Yieldbot is webscale. In 2014 our era collected andprocessed consumer undertaking on 8 billion client sessions across almost 400premium writer computer and mobile web pages. This equated to over 23 billionpage views of first party data and accounting for ad slots, over 50 billionreal time selections. We do not measure “unique guests” since we are a “cookieless”era but on an combination level we likely have one of the largestfootprints in electronic ads.
Certainly we’ve got one of the largest inpremium content material and the largest collection and use of first party publisherdata for monetization. Woohooo!Our growth is fueled completely by the functionality of themedia. We sell on a performance basis – 70% of our campaigns are run on aCost Per Click CPC model. Our good fortune is measured most on functionality of theaudiences we are helping drive to the marketer’s own sites and digitalexperiences. In that regard, we regularly beat the functionality of paid searchand crush other demonstrate applied sciences for our clients ask for the case studies.
Also, 2014 was our first full year of offering mobile ads. 25% of our revenuein 2014 was from mobile. In Q4 it is 33%of our income. Mobile is growing fast. A large a part of our publisher luck is that we deliveredCPMs 2 4X what publishers have become from ad exchanges or other companions suchas SSPs.
Since the spine of our tech is machine learning or as I like tosay equipment “incomes” our most a success publishers are those ourtechnology has been on the longest. They have become CPM in the $5 $10 range. This is on computer and mobile. Maybe our greatest story of the year is a big sitewith 22M unique friends a month that have been a huge Google partner but is nowgetting RPM from Yieldbot as high as 11X Google and for the first time ever hasa partner offering more total monthly income than Google. Cha ching!One other surprising thing happened on the publisher side of ourbusiness this year.
Our good fortune with their first party data result in publishersasking to use our era to serve their own direct sold campaigns andcreation of a carrier model for Yieldbot. These publishers, some of the largeston the net, at the moment are offering CTR functionality 5X what they were providing runningcampaigns with data from their DMP and using their Ad Server waterfall todecision. In addition, they are offering back end metrics that permit them tocompete for budgets going to Facebook and Google. Ask for the casestudies. We did it all in a totally obvious manner, constructing trustwith our direct relationships at the businesses and with publishers.
Ifpublishers are looking to know their maximum RPM pages by referrer source for Mondaymornings we are happy that we can give that data. If advertisers want toknow what inventive headline drove the highest conversion rate in the afternoon weare happy that we may give that data. What we’re most pleased with is that notonly have we built a company that can provide this deep level of learning, butalso by the point we have that data it is already being acted upon by our techto enhance business effects. A sure group of publishers have sophisticated light yearsahead of the existing discourse. Advanced beyond the postulate of programmaticefficiency. Advanced beyond the idea of developing segments by coping with cookies.
Advancedfar beyond unactionable analytic reviews related to viewability. Thesepublishers, many of whom are essentially the most respected names in media, have deployed device learning and synthetic intelligence in their media. They have movedbeyond optimizing delivery of impressions to predictive algorithms optimizing adserver selections in real time against the performance of their media. Publishers won’t live on in a world where they do notknow when, why, where, how and a person is strolling into their store. They won’t live to tell the tale in the event that they don’t know what customers are buying and the way much they’re paying.
No enterprise could survive that lack of dataand intelligence. In fact, no customer really wants that form of store either. Customers want to buy products that serve their supposed goal. Customers want sellers to recognize — even expect — what they might be attracted to. Customer event extends to purchasing media the same as buying anything else. This means marketer performance.
The data is a window into the client mindset and journey. Importantlyfor publishers it is an express first party value exchange among thepublisher and the client. It is an trade of intent for content, of mindsetfor media. It is what brands want more than anything else. The moment, thezeitgeist, the exact time and place a consumer is considering, discovering,evaluating, evaluating, studying, and coming across what and where to buy.
It isthe single most valuable moment in media. It is ideal time, right place, rightmessage. It is uniquely digital, uniquely first party and owned solely by thepublisher. It is a gold mine that Facebook and Google admire and they have focusedtheir recent writer side tasks on capturing from publishers either unsuspectingor incapable of extracting the value themselves. As publishers begin to recognize these moments themselves, activatethem for retailers and optimize the performance of the media towards them, anamazing thing occurs. The basic value of the media increases.
