by Arthur Hainline | Mar, 2021 | Content, Native Advertising, Product and Technology
While Google’s recent press release on the deprecation of third-party cookies offered little in the way of new information, it forced the ad tech world to face the future with a bit more clarity and perspective. Google dashed any last hopes of a silver bullet with regard to unique identifiers (I now realize there was a small part of me still hoping for a browser based Ad ID), and made it clear that they would not involve themselves in any alternative, consent-based identifier framework (though it also seems that they will not obstruct such efforts). The programmatic advertising world is facing their options having heard Google’s qualifications, but for real this time.
Three core tactics present themselves as Chrome depreciates the third-party cookie.
1. Consent-based, Cookie-less Identifiers and Strategies
One clear path will be to use consent-based, cookie-less identifiers. Many large players in the industry are working towards these identifiers including but not limited to LiveRamp, The Trade Desk, and Lotame. These solutions seek to preserve the framework by which the Lumascape primarily transacts programmatic advertising –identity–while sourcing identity in a privacy compliant manner. While Google clarified that they will not be investing in this type of solution, these new identity products should still work, with some caveats. The size and scale of the addressable web within the scope of these new identity products remains unknown, and the big question is, what portion of the web are advertisers willing to sacrifice in order to maintain the status quo? Quite a lot, I suspect, especially upon consideration of the investments made into DMP’s, first party data, and other countless data strategies over the years.
The typical tactics of audience targeting, re-marketing, and others will continue to work in some capacity, and agencies and brands will continue to be able to communicate those identities to DSPs and down the line for execution. The scale and addressability will all just be some amount worse, as it is now compared to a year ago, and as it was a year ago compared to the year before that.
2. Google’s Proposal: FLoC Buy In
Another alternative is Google’s proposed set of solutions. The Federated Learning of Cohorts (FLoC) adheres to the rigors of privacy restrictions (self-imposed or otherwise) without requiring user consent. It sounds great, but there are a few trade-offs, most of which remain murky. Google claims “that advertisers can expect to see at least 95% of the conversions per dollar spent when compared to cookie-based advertising,” but the methodology with which they came to that conclusion is not transparent. Google additionally proposed that audience creation and measurement can be accomplished using FLoC; these proposals, however, remain incomplete and subject to change. The lack of clear information and plans from Google makes FLoC difficult to plan for and invest in. No one is clear about how well it’s going to work, and how much work it will take to get it to work.
We can at least say that FLoC will be one solution among three that will make a large portion of the web semi-addressable, with “semi” being a flexible and unknown qualifier for some time. Another large portion of the web (everything but Chrome) will remain non-addressable.
3. Just Let All These Post-Cookie Identifiers Go
While staring down the prospects of unique identity and FLoC in the future, one must ask, is it really worth it? Have we strayed too far from the light? Are all these sacrifices and efforts worth hanging on to whatever scraps are left of addressable programmatic media?
Each will have to answer that question for themselves; however, we can say that with a combination of intelligent contextual targeting with aggressive optimization based on non-identity signals, audience targeted programmatic does not have a monopoly on digital performance marketing. In fact, in the grand scheme of targetable audiences that exist in the ecosystem now, very few can justify their price against their performance. Dynamic and algorithmic optimization that goes beyond bid modifiers using non-identity signals reign supreme in performance, and letting go of identity seems like the sustainable bet.
Furthermore, while the addressable web shrinks, the rest of it goes on sale. As buyers concentrate their bidding power on those auctions where the user is known and consented, those auctions where the user is not known will have fewer bidders and ultimately lower media cost.
In the end, the smartest strategy will be one that can make the best use of all the available tools. A programmatic media plan that can work with the addressable web in the new era and understand its limitations while also testing and applying non-identity based programs will be the winner. The best planners and executors will also have a keen understanding of how their different identity and non-identity tactics will interact with their measurement and attribution solutions.
Arthur Hainline is the VP of Product at Bidtellect
by Arthur Hainline | Feb, 2021 | Product and Technology
Bid shading is a campaign capability that can save advertisers money. Here’s what it means and how it works.
First, some background: Second price auctions incentivized bidding at the price of the value. First price auctions incentivize bidding at the lowest price that will win the auction, or the price of the value – whichever is lower. First price auctions now make up the majority of programmatic auctions.
Bid shading reduces the bid price to save money on first price auctions (first price sealed-bid auctions/blind auctions). And because of the scale of programmatic first price auctions, Bidtellect is in a position to have a good sense of what the lowest winning price of a slice of inventory is.
Bidtellect already helps our advertisers understand what the value of any slice of inventory is worth to their optimization goals, but now we can help our advertisers make that second part of the decision: What is the lowest price we can bid and still win?
