Archive for the digital marketing tag

Joni

What is a “neutral algorithm”?

english
What is a “neutral algorithm”?

1. Introduction

Earlier today, I had a brief exchange of tweets with @jonathanstray about algorithms.

It started from his tweet:

Perhaps the biggest technical problem in making fair algorithms is this: if they are designed to learn what humans do, they will.

To which I replied:

Yes, and that’s why learning is not the way to go. “Fair” should not be goal, is inherently subjective. “Objective” is better

Then he wrote:

lots of things that are really important to society are in no way objective, though. Really the only exception is prediction.

And I wrote:

True, but I think algorithms should be as neutral (objective) as possible. They should be decision aids for humans.

And he answered:

what does “neutral” mean though?

After which I decided to write a post about it, since the idea is challenging to explain in 140 characters.

2. Definition

So, what is a neutral algorithm? I would define it like this:

“A neutral algorithm is a decision-making program whose operating principles are minimally inflenced by values or opinions of its creators.” [1]

An example of a neutral algorithm is a standard ad optimization algorithm: it gets to decide whether to show Ad1, Ad2, or Ad3. As opposed to asking from designers or corporate management which ad to display, it makes the decision based on objective measures, such as click-through rate (CTR).

A treatment that all ads (read: content, users) get is fair – they are diffused based on their merits (measured objectively by an unambiguous metric), not based on favoritism of any sort.

3. Foundations

The roots of algorithm neutrality stem from freedom of speech and net neutrality [2]. No outsiders can impose their values and opinions (e.g., censoring politically sensitive content) and interfere with the operating principles of the algorithm. Instead of being influenced by external manipulation, the decision making of the algorithm is as value-free (neutral) as possible. For example, in the case of social media, it chooses to display information which accurately reflects the sentiment and opinions of the people at a particular point in time.

4. Limitations

Now, I grant there are issues with “freedom”, some of which are considerable. For example, 1) for media, CTR-incentives lead to clickbaiting (alternative goal metrics should be considered), 2) for politicians and electorate, facts can be overshadowed by misinformation and short videos taken out of context to give false impression of individuals; and 3) for regular users, harmful misinformation can spread as a consequnce of neutrality (e.g., anti vaccination propaganda).

Another limitation is legislation – illegal content should be kept out by the algorithm. In this sense, the neutral algorithm needs to adhere to a larger institutional and regulatory context, but given that the laws themselves are “fair” this should impose no fundamental threat to the objective of neutral algorithms: free decision-making and, consequently, freedom of speech.

I wrote more about these issues here [3].

5. Conclusion

Inspite of the aforementioned issues, with a neutral algorithm each media/candidate/user has a level playing field. In time, they must learn to use it to argue in a way that merits the diffusion of their message.

The rest is up to humans – educated people respond to smart content, whereas ignorant people respond to and spread non-sense. A neutral algorithm cannot influence this; it can only honestly display what the state of ignorance/sophistication is in a society. A good example is Microsoft’s infamous bot Tay [4], a machine learning experiment turned bad. The alarming thing about the bot is not that “machines are evil”, but that *humans are evil*; the machine merely reflects that. Hence my original point of curbing human evilness by keeping algorithms free of human values as much as possible.

Perhaps in the future an algorithm could figuratively spoken save us from ourselves, but at the moment that act requires conscious effort from us humans. We need to make critical decisions based on our own judgment, instead of outsourcing ethically difficult choices to algorithms. Just as there is separation of church and state, there should be separation of humans and algorithms to the greatest possible extent.

Notes

[1] Initially, I thought about definition that would say “not influenced”, but it is not safe to assume that the subjectivity of its creators
would not in some way be reflected to the algorithm. But “minimal” leads into normative argument that that subjectivity should be mitigated.

[2] Wikipedia (2016): “Net neutrality (…) is the principle that Internet service providers and governments should treat all data on the Internet the same, not discriminating or charging differentially by user, content, site, platform, application, type of attached equipment, or mode of communication.”

[3] Algorithm Neutrality and Bias: How Much Control? <https://www.linkedin.com/pulse/algorithm-neutrality-bias-how-much-control-joni-salminen>

[4] A part of the story is that Tay was trolled heavily and therefore assumed a derogatory way of speech.

Joni

Advertisers actively following “Opportunities” in Google AdWords risk bid wars

english

PPC bidding requires strategic thinking.

Introduction. Wow. I was doing some SEM optimization in Google AdWords while a thought struck me. It is this: Advertisers actively following “Opportunities” in AdWords risk bid wars. Why is that? I’ll explain.

Opportunities or not? The “Opportunities” feature proposes bid increases for given keywords, e.g. Week 1: Advertiser A has current bid b_a and is proposed a marginal cost m_a, so the new bid e_a = b_a+m_a. During the same Week 1: Advertiser B, in response to Advertiser A’s acceptance of bid increase, is recommended to maintain his current impression share by increasing his bid b_b to e_b = b_b+m_b. To maintain the impression share balance, Advertiser A is again in the following optimization period (say the optimization cycle is a week, so next week) proposed yet another marginal increase, et cetera.

If we turn m into a multiplier, then the bid will eventually be b_a = (b_a * m_a)^c, where c is the number of optimization cycles. Let’s say AdWords recommends 15% bid increase at each cycle (e.g., 0.20 -> 0.23$ in the 1st cycle); then after five cycles, the keyword bid has doubled compared to the baseline (illustrated in the picture).

Figure 1   Compounding bid increases

Alluring simplicity. Bidding wars were always a possible scenario in PPC advertising – however, the real issues here is simplicity. The improved “Opportunities” feature gives much better recommendations to advertisers than earlier version, which increases its usage and more easily leads into “lightly made” acceptance of bid increases that Google can show to likely maintain a bidder’s current competitive positioning. From auction psychology we know that bidders have a tendency to overbid when put into competitive pressure, and that’s exactly where Google is putting them.

