Simply Ad Tech

ChatGPT – What’s all the Buzz?

It seems like everywhere you look these days you’ll see another post about ChatGPT and the worldwide impact that it can have on everything from customer service to programming. Historically, chatbots have been set up to give preprogrammed responses to common customer questions. ChatGPT changes that by replying to queries in a more conversational way, using reinforced learning from human feedback in order to learn and to gradually give more humanlike responses. The implications of this can be quite significant: more automated customer service with potential impacts on online customer service jobs, the ability to generate essays with easy in academia, and writing emails to name a few. So far, ChatGPT and other automated language models haven’t reached a level of accuracy where we can consistently use them in everyday applications. However, the technology will only improve over time. Eventually, we may reach a time where we ChatGPT and similar technologies are used with confidence.

How Does It Work?

From the OpenAI, the founders of ChatGPT –

“We trained this model using Reinforcement Learning from Human Feedback (RLHF), using the same methods as InstructGPT, but with slight differences in the data collection setup. We trained an initial model using supervised fine-tuning: human AI trainers provided conversations in which they played both sides—the user and an AI assistant. We gave the trainers access to model-written suggestions to help them compose their responses. We mixed this new dialogue dataset with the InstructGPT dataset, which we transformed into a dialogue format.

To create a reward model for reinforcement learning, we needed to collect comparison data, which consisted of two or more model responses ranked by quality. To collect this data, we took conversations that AI trainers had with the chatbot. We randomly selected a model-written message, sampled several alternative completions, and had AI trainers rank them. Using these reward models, we can fine-tune the model using Proximal Policy Optimization. We performed several iterations of this process.”

So it looks like currently ChatGPT still requires a good amount of human guidance in the form of AI trainers is required in order to optimize its algorithm. AI trainers utilize demonstration data to train policies, and sample outputs are presented from several different models. The outputs are ranked by performance and the results are used to further train the reward model. A more detailed explanation with illustrations can be seen below.

Photo from https://openai.com/blog/chatgpt/

ChatGPT Implications In Advertising

Writing Call-To-Actions and Copy on creatives
- Some outputs may be brilliant whereas other outputs may leave much to be desired. Nevertheless, ChatGPT can be an avenue for creatives to brainstorm successful campaign advertisements to reach their customers effectively.
Analyzing information for more efficient targeting
- With a complex data set and a given set of parameters, ChatGPT can analyze break it down in many different ways to advise on campaign advertising strategies.
Tracking conversions and important user actions
- Marketers can ask ChatGPT to record a conversion event every time a customer performs a desired action, such as download a pdf file, make a purchase, or sign up for a class.
Customer service
- ChatGPT can act as customer service to field customer questions, especially with companies that struggle with finding the bandwith.
Coding
- This can be a huge time saver for front-end web developers. For example, instead of manually adjusting the code to provide additional options for a new bulk edit option, developers can ask ChatGPT to do it for them.

*All of these possibilities and more come with the huge asterisk that ChatGPT is still in its infancy and currently it is best used as a thought starter instead of an end all be all. Marketers should definitely still check ChatGPT’s outputs and use it as a reference instead of blindly using it without any QA.

Sources

https://openai.com/blog/chatgpt/

I didn’t fail the test. I just found 100 ways to do it wrong.
Benjamin Franklin

Video Game Advertising in 2023

__________ William Chen

Raid Shadow Legends. How many times have we seen this ad?

As a gargantuan industry that generates revenue in the hundreds of billions, advertising in gaming is a very strong tool for any marketer’s arsenal. In particular, mobile gaming is an area with massive potential for marketer’s to reach consumers with both scale and accuracy.

As most users are drawn towards free games due to the frontloaded cost savings, developers and marketers alike have been depending more and more on in-game advertising for free games. Whereas it may be unlikely for a user to buy a new game upfront, they are much more likely to watch an ad in order to receive some sort of in game “incentive” to advance their progress in game. For developers, although this won’t generate as much revenue per user per instance depending on CPMs, it can generate a significant amount of ad views, which a user may view as “free” since they spend time instead of money. For marketers, this opens the floodgates for potential to advertise with enormous scale.

In-Game Advertising Formats

According to Adjust, there are 6 different types of dynamic in-game ads.