Below is what a typical Win Rate vs. Bid Price curve looks like. There’s some range of bid prices where Bidtellect won’t do anything because the bid price really impacts the win rate. But take this example: If you value a slice of inventory much higher than $6, you’re really just throwing away inventory because your win rate doesn’t increase substantially. Let’s say you value the slice of inventory at $10 based on your optimization goals. With bid shading turned on, we will take a $10 bid and we’ll reduce it down to $6 and save you that $4 without impacting the outcome of the auction.
Bidtellect has the best optimization to determine price to value for each auction according to our 14 optimization goals and multiple goals
Layering our new bid shading technology will increase performance, save money, and maximize advertiser’s value.
Bidtellect will charge 20% of SAVINGS found by bid shading
If our bid shading did not find savings, you will not be charged
Example: If the bid price before bid shading was $5, and Bidtellect’s bid shading reduced the price to $4, the advertiser will be charged $0.20 as the bid shading fee
by Arthur Hainline | Jul, 2018 | Native Advertising, Product and Technology
It didn’t take long for the use of Artificial Intelligence in marketing and paid media campaigns to catch on as the new shiny buzzword in the ad tech and mar tech industries. Like many of the trends that consume these industries, most were excited about the potential implications – enhanced personalization at scale, real-time adjustments, smarter media buys etc. – followed by an inevitable shadow of doubt once people began to realize that it was a bit more complicated than they first anticipated.
Skepticism around Artificial Intelligence
Before we dig into why artificial intelligence in marketing actually does deliver on its promise, we first need to take a step back. Artificial Intelligence is not the first thing to complicate digital advertising. Programmatic advertising – the automated buying and selling of digital media – introduced incredible complexities. Even though it is predicted that 82.5% of all US digital display ads will be bought via automated channels in 2018, 10 years after it emerged, many people in the industry are still trying to get a solid grasp on it. Nevermind regular consumers.
So it makes sense that when another extremely complex technology, Artificial Intelligence, emerged as the next game-changer in digital advertising, many were excited at its promise, but confused at how it will actually work and then eventually skeptical if it works at all. For the skeptics, I am not going to explain how it is already being leveraged every day (more on that here), but why it really will deliver on its promise of optimization.
Programmatic decisions are made in real-time and are based on the trillions of data points in the digital ecosystem. It is similar to how software developers make use of application intelligence to see all the connection points between the app and database. For marketers, it is really important to extract the maximum amount of value and understanding. However, when it comes to data, it is yet another code that marketers have had a tough time cracking. It is one of a brand’s greatest assets, but also presents the most challenges, ranging from aggregation, organization, and utilization:
- Siloed data across all marketing channels (email, CRM, paid media etc.)
- Analyzing this data and applying insights to make customers feel like each experience is unique and personalized to them
- Using 1st and 3rd party data to scale content while maintaining relevant, personalized experiences for consumers
High consumer expectations for tailored experiences
Content on demand has instilled a need for instant gratification in consumers. Brands want customers to feel like they understand them and what content they require in any given moment. The best way to create this feeling is to tell personalized stories and create tailored experiences for individuals, and then use the best digital marketing strategies to increase the reach of these stories and Grow Your Business as a result.
Before the concept of “real-time” became part of every marketer’s plan, personalization was achieved through batch processing. Data scientists would analyze weeks or months worth of consumer behavior data, derive relevant insights and then activate paid media and marketing campaigns some time later. However, as discussed earlier, consumers have high standards for the content they consume and for the experiences that brands provide them, so data that is weeks or months old, more often than not won’t lead to an optimized experience.
So while there is an element of “learning” that occurred here, it was nowhere near as efficient, effective or impactful as identifying even deeper insights and then using them for real-time media buying decisions.
Personalization made possible by advanced technologies
As human beings, our brains are perhaps the most obvious example of Artificial Intelligence in action. As we go through life, meet more people and experience new things, we are constantly learning and becoming smarter, retaining knowledge that will help us make more insightful decisions in the future.
Now think about this same process, but apply it to the computer that is processing billions of data points per second and automating media campaigns. Consumer actions are expressions of their wants and needs. Might seem insignificant, but every single action taken by a consumer is remembered and made actionable for future decisions.
AI promises to deliver this personalization, but in real-time and with much more relevant data. This is probably why AI has so much value in business and also health analytics. The latter has gained much from machine learning (part of AI), with respect to finding misdiagnoses and clinical decision-making. It is always learning and improving over time, becoming an all-knowing entity. Anyway, AI is driving exponential value out of data, ad tech, and mar technology, extracting insights in real-time and delivering personalized messages.