It’s rational, too. I think that more aggressive bidding can easily take place under the increasing usage of “Opportunities”. Basically, the baselines shift at the end of each optimization cycle. The mutual increase of bids (i.e., bid war) is not only a potential outcome of light-headed bidding, but in fact increasing bids is rational as long as keywords still remain profitable. But in either case, economic rents (=excessive profits) will be competed away.

Conclusion. Most likely Google advertising will continue converging into a perfect market, where it is harder and harder for individual advertisers to extract rents, especially in long-term competition. “Opportunities” is one way of making auctions more transparent and encourage more aggressive bidding behavior. It would be interesting to examine if careless bidding is associated with the use of “Opportunities” (i.e., psychological aspect), and also if Google shows more recommendations to increase than decrease bids (i.e., opportunistic recommendations).

Joni

Digital marketing in China: search-engine marketing (SEM) on Baidu

english
Digital marketing in China: search-engine marketing (SEM) on Baidu

Introduction

China is an enormous market, amounting to 1.3 billion people and growing. Out of all the BRIC markets, China is the furthest in the adoption of technology and digital platforms, especially smartphones and applications.

Perhaps the most known example of Chinese digital platforms in the West is Alibaba, the ecommerce giant with market cap of over 200 $bn. Through Ali Express, Western consumers can order Chinese products – but also Western companies can use the marketplace to sell their products to Chinese consumers. However, this blog post is about Baidu, the Chinese equivalent to Google.

About Baidu

Baidu was founded in 2000, almost at the same time as Google (which was
founded in 1998). Google left China in 2010 amidst censorship issues, after which Baidu has solified its position as the most popular search engine in China.

Most likely due to their similar origins, Baidu is much like Google. The user interface and functionalities have borrowed heavily from Google, but Baidu also displays some information differently from Google. An example of Baidu’s search-engine results page (SERP) can be seen below.

Figure 1   Example of Baidu’s SERP

A lot of Chinese use Baidu to search for entertainment instead of information;
Baidu’s search results page support this behavior. In terms of search results, there is active censorship on sensitive topics, but that is not directly influencing most Western companies interested in the Chinese market. Overall, to influence Chinese consumers, it is crucial to have a presence on Baidu — companies not visible on Baidu might not be considered by the Chinese Internet users as esteemed brands at all.

Facts about Baidu

I have collected here some interesting facts about Baidu:

  1. Baidu is the fourth most visited website in the world (Global Rank: 4), and number one in China [1]
  2. Over 6 billion daily searches [2]
  3. 657 million monthly mobile users (December 2015) [3]
  4. 95.9% of the Baidu visits were from mainland China. [4]
  5. Baidu’s share of the global search-engine market is 7.52% [5]
  6. Baidu offers over 100 services, including discussion forums, wiki (Baidu Baike), map service and social network [6]
  7. Most searched themes are film & TV, commodity supply & demand, education, game and travel [7]

The proliferation of Internet users has tremendously influenced Baidu’s usage, as can be seen from the statistics.

How to do digital marketing in Baidu?

Baidu enables three type of digital marketing: 1) search-engine optimization (SEO), 2) search-engine advertising (PPC), and 3) display advertising. Let’s look at these choices.

First, Baidu has a habit of favoring its numerous own properties (such as Baidu News, Zhidao, etc.) over other organic results. Even up to 80% of the first page results is filled by Baidu’s own domains, so search-engine optimization in Baidu is challenging. Second, Baidu has a similar network to GDN (Google Display Network). It includes some 600k+ websites. As always, display networks need to be filtered for ad fraud by using whitelisting and blacklisting techniques. After doing that, display advertising is recommended as an additional tactic to boost search advertising performance.

Indeed, the best way to reach Baidu users is search advertising. The performance of PPC usually exceeds other forms of digital marketing, because ads are shown to the right people at the right time. Advertising in Baidu is a common practice, and Baidu has more than 600,000 registered advertisers. Currently advertiser are especially focusing on mobile users, where Baidu’s market share is up to 90% and where usage is growing the fastest [8].

How does Baidu advertising work?

For an advertiser, Baidu offers similar functionalities than Google. Search-engine advertising, often called PPC (pay-per-click), is possible in Baidu. In this form of advertising, advertisers bid on keywords that represent users’ search queries. When a user makes a particular serch, they are shown text ads from the companies with winning bids. Companies are charged when their ad is clicked.

The following picture shows how ads are displayed on Baidu’s search results page.

Figure 2   Ads on Baidu

As you can see, ads are shown on top of the search results. Organic search results are placed after ads on the main column. On the right column, there is extra “rich” information, much like on Google. The text ads on Baidu’s SERP look like this:

Figure 3   Text ads on Baidu

The ad headlines can have up to 20 Chinese characters or 40 English characters, and the description text up to 100 Chinese characters or 200 English characters. There is also possibility to use video and images in a prominent way. Below is an example of Mercedez Benz’s presence in Baidu search results.


Figure 4   Example of brands presence on Baidu

It can be easily understood that using such formats is highly recommendable for brand advertisers.

How to access Baidu advertising?

Baidu’s search advertising platform is called Phoenix Nest (百度推广). The tools to access accounts include Web interface and Baidu PPC Editor (百度推广助手).

To start Baidu advertising, you will need to create an account. For that, you need to have a Chinese-language website, as well as send Baidu a digital copy business registration certificate issued in your local country. You also need to make a deposit of 6500 yuans, of which 1500 is held by Baidu as a setup fee and the rest is credited to your advertising account. The opening process for Baidu PPC account may take up to two weeks. Depending on your business, you might also need to apply for Chinese ICP license and host the website in mainland China.

Alternatives for Baidu

There are other search providers in China, such as 360 Search and Sogou but with its ~60% market share in search and ~50% of overall online advertising revenue in China, Baidu is the leading player. Additionally, Baidu is likely to remain on top in the near future to its considerable investments on machine learning and artificial intelligence in the fields of image and voice recognition. Currently, some 90% of Chinese Internet users are using Baidu [9]. For a marketer interested in doing digital marketing in China, Baidu should definitely be included in the channel mix.