Interstitial Ads – Interstitial ads are rich interactive ads covering the entire screen. These ads take advantage of natural pauses in the user’s gaming experience, such as the break between levels. As they don’t interrupt the user experience, interstitial ads generally generate high impressions and conversions. The creative and timing of an interstitial ad are critical to its success.
Native Banners – Native banners appear similar to web banner ads in that they blend seamlessly into the background in which they are placed for an unobtrusive ad experience. Unlike display banners, native ads are embedded and therefore, appear as part of the content. By 2025 native ad spend worldwide is anticipated to reach $400 billion. The increase in ad spend for this ad type is due to continued technical innovation, such as in-feed native ads, which are fully integrated with content.
Contextual ads – Contextual ads are delivered to users based on the user’s state at the time the ad is served. Advertisers share set topics and keywords to their digital media source, which then use algorithms to match the ad with relevant content, keywords, topics, and images. Contextual ads are rising in popularity thanks to the latest privacy regulations that have changed the type of information advertisers can collect. Check out our blog: Contextual advertising’s comeback for further insights on harnessing contextual ads.
Rewarded video ads – As the name suggests, this format of ads rewards users if they watch a full-screen ad. After users watch a 15-30 second video ad, they are given in-game currency, extra lives, more levels, etc. Users tend to prefer rewarded video ads to in-app purchases, and 62% of developers reported increased user retention after introducing rewarded video ads. Ready to get started with rewarded video ads? Our article Understanding rewarded video ads offers 10 best practices when utilizing rewarded ads.
Playable ads – Playable ads are interactive video ads. In this form, users can play a short preview of the gaming app before deciding to download it. This pared-down version offers the app’s essential features so that users can “try before they buy”. By allowing gaming audiences a glimpse of how the app works, playable ads tend to reduce uninstall rates and increase retention.
Advergaming – Technically speaking, advergaming isn’t a single dynamic ad, but rather an entire game built specifically to promote a company or product. In advergaming, ads are embedded throughout the game, all pointing to the singular brand that built the game. For example, it could be a racecar game where users race with an automotive company’s latest model. For more advergaming examples, read Advergaming: Why your brand needs a mobile game.

Fortnite and Jordan Collab – A Real World Example of Brand Advertising in Gaming

As it did with many different brands already, Fortnite announced the continuation of its cooperation with the Jordan brand. Players could compete for the Air Jordan XI Cool Gray, the iconic sneaker model, and discover rooms dedicated to the best players. The MVP would receive access to a virtual museum, a basketball court to test their abilities and an immersive video that advertises Jordan brand. This collaboration would cross-promote both brands by increasing awareness of the already popular Jordan brand and also bringing Jordan fans over to Fortnite.

Photo from epicgames.com

Video Game Advertising in 2023 and Beyond

Historically, mobile advertising has been geared towards attracting users from one mobile game to another. As time goes on and some games reach a significant player base, more and more mobile game ads are now marketing actual brands instead of another game in order to capitalize on the current popularity.
Gaming companies such as Roblox are looking into rolling out gaming advertisements in a virtual reality setting.
Fortnite has invested heavily on branded skins and weapons. More and more gaming companies are likely to follow suit.
Sony and Microsoft are planning on supporting ads in ad-free games on the console, a space that has previously been mostly free from ads.
Video game companies are falling under the scrutiny of the law
- The Federal Trade Commission filed a lawsuit to prevent the Microsoft & Activision Blizzard deal from happening due to antitrust concerns.
- The FTC reached a $520 million dollar settlement with Epic Games over Fortnite’s alleged violation of COPPA and the use of dark patterns.

Sources

Success usually comes to those who are too busy to be looking for it
Henry David Thoreau

Facebook Updates

__________ William Chen

In the world of advertising walled gardens, Facebook is making several changes in the interest of user control and future lawsuit mitigation.

Facebook and Instagram Account Links

Facebook announced that it will no longer link a user’s Facebook and Instagram accounts for advertising purposes without an opt-in, effective immediately. The key idea here is opt-out by default, so the majority of users will automatically be opted out as not many people actually navigate their ad privacy settings. Historically, Facebook allowed users to control access to Facebook and Instagram through the Accounts Center. If a user had signed up for both Facebook and Instagram with the same email address or phone number, Facebook could then have a deterministic match that these 2 accounts had the same user. According to Graham Mudd, Facebook’s VP of product marketing for ads, “the update aligns with trends of offering people more control over how their information is used for ads and is consistent with evolving advertising, privacy and regulatory environments”.

So what will be the impact to advertisers? Advertisers may now see slightly more scale for their media buys, as now a single user may be considered as 2 or more users, depending on how many Facebook owned apps they have. The caveat here is that this will be an artificial increase in scale, since there actually isn’t an increase in users.