More than just tailored consumer experiences
Personalized marketing messages at scale is the glaring opportunity that harnessing Artificial Intelligence offers brands: through real time adjustment and learning, personalized customization can happen in real time. But there are a lot of other benefits. The first is greater ROI: marketers can truly make the most out of every dollar by making decisions based on data from the decision before, thus maximizing marketing budget and performance. The second is innovation. Marketers can optimize media in new ways by taking unstructured, unused data and making them relevant to paid media campaigns. The third is ad tech: AI can optimize data within and across paid media channels in real-time, driving a seamless consumer experience.
AI joins programmatic, mar tech, and ad tech efforts into a unified approach that maximizes data and assets to offer a smoother consumer experience and greater ROI. While skepticism still exists, advertisers, marketers, and brands should learn quickly the power of Artificial Intelligence and harness its capabilities, or risk lagging behind.
Bidtellect now offers the most advance Forecasting technology on the market – by harnessing the power AI. Read more: Forecasting 2.0: The Future of Contextual Targeting
Want more from the Bidtellectual? We have a monthly newsletter!
by Arthur Hainline | May, 2018 | Content, Native Advertising, Product and Technology
Shifting Away from User Data
GDPR. Apple’s iOS 11 Intelligent Tracking Prevention. Server-side header bidding. Internet trends point to an uncertain future for data-driven digital ad placement. User data – often the primary (if not only) driver of ad placement – will take a back seat as tighter privacy regulations shift into place. Cookies and users are slipping through the cracks, and marketers, media buyers, and media sellers are now faced with lower match rates on the audience segments they worked so hard to cultivate.
The push to keep user information anonymous is coming from many fronts, and includes government regulation with GDPR (General Data Protection Regulation), large corporations utilizing Apple’s iOS 11 Intelligent Tracking Prevention, and increasing adoption of Server-side header bidding within the ad tech industry. These three major factors alone point to a 2018 (and 2019) with an increased number of auctions on the open market where the user cannot be identified.
How much is this ad placement worth?
In a perfect world, marketers and media buyers would be able to assign an accurate cost value to each impression on the web. Every marketer and media buyer wants to know what an ad placement is truly worth to them. When marketers have at least some information about the user, they can make that decision quite quickly and rationally. The historical success of audience-based targeting and optimization has largely based the crucial decision of determining the cost of an ad placement on who the buyer thinks the user is.
Now, however, we are confronted with a landscape of increasingly available ad placement for sale and increasingly unavailable information on the user. Without user information for an increasing number of impressions, marketers and media buyers must try to assign a value to this ad placement. Does that mean this ad placement is worthless? Absolutely not. We can still direct a user interested in the marketer’s product to see the ad, but we must use an alternative source of information to decide on the value of impressions. Context information becomes crucial to the digital marketer’s toolkit.
What’s the context?
Imagine an impression for sale on the open market with no user information available (which will be increasingly common). We can still make an informed decision on the ad’s worth. In fact, we can determine the location, the time of day, the day of the week, the size and location of the ad on the page, the site it’s on, the page of the site it’s on, even specifics about what content is on that page. For example, we could learn there is an impression available at the bottom of an article titled “Best Valentine’s Day Gifts for Her” dated February 12th. This page’s information tells us enough about the type of user visiting the page to assign a value to this impression and determine its worth to me as a marketer. Through its context, I can deduce enough about the user’s intentions to decide how much I’m willing to pay for an impression.
Contextual Optimization in a Post-GDPR Landscape
Understanding that we have enough information about ad space without user information means we can face the (more private) future of the industry with far less fear. Furthermore, because of the lingering dominance of user-based optimization, much of the unmatched ad space on the open market is effectively “on sale.” Buyers that have relied solely on user-based optimization will be bidding very little or not at all on these unmatched impressions. And now that you have this contextual information, it can of course work in concert with your user-based buying strategies when they are available to you, making your bidding decisions incredibly powerful and precise. In a post-GDPR landscape, contextual optimization within native advertising will be the superior strategy for digital advertisers.
Read more about Contextual Optimization from Bidtellect CEO Lon Otremba
Want more from the Bidtellectual? We have a monthly newsletter!
by Arthur Hainline | Dec, 2017 | Product and Technology
Arthur Hainline, Director of Analytics & Optimization, Bidtellect
Bidtellect’s optimization algorithms are built on machine learning, using the billions of data points running through our platform to create predictive models and algorithms. The technology is designed to simplify the complexity of digital advertising, make smart buying decisions and ultimately drive conversions for advertisers.