Other prominent digital marketing channels include Weibo, WeChat, Qihoo 360, and Sogou. For selling consumer products, the best platforms are Taobao and Tmall – many Chinese may skip search engines and directly go to these platforms for their shopping needs. As usually, companies are advised to leverage the power of superplatforms in their marketing and business operations.

Sources

[1] Alexa Siteinfo: Baidu <http://www.alexa.com/siteinfo/baidu.com>
[2] Nine reasons to use Baidu <http://richwaytech.ca/9-reasons-use-baidu-for-sem-china/>
[3] Baidu Fiscal Year 2015 <http://www.prnewswire.com/news-releases/baidu-announces-fourth-quarter-and-fiscal-year-2015-results-300226534.html>
[4] Is Baidu Advertising a Good Way to Reach Chinese Speakers Living in Western Countries? <https://www.nanjingmarketinggroup.com/blog/how-much-baidu-traffic-there-outside-china>
[5] 50+ Amazing Baidu statistics and facts <http://expandedramblings.com/index.php/baidu-stats/>
[6] 10 facts to understand Baidu <http://seoagencychina.com/10-facts-to-understand-the-top-search-engine-baidu/>
[7] What content did Chinese search most in 2013 <https://www.chinainternetwatch.com/6802/what-content-did-chinese-search-most-2013/#ixzz4G59YyMRG>
[8] Baidu controls 91% mobile search market in China <http://www.scmp.com/tech/apps-gaming/article/1854981/baidu-controls-91pc-mobile-search-market-china-smaller-firms>
[9] Baidu Paid Search <http://is.baidu.com/paidsearch.html>

Joni

Media agency vs. Creative agency: Which will survive?

english

In space, nobody can hear your advertising.

Earlier today I wrote about convergence of media agencies and creative agencies. But let’s look at it from a different perspective: Which one would survive? If we had to pick.

To answer the question, let us first determine their value-provided, and then see which one is more expendable.

Media agencies. First, media agencies’ value-provided derives from their ability to aggregate both market sides: on one hand, they bundle demand side (advertisers) and use this critical mass to negotiate media prices down. On the other hand, they bundle supply side (media outlets) and therefore provide efficiency for advertisers – the advertisers don’t need to search and negotiate with dozens of providers. In other words, media agencies provide the typical intermediary functions which are useful in a fragmented market. Their markup is the arbitrage cost: they buy media at price p_b and sell at p_s, the arbitrage cost being a = p_s – p_b.

Creative agencies. Second, creative agencies value-provided derives from their creative abilities. They know customers and have creative ability to create advertising that appeals to a given target audience. They usually charge an hourly rate, c; if the campaign requires x working hours, the creative cost being e = c*x. And consequently, the total cost for advertiser is T = e+a. We also observe double marginalization, so that e+a > C, where C is the cost that either agency would charge would they handle both creative and media operations.

Transition. Now, let’s consider the current transition which makes this whole question relevant. Namely, the advertising industry is moving into programmatic. Programmatic is a huge threat for intermediation since it aggregates fragmented market players. In practice this means that the advertisers are grouped under demand-side platforms (DSPs ) and the media under supply-side platforms (SSPs). How does this impact the scenario? The transition seemingly has an impact on media agencies, but not on creative agencies — “manual” bundling is no longer needed, but the need for creative work remains.

Conclusion. In conclusion, it seems creative agencies are less replaceable, and therefore have a better position in vertical integration.

Limitations. Now, this assumes that advertisers have direct access to programmatic platforms (so that media agencies can in fact be replaced); currently, this is not the standard case. It also assumes that they have in-house competence in programmatic advertising which also is not the standard case. But in time, both of these conditions are likely to evolve. Either advertisers acquire in-house access and competence, or then outsource the work to creative agencies which, in turn, will develop programmatic capabilities.

Another limitation is that the outcome will depend a lot on the position towards the client base. Whoever is closer to the client, is better equipped to develop the missing capabilities. As commonly acknowledged, customer relationships are the most valuable assets in advertising business, potentially giving an opportunity to build missing capabilities even when other market players would have already acquired them. But based on this “fictional” comparison, we can argue that creative agencies are better off when approaching convergence.

Joni

A few thoughts on ad blockers

english

Anti-ad-blockers are becoming common nowadays.

Introduction. So, I read an article saying that ad blockers are not useful for the users. The argument, and the logic, is conventional: 1) the internet is not really free; 2) publishers need advertisers to subsidize content creation which in turn is also in the users’ interest, because 3) they don’t have to pay for the content. Without ads, the publishers will 4) either start charging for the content or go out of business. Either way, 5) “free” content will cease to exist. (As a real example, the founder of Xmarks wrote a captivating article about the consequences of free-riding in startup context. I encourage to check it out. [1])

Problem of rationality. The aforementioned logic is quite good. But where I disagree with the article is the following argument:

“as soon as users understand the implications of ad blockers [they will] delete them […].”

Based on general knowledge of human behavior, that sounds too much like wishful thinking. In this particular case, I think the dynamics of the tragedy of the commons (Hardin, 1968) [2] are more applicable. We might, in fact, consider “free” content as a type of common (shared) resource. If so, the problem becomes evident: as user_i starts exploiting the free content [3], there is no immediate effect either on the user in question or other users.

There is, however, a minimal impact on the environment (the advertising industry). But because this effect is so small (a few impressions out of millions), it is left undetected. Therefore, it is as if exploitation never took place. This not only gives an incentive for the user_i to continue exploitation, but also signals to other users that ad blocking is quite alright. In consequence, the activity becomes widespread behavior, as we now have witnessed.

Mathematically, this could be explained through a step function.