Facebook is Updating Audience Forecasts to be Less Specific

Facebook will now provide a range instead of more specific numbers when it comes to audience estimates. Facebook says that “These updates are part of an ongoing effort to improve the way it shares pre-campaign ad estimates and metrics with advertisers and has nothing to do with other ad and measurement-related changes it’s making as a result of iOS 14 and Apple’s AppTrackingTransparency framework”. In the image below, the Facebook platform provides a very wide range, with an Estimated Audience Size of 10M-20M.

A major driver behind this change may be lawsuit mitigation. In 2018 Facebook had a class action lawsuit filed against it for inflating its audience reach numbers. From AdExchanger – “According to one example cited in the suit, Facebook supposedly told advertisers that it had a potential reach of 230 million US adults, out of the 250 million adults in the US as counted by the census in 2018. Problem is, a Pew Research study from that year found that only roughly 68% of US adults in the US – around 170 million people – use Facebook.” With broader audience estimates, Facebook can better defends itself against potential audience estimate claims.

Sources

Tell me and I forget. Teach me and I remember. Involve me and I learn.
Benjamin Franklin

CTV and OTT

____________________ William Chen

MRC Provides Clarity around OTT and CTV Definitions

In order to provide simplicity and to conform to industry standards, MRC has renamed OTT to use the term CTV. According to AdExchanger, “What happened was, people started calling what we called OTT, CTV,” said Ron Pinelli Jr., the MRC’s SVP of digital research and standards. “We’re going to rebrand that as CTV and align that with the industry. [The digital video] had to be displayed to the TV – that was a key piece.”

Breakdown of CTV and Non-CTV from AdExchanger

Over the Top (OTT): Using the internet to stream tv shows, movies, and other content that are traditionally provided by cable providers.

Connected TV (CTV): Any method in which a user accesses OTT on the big screen. CTV is under the umbrella of OTT. (Notice in the graphic above that smart TVs, game consoles, Chromecast, Roku Sticks, and Amazon Fire Sticks all have this capability).

MRC Current Goals for CTV

“The MRC called for more integrations between server-side ad insertion (SSAI) providers – which include companies like Verizon, FreeWheel and Google – and vendors in order to facilitate client-side measurement.”
“And the MRC still hasn’t gotten to the root of the measurement issue – identity. “It’s good to see the paths for ad distribution in OTT fully documented for reference,” one network executive, who declined to be identified, said of the updated definitions. “Unfortunately, the [measurement] issues across the various environments are really tied to the identity solutions – desktop, mobile, CTV – and those have yet to be fully addressed by an MRC framework.””

Combatting Ad Fraud in Modern Times

CTV is still newer and not as developed as other more traditional forms of programmatic advertising, such as display, app, and standard video. Below are some of issues that we are seeing from the “bad actors” of today’s programmatic industry.

Traffic Arbitrage aka Looping. CTV is not as well protected as display when it comes to ad security. Bad actors can have connections in both DSPs to purchase inventory and SSPs to sell inventory. They will then identify cheaper sources of inventory such as standard display banners and then turnaround and sell it as premium CTV inventory to unsuspecting buyers. Advertisers can combat this by making sure to incorporate sellers.json across their ad buys.
Misrepresenting Bundle IDs. Bad actors will also purposely miscategorize bundle ids to include far more premium inventory than they actually do, monetizing at 2-3x the market rate for the sold inventory.
Lack of Clear Video Standards. Bad actors can take advantage of the rather underdeveloped definitions and guidelines of the CTV world. “This makes it possible for fraudsters to misrepresent unsold video inventory according to type (in-stream, interstitial, in-banner, in-feed); device and channel (smartphone, desktop, set-top box, tablet, TV, game console); environment (mobile web, in-app, desktop); and ad break type (pre-roll, mid-roll, post-roll) for CTV. “

An Overview of Major CTV Platforms

Roku

Reached 70 million people in Q1 2021
Added 14 million new users in 2020
Provides over 40,000 free movies and TV shows to users

Crackle Plus

30 million monthly users
Partnering with Plex to sell advertising across the entire platform

Fox + Tubi

30,000 owned and operated titles
140 hours of original programming

Samsung

Reaches 45 million households
84% YoY growth in monthly users
34% YoY increase in time spent on the service

Vizio

12 million monthly users of Vizio SmartCast
Represents 20% of the Smart TV marketplace

Amazon

Reaches 120 million unique people, a 100 million increase since January 2020
IMDB TV, and Amazon subsidiary, has seen a 138% YoY increase in viewership