It increasesbecause of intelligence. It increases because of functionality. Most essential, itincreases by reason of value beingdelivered to clients. It also opens up new budgets. Over time, these systems will get smarter. With more data, publishers may even begin to sell based on performance thereby disposing of a host of issues around impression based buying and increasing usual RPM by orders of magnitude with higher efficient CPMs and smarter, more effective allocations of impressions.
Recall term frequency is described as a role of the document id and term, tfdocument, term. At this point we’ve got an RDD of raw term frequencies, but we want normalized term frequencies. We use the flambo inline anonymous function macro, f/fn, to define an nameless Clojure position to normalize the frequencies and map our doc term seq RDD of tuples to an RDD of key/value, ], tuples. This new tuple format of the term frequency RDD will be later used to affix the inverse doc frequency RDD and compute the final tf idf weights. Marceline’s publicity of Storm metrics has been very useful for monitoring the behavior of Yieldbot’s topologies. Friction around instrumentation has been enormously decreased.
Code smells are down. Metrics now entail fewer lines of code and less duplication. An additional architectural advantage is that dependencies on external facilities can be remoted to particular person topology additives. It is painless to add ordinary metrics while sustaining enough flexibility for custom metrics when necessary. We have designed Marceline’s metrics in particular with the goal to leverage Storm’s metrics API unobtrusively. This funding but it isn’t about defining Yieldbot.
Ourfocus is on our customers. It’s more accurate to say it’s about redefining whatis possible with exhibit advertisements. It’s about new buyers into new inventory. It’s about new generation applied in new ways. It’s about new levels ofperformance for sellers and publishers.
It’s a few new channel of real timemedia it’s created explicitly by consumers in a privacy friendly way. We arecreating media value in the electronic ad ecosystem at levels no one ever hasbefore and in ways no one ever has before. That is why we raised money. It isvalue we can continue to invest in for marketers, publishers and consumers. Other strategies, by not optimizing according to relevance, are by definition optimizing on the inaccurate thing for the user and writer and advertiser for that matter. Retargeting for instance optimizes to a cookie that was set some time in the past on some other domain, with out regard for even if the message is relevant to the user at that moment.
That’s why retargeting examples are so jarring. More often than not their message is no longer applicable, and the experience is a reminder to the user that they’re being followed across the web; in the process detracting from the adventure on the present writer site. Contextual, as an alternative example, only looks at one of the elements for relevance and might only drive simple concentrated on rules which are greatly described. Yieldbot hascreated a new advertising channel in the top class writer environment leveragingintent alerts and the buying and matching concepts of Search 1stparty data, in session/real time matching, CPC pricing, and key phrase focused adcreative that does both. Yieldbot closes the loop on consumers by assembly thereal time consumer intent with a relevant ad to drive immediate action and builds demand for merchandise byputting in view, seen relevant ads when clients are most receptive to themessage, during lean ahead content experiences eventually expanding the likelihoodof ad engagement and message receptivity.
The campaignresults spoke volumes. Yieldbot, targeting real time intent groups for energy,healthy living, health, wellbeing, and healthful eating drove a campaign average19% coupon download rate, and a 33. 7% lift in consumer purchase intent for thosethat either “definitely will” or “probably will” purchase the product withinthe next 30 days. Additionally, the campaign drove 6x above industry averageengagement rates CTR. We were not given exact data points for othermarketing channels Search, demonstrate, video nonetheless it when budget cuts for thebrand had to be made Yieldbot was still status.
Our ROI performance allows any other constituent in our marketplace and the source of the information, top class publishers, to attach directly to SEM budgets for the first time because the inception of Paid Search. Considering the mounting monetization demanding situations premium publishers face this is a big market development. We’re lucky to count most suitable media businesses like Meredith, Rodale, Remedy Health, BlogHer, BabyCenter, Internet Brands and plenty of others as partners. We’re closing in on 1. 5 Billion Page Views a month of real time intent. Compared to look query volume this can make Yieldbot the 3rd largest real time intent source on the internet behind Google and Yahoo/Bing.
This post celebrates an anniversary. A year ago today and in response to two years of prior work building our real time intent era, we launched what we expect is by far the simplest company model for the net. We launched a real time keyword marketplace. It’s one of the few ever created external of Search. A year ago we began with 3 writer sites and 1 advertiser.
Yieldbot now is deployed on over 60 premium web pages and content networks within the three verticals serving 125 cost per click campaigns. We all love stats at Yieldbot so here are some more stats about Yieldbot. We’ve made 5,753,199,447 real time match choices most were NOT to serve an ad Yieldbot crunches 14TB of data a day Revenue has doubled every 2 months since launch. The most essential stat is needless to say consequences. The vast majority of Yieldbot advertisers get performance at or exceeding their Search Marketing efforts.