Figure 1 An example of step functions (Stack Overflow, 2012) [4]

The problem is that the negative effects are not linear, but only become an issue when a certain threshold is met. In other words, it is only when user_n exploits (uses ad blocker) when the cumulative negative effects amount to a crisis. At that point, we have a sudden change in the environment which could have been prevented if the feedback loop would be working and accurately reflecting on user behavior.

Complexities. However, the issue is slightly more complex. As many anecdotal and empirical examples show (boiling frog, slippery slope, last straw, etc.), the feedback loop could only work if it had a predictive property, because each transition from state S_t to S_t+1 does not cause an observable effect which would be large enough to justify change of behavior. Thus, prediction of outcomes of a particular behavior is required — something which humans are poor at, especially at a collective level. Second, the availability of information is not guaranteed: the user_i may not be aware of the actions of other users. To solve this problem, a system-level agent with information on the actions of all agents (e.g., ad block users) is required.

Why does ad blocking take place? Indeed, if it’s so harmful, why do people do it? First of all, people may not be aware of it. Advertisers should not be over-estimating users’ rationality or their ability to predict systemic changes; it is not uncommon that systemic problems are ignored by most people. They simply don’t think about the long-term consequences. But even if they did, and realized that the ad blockers ultimately decimate free content, they might still block the ads. Why? Well, for two reasons:

First, 1) the gains from using ad blocker are immediate (getting rid of ad nuisance) and short-term, whereas the gains from not using ad blocker are long term (keeping the “free” content) and give a higher pay-off for others.

Generally speaking, people have a tendency to prefer short-term rewards (instant gratification) over long-term rewards, even if they’d be much higher. That’s why many people buy a lottery ticket every week instead of working hard to realize their dreams. Also, although the long-term benefit of ads does introduce a pay-off for user_i, that payoff is lower than the “service” he is doing for others, so that reward(u_i) < reward(u_I), where I includes i. Under some circumstances, the psychological effect might be to over-emphasize own immediate benefit over a larger long-term benefit when there are others to share it. In fact, such behavior is rational in a way rationality is usually defined: making decisions based on self-interest.

Second, 2) users might expect someone else to fix it; free content is taken for granted and the threat for its existence is not taken seriously. This is commonly known as “somebody else’s problem”. Yes, we know that keeping lights on in the university toilet wastes energy, but let someone else turn them off (this is a real example based on author’s own perceptions…). The user_i perceives, perhaps correctly, that his contribution to the outcome is marginally low, and therefore does not see any reason to change behavior. If you think of it, it’s the same reason why some people don’t see voting worthwhile; what good does one vote do? Paradoxically, it makes all the difference when that logic prevails in large part of the electorate.

Third, 3) they just might not care. The value of free content might not outweigh the nuisance of ads; user_i might just be without content rather than seeing ads.
Even if this scenario seems a tad unrealistic when viewing a users’ entire media consumption, it might apply to a particular publisher. For example, when publisher_j introduces anti ad blockers, the user simply frequents the website of publisher_k instead.

Two drivers are in favor of this development:

  1. Low switching cost – the trouble for going to another site is close to zero,
    so no individual publisher can impose a lock-in (and, deriving from this proposition, they could do so only by forming a coalition, where publisher_j and publisher_k both introduce ad blockers).
  2. Race to the bottom – there is an incentive for a publisher to allow ad blockers
    and think of alternative ways to monetize their content. This is commonly known as “race to the bottom” which means that due to heightening competition, supply-side actors willingly decrease their pay-offs even when there is no definitive signal from the demand side (again, a coalition of strict adherence could solve this).

Conclusion. Many of these problems are modeled in game theory and have no definitive solutions. However, there is some hope. We can distinguish short-term rationality and long-term rationality. If the latter did not exist, anything that requires momentary sacrifice would be left undone. For example, individuals would not get schooling because it is more satisfactory to play Pokémon GO than to go to school (for most people). But people do go to school, and they do (sometimes) make sacrifices for the greater good. Such behavior is also driven by socio-psychological phenomena: say it would be a strict norm in the society not to use ad blockers, i.e. their use would not be socially approved. The norms and values of a community are strong preventers of undesireable behavior: that is why so many indigenous cultures have been able to thrive under harsh circumstances. But in this particular case (and maybe in the West altogether, where common value base is perhaps no more), it is hard to see the use of ad blockers becoming a “no no”. If anything, the youth perceive it as a positive behavior and take cue.

Suggestions. According to the logic of the commons problem, everyone suffers if no mechanism for preventing exploitation is not developed. But how to go about it? I have a couple of ideas:

1) It is paramount that the publishers acknowledge the problem – many of them still run their advertising operations without really thinking about. They say: “Sure, it’s a problem” and then do nothing. In a similar vein, blaming the users is an incorrect response, although it might be empathically understood when examining advertising as a social contract. For example, publishers see that users are violating the implicit contract (exposure to ads –> free content) by using ad blockers, whereas users see that publishers are violating the contract by placing too many ads on the website (content > ads). According to the previous example, there is not a common understanding or definition of the contract — perhaps this is one of the root causes of the problem. People know they are shown ads in exchange for consuming content they don’t have to pay for, but what are the rules of that exchange? How many ads can be placed? What type of ads? Can they circumvent the ads? Etc.

Second, 2) the motives for ad blocker usage need to be clarified in depth – what are they? From my own experience, I can tell I use ad blockers because they make surfing the Web faster. Many websites are full of ads which makes them load slowly – the root cause here would be ad clutter, or (seemingly) willingness to sacrifice user experience over ad money. I’m just one example, though. There may be other motivations as well, such as ads seem untrustworthy, uninteresting, or something else.

Whatever these reasons are, 3) they need to be taken seriously and fixed, going to the root of the problems. Solving the ad blocker problem requires systemic thinking – superficial solutions are not enough. It’s not a question of introducing paywalls or blocking blockers by technical means; rather, it’s about defining the relationship of publishers, users, and advertisers in a way that each party can accept. Because in the end, ad blockers belong to a complex set of problems that can be described as “no technology solution problems” [5], or at least technology is only a part of the solution here.

References

[1] End of the road for Xmarks. Available at: https://web.archive.org/web/20101001150539/http://blog.xmarks.com/?p=1886

[2] Hardin, G. (1968). The Tragedy of the Commons. Science, New Series, Vol. 162, No. 3859 (Dec. 13, 1968), pp. 1243-1248.

[3] Essentially this is equivalent to resource exploitation, although nominally it seems reverse.

[4] Stack Overflow (2012) Plotting step functions. Available at: http://stackoverflow.com/questions/8988871/plotting-a-step-function-in-mathematica

[5] Garrity, E. (2012). Tragedy of the Commons, Business Growth and the Fundamental Sustainability Problem. Sustainability, 4(10), 2443-2471.

Joni

Problems of standard attribution modelling

english

Attribution modelling is like digital magic.

Introduction

Wow, so I’m reading a great piece by Funk and Nabout (2015) [1]. They outline the main problems of attribution modelling. By “standard”, I refer to the commonly used method of attribution modelling, most commonly known from Google Analytics.

Previously, I’ve addressed this issue in my digital marketing class by saying that the choice of an attribution model is arbitrary, i.e. marketers can freely decide whether it’s better to use e.g. last-click model or first-click model. But now I realized this is obviously a wrong approach — given that the impact of each touch-point can be estimated. There is much more depth to attribution modelling than the standard model leads you to believe.

Five problems of standard attribution modelling

So, here are the five problems by Funk and Nabout (2015).

1. Giving touch-points accurate credit

This is the main problem to me. The impact of touch-points on conversion value needs to be weighed but it is seemingly an arbitrary rather than a statistically valid choice (that is, until we consider advanced methods!). Therefore, there is no objective rank or “betterness” between different attribution models.

2. Disregard for time

The standard attribution model does not consider the time interval between touch-points – it can range anywhere from 30 minutes to 90 days, restricted only by cookie duration. Why does this matter? Because time generally matters in consumer behavior. For example, if there is a long interval between contacts A_t and A_t+1, it may be that the effect of the first contact was not very powerful to incite a return visit. Of course, one could also argue there is a reason not to consider time, because any differences arise due to discrepancy of the natural decision-making process of the consumers which results in unknown intervals. Ignoring time would then standardize the intervals. However, if we assume patterns in consumers’ decision-making process, as it is usually done by stating that “in our product category, the purchase process is short, usually under 30 days”, then addressing time differences could yield a better forecast, say we should expect a second contact to take place at a certain point in time given our model of consumer behavior.

3. Ignoring interaction types

The nature of the touch or interaction should be considered when modeling customer journey. The standard attribution model assigns conversion value for different channels based on clicks, but the type of interaction in channels might be mixed. For example, for one conversion you might get a view in Facebook and click in AdWords whereas another conversion might have the reverse. But are views and clicks equally valuable? Most marketers would not say so. However, they would also assign some credit to views – at least according to classic advertising theory, visibility has an impact on advertising performance. Therefore, the attribution model should also consider several interaction types and the impact each type has on conversion propensity.

4. Survivorship bias

As Funk and Nabout (2015) note, “the analysis does not compare successful and unsuccessful customer journeys, [but] only looks at the former.” This is essentially a case of survivor bias – we are unable to compare those touch-points that lead to a conversion to those that did not. By doing so, we could observe that a certain channel has a higher likelihood to be included in a conversion path [2] than another channel, i.e. its weight should be higher and proportional to its ability to produce lift in the conversion rate. Excluding information on unsuccessful interaction, we risk getting Type I and Type II errors – that is, false negatives and positives.

5. Exclusion of offline data

The standard attribution model does not consider offline interactions. But research shows multi-channel consumer behavior is highly prevalent. The lack of data on these interactions is the major reason behind exclusion, but the the same it restricts the usefulness of attribution modelling to ecommerce context. Most companies, therefore, are not getting accurate information with attribution modelling beyond the online environment. And, as I’ve argued in my class, word-of-mouth is not included in the standard model either, and that is a major issue for accuracy, especially considering social media. Even if we want to measure the performance of advertising channel, social media ads have a distinct social component – they are shared and commented on, which results in additional interactions that should be considered when modeling customer journey.

Solutions

I’m still finishing reading the original article, but had to write these few lines because the points I encountered were poignant. Next I’m sure they will propose solutions, and I may update this article afterwards. At this point, I can only state two solutions that readily come to mind: 1) the use of conversion rate (CVR) as an attribution parameter — it’s a global metric and thus escapes survivorship bias; and 2) Universal Analytics, i.e. using methods such as Google’s Measurement Protocol to capture offline interactions. As someone smart said, solution to a problem leads to a new problem and that’s the case here as well — there needs to a universal identifier (“User ID” in Google’s terms) to associate online and offline interactions. In practice, this requires registration.

Conclusion

The criticism applies to standard attribution modeling, e.g. to how it is done in Google Analytics. There might be additional issues not included in the paper, such as aggregate data — to perform any type of statistical analysis, click-stream data is a must have. Also, a relevant question is: How do touch-points influence one another? And how to model that influence? Beyond technicalities, it is important for managers to understand the general limitations of current methods of attribution modelling and seek solutions in their own organizations to overcome them.

References

[1] Funk, B., & Abou Nabout, N. (2016). Cross-Channel Real-Time Response Analysis. in O. Busch (Hrsg.), Programmatic Advertising: The Successful Transformation to Automated, Data-Driven Marketing in Real-Time. (S. 141-151). Springer-Verlag.

[2] Conversion path and customer journey are essentially referring to the same thing; perhaps with the distinction that conversion path is typically considered to be digital while customer journey has a multichannel meaning.

Joni

Programmatic ads: Fallacy of quality supply

english

A major fallacy publishers still have is the notion of “quality supply” or “premium inventory”. I’ll explain the idea behind the argument.

Introduction. The fallacy of quality supply lies in publishers assuming the quality of certain placement (say, a certain website) is constant, whereas in reality it varies according to the response which, in turn, is a function of the customer and the ad. Both the customer and the ad are running indices, meaning that they constantly change. The job of a programmatic platform is to match the right ads with right customers in the right placements. This is a dynamic problem, where “quality” of a given placement can be defined at the time of match, not prior to it.

Re-defining quality. The term “quality” should in fact be re-defined as relevance — a high-quality quality ad is relevant to customers at a given time (of match), and vice versa. In this equation, the ad placement does not hold any inherent value but its value is always determined in a unique match between the customer, ad and placement. It follows that the ad itself needs to be relevant to the customer, irrespective to the placement. It is not known which interaction effect is stronger, ad + customer, or placement + customer, but it is commonly assumed that the placement has a moderating effect on the quality of the ad as perceived by the customer.

The value of ad space is dynamic. The idea of publishers defining quality distribution a priori is old-fashioned. It stems from the idea that publishers should rank and define the value of their advertising space. That is not compatible with platform logic, in which any particular placement can be of high or low quality (or anywhere between the extremes). In fact, the same placement can simultaneously be both high- and low quality, because its value depends on the advertiser and the customer which, as stated, fluctuate.

Customers care about ad content. To understand this point, quality should be understood from the point of the customer. It can be plausibly argue that customers are interested in ads (if at all) due to their content, not their context. If an ad says I get a promotion on item x which I like, I’m interested. This interest takes place whether the ad was placed on website A or website B. Thus, it is not logical to assume that the placement itself would have a substantial impact on ad performance.

Conclusion. To sum up, there is no value in an ad placement per se, but the value realizes if (and only if) relevance is met. Under this argument, the notion of “premium ad space” is inaccurate and in fact detrimental by its implications to the development of the programmatic ad industry. If ad space is priced according to inaccurate notions, it is not likely to match its market value and, given that the advertisers have choice, they will not continue buying such ad inventory. Higher relevance leads to higher performance which leads to advertiser satisfaction and a higher probability of repurchase of that media. Any predetermined notion of “quality supply” is not relevant in this chain.

Recommendations. Instead of maintaining the false dichotomy of “premium” and “remnant” inventory, publishers should strive to maximize relevance in match-making auctions at any means necessary. For this purpose, they should demand higher quality and variety of ads from the advertiser. Successful match-making depends on quality and variety at both sides of the two-sided market. Generally, when prices are set according to supply and demand, more economic activity takes place – there is no reason to expect otherwise in the advertising market. Publishers should therefore stop labeling their inventory as “quality” or “premium” and instead let markets decide whether it is so. Indeed, in programmatic advertising the so-called remnant inventory can outperform what publishers initially would perceive as superior placements.

Joni

Is “premium” ad space a hoax?

english

Answer: It kinda is.

“Premium publishers” and “premium ad space” — these are often heard terms in programmatic advertising. But they are also dangerously fallacious ideas.

I’ll give three reasons why:

  1. A priori problem
  2. Uniformity problem
  3. Equilibrium problem

First, publishers define what is “premium” a priori (before results) which is not the right sequence to do it (a priori problem). The value of ad space — or the status, premium or not — should be determined a posteriori, or after the fact. Anything will risk biases due to guess-work.

Second, what is “premium” (i.e., works well) for advertiser A might be different for advertiser B, but the same ad space is always “premium” or not (uniformity problem). The value of ad space should be determined based on its value to the advertiser, which is not a uniform distribution.

Third, fixing a higher price for “premium” inventory skews the market – rational advertisers won’t pay irrational premiums and the publisher ends up losing revenue instead of gaining “premium” price (equilibrium problem). This is the exact opposite outcome the publisher hoped for, and arises from imbalance of supply and demand.

Limitations

I defined premium as ad space that works well in regards to the advertiser’s objectives. Other definitions also exist, e.g. Münstermann and Würtenberg (2015) who argue the distinctive trait between premium and non-premium media is the degree of its editorial professionalism, so that amateur websites would be less valuable. In many cases, this is an incorrect classifier from the advertiser’s perspective — e.g., placing an ad on a blogger’s website (influencer marketing) can fairly easily produce higher rents than placing it alongside “professional” content. The degree of professionalism of the content is not a major cue for the consumers, and therefore one should define “premium” from the advertiser’s point of view — as a placement that works.

Conclusion

The only reason, I suspect, premium inventory is still alive is due to the practice of private deals where advertisers are more interested in volume than performance – these advertisers are more informed by assumptions than data. Most likely as the buyers’ level of sophistication increases, they become more inclined to market-based pricing which has a much closer association with performance than private deals.

Joni

Algorithm Neutrality and Bias: How Much Control?

english

The Facebook algorithm is a global super power.

So, I read this article: Facebook is prioritizing my family and friends – but am I?

The point of the article — that you should focus on your friends & family in real life instead of Facebook — is poignant and topical. So much of our lives is spent on social media, without the “social” part, and even when it is there, something is missing in comparison to physical presence (without smart phones!).

Anyway, this post is not about that. I got to think about the from the algorithm neutrality perspective. So what does that mean?

Algorithm neutrality takes place when social networks allow content spread freely based on its merits (e.g., CTR, engagement rate); so that the most popular content gets the most dissemination. In other words, the network imposes no media bias. Although the content spreading might have a media bias, the social network is objective and only accounting its quantifiable merits.

Why does this matter? Well, a neutral algorithm guarantees manipulation-free dissemination of information. As soon as human judgment intervenes, there is a bias. That bias may lead to censorship and favoring of certain political party, for example. The effect can be clearly seen in the so-called media bias. Anyone following either the political coverage of the US elections or the Brexit coverage has noticed the immense media bias which is omnipresent in even the esteemed publications, like the Economist and Washington Post. Indeed, they take a stance and report based on their stance, instead of covering objectively. A politically biased media like the one in the US is not much better than the politically biased media in Russia.