Performance will always be a top priority for advertisers when evaluating their digital investments. Trusting that there is automated technology in place to ensure spend is consistently optimizing toward desired performance metrics is really important for advertisers’ confidence in their digital marketing efforts. As with all of our technology innovations and enhancements, we are always evaluating the needs of our clients and enhancing our technology to exceed these expectations. Whether the need for Dynamic Creative digital ads are evident or to reshape the current ones out there, we will always see what we can do that will appeal to and attract customers. It is important that we keep the relevant data on our clients so they are in contact with us and we can reach them at specific times. So we may need to verify a postal address through targeted software which can help with accurate records to aid us and our clients.
Focus on Conversions and Performance
Many of the new capabilities in this latest iteration of our optimization technology have a primary focus on conversion based optimizations and post-click activity. This focus improves scaled performance for marketers across multiple campaign metrics. In addition, we made enhancements to existing algorithms to ensure the smartest buying decisions for advertisers.
Multiple KPI optimization, or the ability to optimize toward one or more KPIs in a given campaign, has proven to be highly valuable for advertisers. However, before moving to multiple KPI optimization, it is necessary to learn a bit about what KPIs are and how they can be useful for an organization. KPIs are useful business metrics that can reflect performance. They can let individuals know what they need to analyze to determine the basis for their OKRs (objective and key results). But what are OKRs? OKR is a goal-setting technique that can enable individuals to drive performance improvement and make changes. KPIs and OKRs are both measurable and can help in assessing team performance. (Note: Those interested to learn more about KPIs and OKRs in detail can lookup kpis vs okrs on the Internet.)
That said, once individuals have leant more about OKRs, especially KPIs, they will understand how the latest advancements like multiple KPI optimization can help. Before, advertisers were able to optimize toward each goal for a percentage of the campaign. For example, 20% of the campaign optimized toward CTR and 80% of the campaign optimized toward conversions. However, with these latest changes, advertisers can now optimize toward multiple goals and in concert throughout the campaign.
In addition, we introduced the ability to enter KPI targets. When the user enters a specific target that is what the campaign will optimize toward, leading to stronger performance. If the target is left blank, the campaign will optimize against the average established by the campaign, striving to continually improve that average.
View through conversions is a new metric to report against and optimize towards with unique optimization algorithms. With this conversion type, the user has the ability to change the view through attribution window, ranging between 1-30 days. The combination of the new flexible multi-goal methodology, the ability to change views through and click through attribution windows, and the ability to enter targeted CPA goals, we can accommodate clients that have different standards for measuring attribution. Business owners can look at agencies such as Samba TV who can help them optimize campaigns with an accurate omniscreen identity and holistic solutions with identity resolution and industry-leading TV data.
Optimizing toward post-click activity is really important for advertisers to understand how consumers engage with their content. We developed a new type of optimization, Quality Optimization, designed to enable advertisers to optimize toward this post-click activity when they are unable to implement code on the content. Quality optimization is a unique way for advertisers with engagement or conversion goals to optimize toward high-quality post-click activity.
Video is an integral piece of marketers’ strategies. Recognizing this growth, we introduced two new video goal types, Play Rate and Cost per Completed View (CPCV), providing marketers with multiple options for their video campaigns, depending on objective and type of content.
With these most recent changes, Bidtellect’s platform now offers 11 unique goal types with 66 possible goal combinations. Every client has their own set of unique goals. Bidtellect recognizes that a major part of client performance is offering unique optimization goals that exactly match their needs.
Additional Enhancements
Additionally, advertisers have four new pricing options to choose from. flat Cost per Play (CPP) and flat Cost Per Completed View (CPCV).
Flexible bidding strategies in line with real-time performance have always been a crucial part of how Bidtellect approaches media buying. To support this, we added two new bid types that enhance our ability to buy ads at a price that reflects their value – Dynamic CPC and Dynamic CPM.
Our objective is always to deliver successful campaigns for advertisers and often times this means considering the end user experience. User-level CTR estimates which monitor an individual’s ad exposure and interaction (clicks, engagements etc.) and adjusts delivery based on these insights to avoid ad fatigue and to not inundate consumers with repetitive messaging.
Importance of a Smooth Pacer
Underlying Bidtellect’s optimization technology is the power of a smooth pacer. A smooth pacer enables advertisers to reach their widest audience by pacing out the spend of budget. A built-in, automated pacer is more effective and efficient than frequency capping. This makes planning and budgeting easier for advertisers because the pacer, backed by a real-time data pipeline, ensures smooth and effective pacing to the make the most of the ad spend.
We are now bidding more frequently when inventory has driven optimal KPIs in addition to modifying the actual bid amount. Our pacing algorithm drives smooth and consistent delivery, considering time of day and volume levels in its predictive algorithm.
Recent Comments