Our work has no precedent there. Our second advertiser gave us an idea we were onto something. After 6 weeks buying on Yieldbot they reported that our conversion rate was 35% better than Paid Search. Then our third advertiser told us a similar story. We’ve kept listening to it and it never gets old.
Yieldbot industry connects the two biggest channels of electronic ad supply and demand. Premium publishers get the performance era they absolutely should live on electronic. Search Marketers get much needed new inventory. Parties that never did business with each other before now do it transparently and in real time through Yieldbot. It’s a wonderful thing to behold. You might be asking of yourself how does this work?What is Real Time Intent?People browse the net with aim and course.
A real time “why” is the most powerful concentrated on dimension Marketing has ever known. Yieldbot works to understand the genuine time “why” a person is clicking to any and every page of web page. We try this via our superior keyword navigation path analytics, our utilized data science and our intelligent real time match choices. So 365 days in and we’ve won the trust of many of the most popular media brands and advertising budgets on the internet. Now it’s time for Yieldbot to construct on that basis.
We’ve introduced our Mobile solution and our real time intent has caught interest in Social advertisers as well as Search. We’re psyched. We still fully manage the campaigns ourselves and source our own demand. Until we fully make it a self serve enterprise we all know to a big extent we won’t truly be a market. We are working towards starting Yieldbot up via APIs to quite a number demand partners by early 2014. Real time click stream intelligence is a place where few have ever ventured.
Our data science team, our engineering team, our sales and company development team and our account strategy teams are learning for ourselves what Yieldbot is, what it can be and what it should be. The early outcomes are astounding to even us. Year 1 was an amazing journey. The journey keeps. We hope you join us on it.
BYOD in online advertisements is a part of a broader trend toward addressable media that goes well beyond online. For simplicity, let’s define addressable media as the concentrated on of ads to a specific shopper or family through data via a media channel. The data can come from either the advertiser side or the writer side, once in a while a combination of the two, and frequently will include 0,33 party data company. For instance, an insurance company ran a TV crusade currently for renter’s insurance in collaboration with two satellite TV companies targeted only to households that were identified as rentals. It’s indispensable that the online publishing neighborhood embraces a BYOD technique of their very own. Publishers must look to their viewers and their behaviors and create meaningful insights that fuel ad fits with the needs of dealers.
This BYOD effort by publishers has to begin with truefirst party data about the site and the user adventure. While many publishers are embracing DMP’s Data Management Platform, consider the info sets being utilized by the DMP. They are often commoditized third party data sets that advertisers already use on exchanges and in other places. And unless you have big scale as a publisher, this commoditized data won’t really be advantageous. Publishers should create new first party data that’s differentiated and utilize it to satisfy the purposes of advertisers and likewise their site’s users who are searching for a relevant, timely adventure both in the content material they seek and the ads they have interaction with. The chance has never been bigger and by all measures the ordinary online ad spend will proceed to grow.
But, the risks are also high and the distribution of dollars between search, social, mobile, reveal, exchanges and new platforms continues to be determined. Advertisers have lots of options and when you are a writer whose data would you favor to be operating with?This isn’t a winner take all market, but publishers that include BYOD and create true data and insights into their site experience and users stand a strong chance to create separation available in the market and drive value for their advertisements companions and their viewers. The motivating precept behind what we would have liked in our configuration database was a reliable concept of versioning. We had previously tried to manually implement some idea of versioning in the context of available database generation. This would keep around older types of objects in a separate area with a version number picking them, and application logic would move copies of whole items around as alterations were made to them. Data events would comprise types of the items that they were linked to.
While this did a few of what we wanted, it was clear that this was not the answer we were looking for. The simplicity of the mindset hit two strong notes for us. First was that the simplicity brought with it a kind of elegance. It is straightforward to know how the database handles history and to reason in regards to the variety of the data in the database. We also immediately got capability like audit logging built into the database for free.
This is one reason Yieldbot ordinary platform CTR click via rate has gone up each month whilst influence levels have skyrocketed. There are efficiencies created as markets get larger and people will advantage both Yieldbot advertisers and publishers. We head into 2013 fully aware that we haven’t accomplished something near our goals and we are still at the start of building our company. We have 2 new verticals launching Q1 and more growth to manage in advance of us. There is also pressure that comes from the sheer enormity of the opportunity in front of us. That’s a great thing.
It was Billie Jean King that said “force is a privilege” and we’re privileged to be solving complications that bring the best quality clients, world’s top marketers and top rate content publishers together in a way that delivers relevance and cost to each concurrently. Creative – if there has been one course and point on which it seemed almost all of these specialists agreed it was that new inventive formats, creative applied sciences, inventive optimization methodologies will proceed to be at the forefront of the industry’s innovation. As GYB Google, Yahoo, Bing and the other “basic” platforms proceed to experiment and innovate with SERPs that are more dynamic and visually pleasing for the twin applications of better engagement and the potential to allure more dollars from brands with higher levels of fear about image, there are interesting groups well represented which are doing cool stuff to leverage this need. One that comes to mind is Datapop a real era innovator in this space. Then there’s the easy blockading and tackling of better text copy optimization being tackled by folks like BoostCTR less a era and more a time saving method for the drudgery of copy testing. New Platforms – the other area where much of the most appealing discussions happened and where the opportunities for industry growth seem most fertile were in new platforms.
Everything from expanding growth of mobile and tablet search and its alternative nuances from laptop to the arriving of more vertical search encompassed by Yelp, Amazon and others surprisingly none of which have been represented at SIS except for the very cool Intent Media but were much mentioned. Even really cool but maybe a bit scary from a privacy attitude applied sciences like visual search from Xbox were hot topics. These new platforms create new markets and new audiences on top of which the best practices of the hunt industry are being built and their rapid growth is consultant of their promiseIt was never clearer to me than at this great conference that content material publishers not to be confused with eCommerce publishers are such a obvious afterthought to the optimum innovators of essentially the most a success and impactful a part of the electronic advertisements industry… the quest community. There was not a content publisher in the room nor was there a mention of one on the stage except in brief by me. Yet, I bet that if asked in a vacuum and I did a little bit asking that almost all of this proficient group knows that top class content material publishers are hardcore SEO buffs, often times buy Paid Search and are sitting on a treasure trove of first party data.
When properly harvested as we do at Yieldbot this information can illuminate the “search like” behaviors of web visitors in their classes. Selling these company’ top class writer intent in the currency of keywords to the very agents that makeup the search ecosystem that was so well represented at SIS represents an enormous opportunity for market expansion. Not just for those publishers to access search budgets but additionally for search dealers in finding new ways pin pointing the user that they want in real time as they are expressing attention in a particular class. Utilization of this real time data can and does yield often times outcomes even better than traditional search itself far and wide the marketing funnel from branding to conversion. We heard and talked about the usage of third party search data and third party site visitation in dealers’ and eCommerce platforms as a new data set for the search marketplace to leverage its methodologies in buying performance media. There is no doubt that there’s a spot for that and a few have done quite well.
But nowhere in this paradigm is using FIRST PARTY DATA in REAL TIME within publisher content to make ad selections on a similar pubs from which that data is harvested being mentioned. In this paradigm the publishers can take part no less than on par in the hunt and performance electronic market like every these old and new search platforms that agents know and wish more of. Quietly taking potential of writer fumbling is Google. Google has extended its grip on not only the publisher media through its Ad Exchange and AdSense but also their data via Google Analytics and its DFP ad server. Most publishers will openly admit that Google knows how much cash they make and more about their viewers then they do.
Note in the diagram above that apart from the same old usage of git among developers Erik and Sean, the Chef server also is set up with GitHub as the remote. Below the Chef server are two examples of servers UI and DB which are set up with the Chef server as their remote. Note that UI and DB really only pull from the Chef server repo. The Chef server in turn mostly pulls from GitHub, although it does ward off to GitHub the state of tags as manipulated during the deploy. Features/BenefitsYou can easily change the region of the “manufacturing” tag on the Chef server repo and then resync the servers in the platform to deploy any level of code desired.
During a deploy, our servers aren’t based on contacting GitHub. To start a deploy we sync the Chef server’s repo with GitHub, but we could just as easily push to the Chef server’s repo from elsewhere, or even add a special remote to the Chef server’s repo to drag from an alternative region as an alternative. If any adjustments are made to a system to make it deviate from what was last deployed, a “git diff” in the repo on that server can expose exactly what those alterations were made on that genuine server if you’ve worked with deployed code before you recognize this idea is huge. Next UpIn a follow on blog post we’ll cover some more advanced uses of assorted git remotes that we leverage in our development environments. Sideways site visitors that driven from search, portals, aggregators, et al. was a special story.
