, but with little to no organic traffic and thus also with few or no repeat direct visitors (relative to direct bot traffic). I would not expect it to be useful as an ad serving mechanism in terms of driving viewers to other websites who may be actually interested in those other websites. The only conclusion in all probability regarding a website with proportionately little to no organic traffic relative to direct traffic but where it is serving ads via a given ad serving system anyway (acting as a bidder on ad campaigns) is that such a website is a mechanism for click fraud, especially if it has proportionately few or no backlinks from websites with relatively high organic visitor rates. This is borne out especially if the site is inexplicably generating a high number of clicks on the ads it is serving in proportion to a low number of organic visitors where proportionately high numbers of direct visitors based upon relative site popularity is improbable, but the criteria I'm suggesting is designed to be utilized before even accepting such a site onto the ad serving network in the first place.