Sources

There are only two ways to live your life. One is as though nothing is a miracle. The other is as though everything is a miracle.
Albert Einstein

Supply Chain Fee Breakdown and Transparency

__________ William Chen

It’s no secret that advertising in the programmatic marketplace comes with a plethora of fees. In some instances, the transactions may be transparent. In other instances, transaction and their fees may be opaque, or even conducted in the hidden veils of a black box. In addition to opaque transactions, the a significant proportion of the advertising dollar never reaches the publisher at all. The graphic below is provided by the ISBA (Incorporated Society of British Advertisers), when they conducted a study on transparency on the programmatic supply chain, and how much of the advertising dollar actually reaches the publisher.

According to the study, a staggering 49% of the advertising dollar never reaches the hands of the publishers. This means that for every dollar spent, the publisher only receives 51 cents.

Agency Fee

A few different methods for agency fee calculations are below. The agency is meant to consult and conduct advertising on the advertiser’s behalf, which is what the fee is charged for.

● Fixed fee for all services rendered
● Fixed fee for programmatic
● Commission
● Agency and adtech aggregated fee

DSP Fee

Advertisers/Agencies utilize DSPs buying engines in order to buy programmatic media in real time. In the study with ISBA, the actual fee averaged to around 8%. Note that the actual fees and contracted fees sometimes varied. This could be due to make-goods or estimated fees, depending on the contract.

Tech Fees

Tech Fees are mostly demand side. In the ISBA study, this resulted in about 10% of advertiser spend.

Ad Server Fee: Fee to utilize an ad server (generally Google Campaign Manager) in order to serve ads.
Verification Fee: Verification of ad serving and viewability, etc. Vendors in this space include DoubleVerify, WhiteOps, MOAT, Integral Ad Science
Data and Other Tech Fees: Fee to serve data on top of media in order to achieve a targeted message and a higher likelihood of conversion. Vendors in this space include LiveRamp, Oracle, Factual, etc.
Exchange Bidding Fee (EB): Google charges a fee for usage of its exchange bidding product, Google’s answer to header bidding.

SSP Fee

In the ISBA study, we see that the SSP fee was at around 14% of publisher revenue. Similar to the DSP fee, the actual fee and the contracted fee differed slightly due possible reconcilliations.

Unknown Delta

According to the study, a staggering 15% of the advertising dollar is unable to be accounted for. According to the ISBA study, the DSP spend was on average 15% higher than what the SSP recorded as gross revenue. What’s more, even in a “disclosed” model, as much as 33% of the supply-chain costs can still remain undisclosed.

Advertising Spend Breakdown by Media Type and Marketplace Type

In the graphs below, we can see that publisher revenue is least affected in video, but most affected in the Open Marketplace (OMP).

Current Challenges

Lack of formal structure on how different parties share data and where permission needs to be granted. For example, an SSP may need permission from all 3 of the advertiser, agency, and DSP in order to receive campaign data, whereas another SSP may only need permission from the DSP.
Complex Supply Chain. Advertisers may have tens and even hundreds of different supply chains just to reach a few distinct publishers. This could result in bid duplication, and definitely introduces complexity when trying to analyze the results of advertiser campaigns.
Delays in providing access to information.
In-congruent data structure and naming between platforms. For example, in Google Ad Manager the metric “Product” refers to video, display, Mobile-in-app, etc. In other platforms, this “Product” dimension could be named “Media Type” or something else, causing problems with data consolidation between platforms.

Possible Solutions for the Future

Full Transparency throughout the Supply Chain: Some partners may want to protect their data, but a strong and structured transparency format of the supply chain needs to be put in place so that players in the ad industry know exactly where the advertising dollar is going. For “disclosed” transaction models, 100% transparency should be enforced, otherwise it isn’t truly disclosed or transparent.
Cleaner Supply Path: Buyers should double down on investigated their own supply paths, removing partners that may not hold any value. At some point, the infrastructure costs and cannibalization costs of seeing just a little more inventory offset the benefits.
Format Structure for Report Consolidation: The unknown delta of 15% in the ISBA report is staggering. Advertisers need to know exactly where their advertising dollars are going, and not see it disappear inside a black box. DSPs and SSPs need to figure out where exactly that 15% revenue discrepancy is coming from. Ideally, discrepancy checks should be conducted before vendors conduct business in order to avoid issues such as these.

Sources

Before anything else, preparation is the key to success.

— Alexander Graham Bell