While it helped drive unique user growth, the traffic was mostly inside and out. After touchdown on an article, we were lucky to see a user turn an extra page before leaving the site. We tried a couple of how to tackle this: we employed applied sciences to surface connected or advised content material based on the content material of the page, conventional field matter, even social sharing. We made assumptions in keeping with referral source and surfaced headlines accordingly – people that came from the Yahoo homepage or Facebook must be attracted to typical, non business articles. If they came from a Y!Finance quote page they needs to be interested in articles in regards to the agency ticker they were looking up. And if they came from Drudge they must be looking for Opinion Page content.
But despite these efforts, we never managed to considerably move the needle on those page views per guest numbers. Let’s say site XYZ. com gets 10M UU’s per 30 days. Any good SEO person will let you know that the sweet spot for search traffic is about 20%. Any less and you’re not optimized. Much more and you’re one Google set of rules tweak away from falling off a site visitors cliff.
The other 8M users are coming at once or thru other “sideway” means equivalent to partner sites, portals, social media, etc. If that bills for 20% of total traffic, you’re doing well – give those biz and viewers development folks raises!But most definitely that traffic is driving 2 or less PV’s/visit 4M PVs/month. By knowing why those users come for your site, surfacing content material it really is of attention to them and getting them to show an additional page, that’s now 6M PVs per month or an increase of 30%. If you’re a subscription based site or have a e-newsletter product, recall to mind applying that knowing to help drive subs!But a funny thing came about in electronic behavior. People skipped over the front inside cover and went right to content material that was relevant to them.
Search’s means to fracture content material hierarchies and convey relevance not only became probably the most loved and valuable application of the internet, it destroyed the postulate of premium content material all in combination. In truth, top class never really existed in a user managed medium as it was never according to something that had to do with what the user wanted. It was based on the basic ad metric of “reach” when during this medium, decisions about what is premium are determined by on demand ability and relevance. The great thing about this medium is in the size of it. Validation for the drowning of top class beyond the proven fact that Wikipedia destroyed Encyclopedia Britannica rests in the functionality of electronic media. A funny thing took place as advertisements functionality became more measured.
Advertisers found out premium didn’t nearly matter up to they theory. There were better ways to drive performance that yielded better and more measureable outcomes. The ability to match messaging to peopleon request and in a applicable way was more helpful during this medium than some content issuer idea of what was “top rate. ” In this medium the general public not the publisher determines what’s top rate. As realtime rules based matching era keeps to improve performance ads and advertising itself keeps to grow at the cost of top class advertising. Today, regardless of those trying to hold on to the past, premium is little more than an train in brand borrowing and little else.
Despite the simplest efforts of the IAB to bring Brand ads to Digital it has fallen as a percent of ad spend for five directly years. In the area we are living in today Mr. Emanuel’s $9 billion dollar prematurely for network TV primetime advertising is $1. 5 billion less in ad earnings than Google made last quarter. Like Search, Publishers should have two critical components to their marketplaces. They need the anxiety of shortage in the industry.
That will drive up demand and force advertisers to spend the time operating on getting better their functionality. This was the cherry on the sundae for Google as a $1billion industry Conversion Testing and Content Targetinggrew out of nowhere to help spends in Search. Most every dollar saved with optimization went to drive more volume – or back to Google. They need a unique currency for the market. Keywords were a completely new way to buy media.
Nothing has ever worked better. Facebook is promoting Actions with OpenGraph. Ultimately advertisers are buying clients not key phrases or actions but there is a completely unique window of chance for publishers at this moment in time to create something new and uniquely people, not page focused. The systems used to fuel these innovations all depend on one herbal useful resource data. Publishers have diamonds and gold in below the floor of their properties.
Mining these data nuggets and using them to improve the performance of their media is the only hope publishers have competing in the world of “CPM Zero. ” Only publishers can uniquely wrap their data with their media and drive functionality in a manner unique to the industry. That’s what Google does. That’s what Facebook does. That’s what Twitter does.
The scarcity discussed above is created because the realtime knowing of site visitor interest and intent is only derived using first party data as rules and integration with the writer ad server for beginning. So pubs are really left with one choice – take manage of their data and use it for their improvement growing an understanding of WHY persons are buying their media and the way it performs. Or let Google, Facebook, third party et al are available and grab their data and know not anything about why it’s being bought and the way much it’s being sold.