AdMonsters PubForum Montreal: The Live Blog

(While it’s fun to journey to the past… Why not sign up for the next PubForum, November 2017 in Nashville?)

The AdMonsters team has successfully made it to Montreal for the 42nd Publisher Forum–which means as much as any of us has misbehaved in the past, evidently it wasn’t incriminating enough for border control to raise an eyebrow. The PubForum attendees are here, we’re slowly remembering some of our high school French, and people are reconnecting with friends and colleagues. The assembled crew has been talking about the kinds of discussions they want to have while we’re here, and all are ready to dive into the business of digital media and stay there for the next few days.

The main room is filling up this morning at hotel Fairmont the Queen Elizabeth in downtown Montreal. If you weren’t able to be here with us this week for PubForum, we’ll be liveblogging from this room on Monday and Tuesday, Aug 14 and 15. Follow this space to keep up with keynotes and other full-group sessions, and check back on it on the AdMonsters site, where it’ll live in perpetuity for your and your team’s reference.

It’s shortly after 9 a.m., and AdMonsters Chairman Rob Beeler is about to take to the stage and welcome the crowd.

9:05: Rob warns us he’ll be dropping a lot of Game of Thrones references today, in light of the big episode last night–but no spoilers. (Thanks to GeoEdge for connecting the GoT fans to HBO here at the hotel.) One big topic we’re going to be exploring is… Bran Safety. (Collective groan from the audience. Thanks.) And the theme of this Forum, Rob says, is… “Sh** Just Got Realer.”

9:13: We normally thank our sponsors by clapping loudly at this point (Claps per Mille? Cost per Engaged Clapping?), but in the spirit of moving with the times, