Do Targeted Digital Ads Work Better?

In a previous post, I predicted that Google's double digit growth will come to an end most likely before 2020. I argued that this was due to the fact that advertising spend has been pegged at about 1% of GDP for a century, and that this hard ceiling would make it challenging for Google to continue fast growth amidst increasing competition from Facebook and other social networks.

Recently it was reported that P&G slashed 140 million USD from its digital ad spending, but saw sales rise for 2017Q2. The article also mentions that

Over five years, P&G is aiming for $2 billion in marketing cuts, including media, with a heavy emphasis on cleaning up the digital supply chain.

Despite high double digit growth for the current quarter, it is clear that there are clouds in the horizon for Google.

The ceiling for ad spending

The Bloomberg article mentioned in my previous post presented the following chart showing just how constant ad spending has been as a percentage of GDP. The earliest data point in this chart goes back to 1926, which is very much the beginning of advertising as we know it today. This is when corporations started to take advantage of the government propaganda techniques that had been employed during World War I, in a massive effort to get young men to enlist in the military forces.

We can see that even as the advertising media shifted from street posters to radio to television and finally the mobile Internet, nothing has significantly grown the advertising market relative to GDP. The fact that ad spending has not significantly grown since the era when all we had were street posters is remarkable when you think about it. Ads used to only be on the streets but radio allowed private time with families to be targeted as well. Even then, ad spending remained constant. Importantly in the context of digital advertising, the advent of highly targeted digital ads which collect all sorts of private information about virtually person on the Internet have not detectably increased total ad spending.

The ceiling for ad spending is very robust indeed.

Sophisticated analytics in digital ads

A lot has been made about the highly sophisticated analytics that digital advertising makes possible (often at the expense of privacy, of course). By use of tracking across multiple websites, it is possible to see whether customers who saw banner ads actually came to an e-commence site to make the purchase. All sorts of techniques have been devised to even connect online behaviour to purchases at physical stores. All this should make it possible for the advertisers to see whether their online ads were useful or not. At the very least, there should be an improvement compared to the classical dilemma; "Half the money I spend on advertising is wasted; the trouble is I don't know which half."

The P&G example above suggests however that the improvement may have been illusional. Despite all the analytics that suggested otherwise, 140 million USD of their digital advertising budget apparently belonged to the wasted half. Maybe all the analytics that sacrifices user privacy does not provide any value after all.

The truth of targeting

Targeting means showing your ads to the people who are most likely to respond positively. Targeting is also purportedly beneficial to the end-user who will end up seeing more "useful" advertisements. Because of these benefits, argue the ad companies, collecting all your personal information is a reasonable compromise. However, despite their best efforts, P&G's CFO Jon Moeller did not have kind words to say.

Clearly we don't need to be spending money that is seen by a bot and not a person. Clearly we don't need to be spending money on ads that are placed in inappropriate places, and that's why you see a significant reduction.

The targeting capabilities that we have today apparently are not very good at distinguishing between a bot and a real human being with a profile that matches what P&G desired. From this, it is reasonable to assume that our current targeting algorithms are even worse at spotting the difference between two humans with different profiles.

Again, collecting all that personal information seems to have been in vain.

Future developments to watch

P&G is not the only company cutting back on digital advertising. Unilever is reportedly doing the same. As a result, in the next few quarters, we should get a better picture as to whether these companies will continue to see strong sales growth despite cutting back, or whether they will see negative impacts and eventually come re-invest in digital. We will also be able to discern whether or not there will be a ripple effect as other companies reconsider their ad portfolios.

If P&G and others do not observe a negative impact despite a continued cut-back on digital spending, then this will significantly blunt the growth of digital advertising. Although I expect this to slow down considerably within the next few years anyway, this may come earlier than I previously though. In the short term however, this will actually benefit Google because they are most likely to be able to maintain advertiser trust by implementing measures to counter the bot issues. In aggregate, I would expect a short term boost followed by an earlier slowdown for Google and Facebook , and an much more imminent downfall for other digital advertising companies.

One possible change that I would very much welcome, is a better understanding of how valuable our personal data really is. Television and magazine advertisements do not collect your personal information, but are nonetheless targeted to a certain degree based on the programme or genre that you are viewing. Targeting itself does not necessarily need personal information as long as the ad placement itself is intelligent, and targeting does not need to stalk your whole Internet browsing habits. Digital advertising might not necessarily need to stalk you. By understanding the true value of these privacy intrusions, our society should be in a much better position to discuss whether we have to make these concessions or not.

Can advertising grow beyond the 1% ceiling?

Given that a large number of Internet companies rely on advertising as their main revenue source, the presence of a hard ceiling should worry venture capitalists who are pouring ever increasing amounts of money into them. If tech is to continue to grow as a whole by high double digits, then tech needs to find a way to either break out of this, or to develop new business models with a more direct revenue.

However, as I have argued, I consider it unlikely that digital advertising is more revolutionary than radio or television advertising, and I strongly doubt that the 1% ceiling will be broken. As digital advertising saturates the ad market as a whole, this market will become a zero-sum game and will not contribute to the growth of overall tech.

Therefore, my belief is that tech needs to stop relying on advertising and that this is starting to be an urgent issue. As the tech advertising space saturates, the current incumbents will become stronger and stronger albeit with slower growth rates. On the flip side, it will be harder and harder for new entrants with an advertising business model to make it. Advertising will quickly cease to be a viable revenue strategy for start ups.

Effectiveness of Online Ads

Jordan Weissmann an article that is very much in alignment with my views and experiences with online advertising.

“We Have No Idea If Online Ads Work: The Internet has given us an ocean of data. Turns out, most of it is pretty useless.”

Some excerpts;

Last year, a group of economists working with eBay’s internal research lab issued a massive experimental study with a simple, startling conclusion: For a large, well-known brand, search ads are probably worthless.

For instance, companies like to run large ad campaigns during major shopping seasons, like Christmas. But if sales double come December, it’s hard to say whether the ad or the holiday was responsible. Companies also understandably like to target audiences they think will like what they’re selling. But that always leads to the nagging question of whether the customer would have gone and purchased the product regardless. Economists call this issue “endogeneity.” Derek Thompson at the Atlantic dubs it the “I-was-gonna-buy-it-anyway problem.”

In the end, it all comes down to the evergreen challenge of distinguishing correlation (e.g., a Facebook user saw an ad and then bought some shoes) from causation (e.g., a Facebook user bought some shoes because he saw an ad).

This is exactly the reason why we stopped using AdWords for our antibody search service. Visitor numbers dropped, but our profits didn’t.

Maybe brands that are just starting out need to use AdWords. Maybe if they never gain brand recognition or loyalty, these brands might have to use AdWords forever. However, if your brand has managed to get some recognition, or if your product offerings are unique and listed high in organic search, it is very likely that AdWords is reducing your profit rather than increasing it.

Google and other internet advertising companies are telling advertisers how much data they have and how effectively they can target customers. However, as a recipient of these advertisements, I’m seldom amused when they blatantly show ads from websites that just I visited the day before. If this is all that big data can do, then Internet advertising companies have a big problem.

Google Plus is an SEO Tool

There was a good article on the New York Times about Google’s spooky social network, Google Plus.

Some quotes from the article;

Thanks to Plus, Google knows about people’s friendships on Gmail, the places they go on maps and how they spend their time on the more than two million websites in Google’s ad network. And it is gathering this information even though relatively few people use Plus as their social network. Plus has 29 million unique monthly users on its website and 41 million on smartphones, with some users overlapping, compared with Facebook’s 128 million users on its website and 108 million on phones, according to Nielsen.

Starbucks, for instance, has three million followers on Plus, meager compared with its 36 million “likes” on Facebook. Yet it updates its Google Plus page for the sake of good search placement, and takes advice from Google representatives on how to optimize Plus content for the search engine.

“When we think about posting on Google Plus, we think about how does it relate to our search efforts,” said Alex Wheeler, vice president of global digital marketing at Starbucks.





米国のデータです。それほど大きなスパイクではありませんが、Chromeは週末に多く使われているのがわかります。それに対してFirefoxはスパイクが全く見られず、職場でもプライベートでも使われていることがわかります。参考までにInternet Explorerは週末に下にスパイクしており、週末には比較的使われていないことがわかります。

StatCounter browser US daily 20130620 20131120


StatCounter browser JP daily 20130620 20131120


StatCounter browser DE daily 20130620 20131120

ブラジルのデータです。ここはInternet Explorerが週末に利用が増えていて、Firefoxは平日に利用が増えています。またChromeは余りはっきりしません。

StatCounter browser BR daily 20130620 20131120




StatCounter browser BR daily 20091120 20111120


StatCounter browser BR daily 20111120 20131120


StatCounter browser US daily 20091120 20111120


StatCounter browser DE daily 20111120 20131120



  1. 新しい製品が浸透するのは、職場よりも圧倒的にプライベートでの利用が早いです。ブラジルにおいてはChromeが職場に浸透してきましたが、他の国(米国、ドイツ)でChromeが浸透していくかは未知数です。
  2. 職場ではInternet Explorerが強みを発揮していますが、standards compliantなブラウザが必要なときは歴史的にFirefoxが使われてきたと思われます。いったん確立されたこのFirefoxの立場は、容易にはChromeに変わらないのでしょう。



NetMarketShareもStatCounterもアクセスログ分析ツールです。Google Analyticsと同じようなものです。そしてNetMarketShareやStatCounterのツールが導入されているウェブサイトのすべてからデータを集めて、ブラウザの使用シェア統計を算出しています。

しかしここが大きな問題です。ウェブに関わっている人は一人残らずGoogle Analyticsのことを知っていますが、NetMarketShareにしてもStatCounterにしても、全く聞いたことがない人が多いのではないでしょうか。

それもそのはずです。Wappalyzerというブラウザ拡張機能を使って調査されたデータを見る限り、68%のウェブサイトはGoogle Analyticsを導入しているものの、StatCounterはわずか2%のサイトしか導入していないのです。NetMarketShareはこれよりもさらに導入数が少なくなっています。つまりStatCounterおよびNetMarketShareはウェブ全体から見ると極小さい標本サイズしかなく、なおかつ標本の中にバイアスがある可能性が高いからです。


  1. シェア(何%)の絶対値は信用しません。例えば2013年12月時点でInternet Explorerのシェアが57.91%というのは基本的に信用できません。またChromeの16.22%も絶対値として信用しません。原則として傾向だけを見ます。
  2. 複数の国を合算した数値は信用しません。国によって標本サイズが違うからです。StatCounterはこのあたりを公開していて、例えばトルコの標本サイズが異常に高いことがわかります。したがって合算した数値よりは、個々の国の数値の方が意味を持ちます。それでも個々の国を見るのではなく、複数の国を見て傾向を判断することが必要です。
  3. 時系列の変化はある程度信用できますが、実際にグラフを見てみると不思議な挙動が起こることが頻繁にあります。したがって時系列の変化も鵜呑みにできません。






  1. メーカーウェブサイトのボリュームの大半は製品ページであって、これはほとんど更新することがありません。更新するとしても印刷版のカタログと同期させることが(本来は)必要なので、どこかで更新履歴がまとまっていないと印刷版との同期がおかしくなります。
  2. メーカーウェブサイトの更新の大半はキャンペーンや新製品です。どれも新しいページの作成が必要です。特にキャンペーンページはデザインが重要なので、ウェブデザイナーに依頼します。製品担当が直接ページを作成することはありません。新製品の場合、注力製品であればイメージなども多いので、やはりウェブデザイナーに依頼するのが無難です。




  1. CMSはメーカーにとってはメリットがない
  2. CMS導入業者にとっては継続的に管理費やカスタマイズ費がもらえるので、メリットだらけ


  1. CMSはテンプレートでデザインの統一性が保てます。DreamWeaverにもテンプレート機能が充実しています。そしてCMSのテンプレートとDreamWeaverのテンプレートの最大の違いは、CMSだとプログラマーが変更しなければならないのに対して、DreamWeaverのテンプレートはウェブデザイナーが変更できる点です。
  2. CMSに機能拡張をすれば、価格などの情報をデータベースから引っ張ったりできますので、常に最新の価格情報を保証することができます。しかしこれはまず非常に高価です。そしてカスタムなので業者依存がますます強くなります。このあたりは私たちが新しく用意しているサービスを利用すると、うんと安くできます。業者依存も生じません。







  1. 全調査データ


いろいろ思うことはあるのですが、特にスマートフォンの普及率が興味深いです。高校生のスマートフォン普及率は既に14.9%に達しているそうです。それに呼応するかのように、mixi (23.4%)よりもTwitter (34.0%)の方がすでに利用率が高くなっているとのことです。Facebook (12.7%)も決して少ない数字ではないです。これは驚きでした。






バイオの買物.com まとめて抗体検索


目標が達成されているかどうか、それはこのページを見ているご自身で判断ください。Twitterの @naofumi もしくは @BioKaimono に感想をいただければうれしいです。

Exact Antigen 改め



ページのトップからAntibody Searchを探して抗体検索のページに移動します。



検索システムで特徴的なのは蛍光色素の選択の仕方です。同じ蛍光波長のものでも、メーカーによって使用する蛍光色素は大きく異なります。特に緑や赤の領域は同じような蛍光色素がたくさんあります。そこでBiocompareでは蛍光波長によってグループ分けし、Blue, Green, Yellow, Orange…などと選択できるようにしています。これは非常に便利です。













しかし残念ながらBiocompareのように蛍光波長でグループ分けせず、蛍光色素のブランドでグループ分けしています。例えば “Alexa Fluor”とか”Cy”とかいうグループ分けです。正直、これじゃ何の色が選ばれるのか全くわかりません。










  1. 収益モデルがウェブサイトの使い勝手を決める例:バイオ百科
  2. バイオ百科で抗体検索:僕が使いにくいと感じるところ

















Facebook, Twitter, Mixiのデモグラフィックを見てみる

Facebookが日本で普及し始めていることを受けて、斉藤 徹さんがブログ”In the looop”に書いた記事を先日紹介しました。













DoubleClick Ad Plannerに見るデモグラフィックの違い

各ウェブサイトを利用しているユーザのデモグラフィックを見るツールとして、Google提供のDoubleClick Ad Plannerがあります。ここのデータの信憑性はハッキリ分かりませんが、他の方法では入手しがたい有益なデータが得られます。FacebookのデータMixiのデータ


Facebook access trend


2011 02 18 00 36 22



恐らく斉藤 徹さんの意見は45歳以降の世代の考えを代表していて、「もとまか日記」さんの意見はそれより前の世代の考えを反映しているのではないでしょうか。お互いの意見が異なるように見えますが、単に違う世代、違うデモグラフィックについて語っているだけのように思えます。


最後に結論ですが、もし斉藤 徹さんの意見が45歳以降の動向を正確に見極めたものであるならば、日本の人口構成を考えたとき、間違いなくFacebookはMixiに迫るかもしくは追い抜く存在になるでしょう。