It is clear that free channels of expression enable the proliferation of alternative views, whereupon an individual is (theoretically) better off, since there are more data points to base his/her opinion on. Thus, social networks (again, theoretically) mitigate media bias.

There are many issues though. First is the one that I call neutrality dilemma.

The neutrality dilemma arises from what I already mentioned: the information bias can be embedded in the content people share. If the network restricts the information dissemination, it moves from neutrality to control. If it doesn’t restrict information dissemination, there is a risk of propagation of harmful misinformation, or propaganda. Therefore, in this continuum of control and freedom there is a trade-off that the social networks constantly need to address in their algorithms and community policies. For example, Facebook is banning some content, such as violent extremism. They are also collaborating with local governments which can ask for removal of certain content. This can be viewed in their transparency report.

The dilemma has multiple dimensions.

First of all, there are ethical issues. From the perspective of “what is right”, shouldn’t the network prohibit diffusion of information when it is counter-factual? Otherwise, peopled can be mislead by false stories. But also, from perspective of what is right, shouldn’t there be free expression, even if a piece of information is not validated?

Second, there are some technical challenges:

A. How to identify “truthfulness” of content? In many cases, it is seemingly impossible because the issues are complex and not factual to begin with. Consider e.g. the Brexit: it is not a fact that the leave vote would lead into a worse situation than the stay vote, and vice versa. In a similar vein, it is not a fact that the EU should be kept together. These are questions of assumptions which make them hard: people freely choose the assumptions they want to believe, but there can be no objective validation of this sort of complex social problem.

B. How to classify political/argumentative views and relate them to one another? There are different point of views, like “pro-Brexit” and “anti-Brexit”. The social network algorithm should detect based on an individual’s behavior their membership in a given group: the behavior consists of messages posted, content liked, shared and commented. It should be fairly easy to form a view of a person’s stance on a given topic with the help of these parameters. Then, it is crucial to map the stances in relation to one another, so that the extremes can be identified.

As it currently stands, one is being shown the content he/she prefers which confirms the already established opinion. This does not support learning or getting an objective view of the matter: instead, if reinforces a biased worldview and indeed exacerbates the problems. It is crucial to remember that opinions do not remain only opinions but reflect into behavior: what is socially established becomes physically established through people’s actions in the real world. Therefore, the power of social networks needs to be taken with precaution.

C. How to identify the quality of argumentation? Quality of argumentation is important if applying the rotation of alternative views intended to mitigate reinforcement of bias. This is because the counter-arguments need to be solid: in fact, when making a decision, the pro and contra-sides need both be well-argued for an objective decision to emerge. Machine learning could be the solution — assuming we have training data on the “proper” structure of solid argumentation, we can compare this archetype to any kind of text material and assign it a score based on how good the argumentation is. Such a method does not consider the content of the argument, only its logical value. It would include a way to detect known argumentation errors based on syntax used. In fact, such a system is not unimaginably hard to achieve — common argumentation errors or logical fallacies are well documented.

Another form of detecting quality of argumentation is user-based reporting: individuals report the posts they don’t like, and these get discounted by the algorithm. However, Even when allowing users to report “low-quality” content, there is a risk they report content they disagree with, not which is poorly argued. In reporting, there is relativism or subjectivism that cannot be avoided.

Perhaps the most problematic of all are the socio-psychological challenges associated with human nature. The neutral algorithm enforces group polarization by connecting people who agree on a topic. This is natural outcome of a neutral algorithm, since people by their behavior confirm their liking of a content they agree with. This leads to reinforcement whereupon they are shown more of that type of content. The social effect is known as group polarization – an individual’s original opinion is enforced through observing other individuals sharing that opinion. That is why so much discussion in social media is polarized: there is this well known tendency of human nature not to remain objective but to take a stance in one group against another.

How can we curb this effect? A couple of solutions readily come to mind.

1. Rotating opposing views. If in a neutral system you are shown 90% of content that confirms your beliefs, rotation should force you to see more than 10% percent of alternative (say, 25%). Technically, this would require that “opinion archetypes” can be classified and contrasted to one another. Machine learning to the rescue?

The power of rotation comes from the idea it simulates social behavior: the more a person is exposed to subjects that initially seem strange and unlikeable (i.e., xenophobia), the more likely they are to be understood. A greater degree of awareness and understanding leads into higher acceptance of those things. In real world, people who frequently meet people from other cultures are more likely to accept other cultures in general.

Therefore, the same logic could by applied by Facebook in forcing us to see well-argumented counter-evidence to our beliefs. It is crucial that the counter-evidence is well-argued, or else there is a strong risk of reactance — people rejecting the opposing view even more. Unfortunately, this is a feature of the uneducated mind – not to be able to change one’s opinions but remain fixated on one’s beliefs. So the method is not full-proof, but it is better than what we now have.

2. Automatic fact-checking. Imagine a social network telling you “This content might contain false information”. Caution signals may curb the willingness to accept any information. In fact, it may be more efficient to show misinformation tagged as unreliable rather than hide it — in the latter case, there is possibility for individuals to correct their false beliefs.

3. Research in sociology. I am not educated to know enough about the general solutions of group polarization, groupthink and other associated social problems. But I know sociologists have worked on them – this research should be put to use in collaboration with engineers who design the algorithms.

However, the root causes for dissemination of misinformation, either purposefully harmful or due to ignorance, lie not on technology. The are human-based problems and must have a human-based solution.

What are these root causes? Lack of education. Poor quality of educational system. Lack of willingness to study a topic before forming an opinion (i.e., lazy mind). Lack of source/media criticism. Confirmation bias. Groupthink. Group polarization.

Ultimately, these are the root causes of why some content that should not spread, spreads. They are social and psychological traits of human beings, which cannot be altered via algorithmic solutions. However, algorithms can direct behavior into more positive outcomes, or at least avoid the most harmful extremes – if the aforementioned classification problems can be solved.

