Why Robots Will Never Completely Replace Humans: A Deep-Dive Into the Work Of an Ad Quality Team

At AppLift, we like to pass on the mic over to various employees in the company so they can provide insights about their specific fields of expertise. A few months ago, AppLift acquired mobile DSP Bidstalk and we fully integrated their team based in Singapore and India. Today, we turn to Kanika Upadhay, Head of Ad Quality, who gives her view on why robots will never completely replace humans for ad operations in general, and ad quality control in particular.

When advertising became digital, like any up-and-coming industry, everyone scrambled to get the biggest chunk of the advertisers’ budgets. Amidst this race of CPMs,CPCs, CPIs, CPOs (cost per order), CPAs (cost per action) and other pricing models, everyone forgot about the most important part of the equation: the user. This reckless oversight caused a lot of troubles and digital users began disliking the concept of ads to a great extent. For instance, a user who was happy to get an ad about his favorite fast food chain was equally unhappy to get a lingerie ad because he knew his 6 year-old daughter was often playing with her device. Did it matter to anyone else in the industry? No.
Should it matter? Yes, a lot.

And eventually, it did.

There are, off the top of my mind, at least 20 to 30 apps that lost their high rankings on the app stores because their users complained about inappropriate ads. On top of this, needless to say, those new users also shied away from clicking on these ads, in spite of the incentives and offers displayed on the banners.

With evolving technologies and with ever more efficient algorithms, effective targeting became reality paved the way for sophisticated tools to define and assess user preferences. This advanced mapping system also strengthened the span of ad operations and Ad Quality functions within ad networks, DSPs and Ad Exchanges.

In the past 6 years of working within the ad-tech ecosystem, I have seen and been a part of the growing importance of the ad operations and ad quality function. From a simple reactive support function, Ad Ops has become a true sales and business team, defining parameters for ad rendering, understanding specificities of cultures as well as ethnic regional nuances and, most importantly, championing optimization.

What are the top 3 things that an Ad Quality team does ?

Here’s a peek into a day’s work of the Ad Quality team. Our team manually checks and approves every single ad served on our platform.

1. Basic Quality Check

Manually reviewing each demand and supply piece on parameters such as rendering, behavior, ad content, and quality.
With new ad formats such as native ads, very high user interactive ads (e.g. ad carousel), and direct transaction ads for e-commerce domains, the review process has improved considerably with the use of proxies and third-party tools.

2. Cultural Fit Check

Understanding region, cultural and ethnicity specific guidelines to enable correct geographical and demographic targeting.
Mobile ad tech is one of the few industries which has impacted all global markets at almost the same pace. Therefore, from economic giants like the US and European nations to small countries like Sri Lanka and Indonesia, each market has its own complexities and needs to be understood perfectly.

3. Context Check

Optimizing every ad by ensuring they are run on best inventory to generate the best yield.
This involves great in-depth knowledge of available sources, best converting supply, etc. This analytical part gives an edge to any campaign and ensures that advertisers’ goals, such as awareness (branding), acquisition (performance), or transaction (post-install performance) are fulfilled at the best cost possible.

Here are a few examples of ads that our team recently approved and declined, as well as the reasons for doing so:


  • Give clear insights as to the product advertised
  • Mentions the brand name
  • Has no ambiguity or user attention-diverting gimmicks such as flash or blink

Not Cool:

  • Mimics the Facebook design and color patterns

  • Plain fraudulent message

  • Ambiguous message
  • No brand shown
  • Obviously not suited for many publishers/countries/cultures

Will automation completely take over adops in the future?

It goes without saying that a single super fast algorithm, an image recognition program (hello Google), and an automatic script will always be faster than a team of 15 people who manually review each ad piece. They cost less in the long run and maintain great SLAs (service level agreement, which is the minimum time by which an ad is approved and starts to be served). Eventually, automation could greatly benefit ad tech firms by cutting costs and becoming more effective.

However, I strongly believe that ad operations work best by combining the manual review process with the processing power of automated tools. Robots are faster, but only humans are capable of correctly vetting 100% of ads submitted. Ad Quality is unfortunately a field where the 80/20 rule is not applicable; even if robots managed to assess the quality of 99.9% of the ads, the remaining 0.1% false positives could be very detrimental to brand advertisers or to publishers. Therefore, by capturing all the various aspects of ad behaviors and ad content, this collaboration between humans and robots can make the process faster, more accurate and, most importantly, bullet-proof.

In the end, automating the vetting of ad units as well as the selection of publishers can enhance the ad operations’ function but cannot entirely replace it, as teams of 5 to 10 – sometimes more in some bigger ad firms – individuals will always be experts at looking at every ad piece with a keen eye and capture all parts within those banners, videos, and other formats.

Do you have any ad quality horror story? What are your expectations of an ad operations team?