The other part of the equation is education — kids need to be taught from early on about media and source criticism, logical argumentation, argumentation skills and respect to another party in a debate. Indeed, respect and sympathy go a long way — in the current atmosphere of online debating it seems like many have forgotten basic manners.

In the online environment, provocations are easy and escalate more easily than in face-to-face encounters. It is “fun” to make fun of the ignorant people – a habit of the so-called intellectuals – nor it is correct to ignore science and facts – a habit of the so-called ignorants.

It is also unfortunate that many of the topics people debate on can be traced down to values and worldviews instead of more objective topics. When values and worldviews are fundamentally different among participants, it is truly hard to find a middle-way. It takes a lot of effort and character to be able to put yourself on the opposing party’s shoes, much more so than just point blank rejecting their view. It takes even more strength to change your opinion once you discover it was the wrong one.

Conclusion and discussion. Avoiding media bias is an essential advantage of social networks in information dissemination. I repeat: it’s a tremendous advantage. People are able to disseminate information and opinions without being controlled by mass-media outlets. At the same time, neutrality imposes new challenges. The most prominent question is to which extent should the network govern its content.

One one hand, user behavior is driving Facebook towards information sharing network – people are seemingly sharing more and more news content and less about their own lives – but Facebook wants to remain as social network, and therefore reduces neutrality in favor of personal content. What are the strategic implications? Will users be happier? Is it right to deviate from algorithm neutrality when you have dominant power over information flow?

Facebook is approaching a sort of an information monopoly when it comes to discovery (Google is the monopoly in information search), and I’d say it’s the most powerful global information dissemination medium today. That power comes with responsibility and ethical question, and hence the algorithm neutrality discussion. The strategic question for Facebook is that does it make sense for them to manipulate the natural information flow based on user behavior in a neutral system. The question for the society is should Facebook news feeds be regulated.

I am not advocating more regulation, since regulation is never a creative solution to any problem, nor does it tends to be informed by science. I advocate collaboration of sociologists and social networks in order to identify the best means to filter harmful misinformation and curb the generally known negative social tendencies that we humans possess. For sure, this can be done without endangering the free flow of information – the best part of social networks.

Joni

A New Paradigm for Advertising

english

From its high point, the sheep can see far.

Introduction

In Finland, and maybe elsewhere in the world as well, media agencies used to reside inside advertising agencies, back in the 1970-80s. Then they were separated from one another in the 1990s, so that advertising agencies do creative planning and media companies buy ad space in the media. Along with this process, heavy international integration took place and currently both the media and advertising agency markets are dominated by a handful of global players, such as Ogilvy, Dentsu, Havas, WPP, etc.

This article discusses that change and argues for re-convergence of media and advertising agencies. I call this the new paradigm (paradigm = a dominant mindset and way of doing things).

The old paradigm

The current advertising paradigm consists of two features:

1) Advertising = creative + media
2) Creative planning –> media buying –> campaigning

In this paradigm, advertising is seen as rigid, inflexible, and one-off game where you create one advertising concept and run it, regardless of customer response. You are making a one sizable bet, and that’s it. To reduce the risk of failure, creative agencies use tons of time to “make sure they get it right”. Sometimes they use advertising pre-testing, but the process is predominantly driven by intuition, or black-box creativity.

Overall, that is an old-fashioned paradigm, for which reason I believe we need a new paradigm.

Towards the new paradigm

The new advertising paradigm looks like this:

1) Advertising = creative + media + optimization
2) Creative planning –> media trials –> creative planning –> …

In that, advertising in seen as fluid, flexible, and consecutive game where you have many trials to succeed. The creative process feeds from consumer response, and in turn media buying is adjusted based on the results of each unique creative concept.

So what is the difference?

In the old paradigm, we would spend three months planning and create one “killer concept” which according to our intuition/experience is what people want to see. In the new paradigm, we spend five minutes to create a dozen concepts and let customers (data) tell us what people want to see. Essentially, we relinquish the idea that it is possible to produce a “perfect ad”, in particular without customer feedback, and instead rely on a method that gets us closer to perfection, albeit never reaching it.

The new paradigm is manifested in a continuous, iterative cycle. Campaigns never end, but are infinite — as we learn more about customers, budget spend may increase in function of time, but essentially optimization is never done. The campaign has no end, unlike in the old paradigm where people would stop marketing a product even if the demand for that product would not disappear.

You might notice that the paradigm may not be compatible of old-fashioned “shark fin” marketing, but views marketing as continuous optimization. In fact, the concept of campaign is replaced by the concept of optimization.

Let me elaborate this thought. Look at the picture (source: Jesper Åström) – it illustrates the problem of campaign-based (shark-fin) marketing. You put in money, but as soon you stop investing, your popularity drops.

Now consider an alternative, where you constantly invest in marketing and not in heavy spikes (campaigns) but gradually by altering your message and targeting (optimization). You get results more like this:

Although seasonality, which is a natural consequence of the business cycle, does not fade away, the baseline results increase in time.

Instead of being fixed, budget allocations live according to the seasonal business cycles — perhaps anticipating the demand fluctuations. The timing should also consider the carryover effect.

Conclusion

I suspect media agencies and advertising will converge once again, or at least the media-buying and creative planning functions will reside in the same organization. This is already the way many young digital marketing agencies are operating since their birth. Designers and optimizers (ad buyers) work side-by-side, the former giving instructions to the latter on what type of design concepts work, not based on intuition as old-paradigm Art Directors (AD) would do, but based on real-time customer response.

Most importantly, tearing down silos will benefit the clients. Doing creative work and optimization in tandem is a natural way of working — the creative concept should no longer be detached from reality, and we should not think of advertising work as a factory line where ads move from one production line to another, but instead as some sort of co-creation through which we are able to mitigate advertising waste and produce better results for advertising clients.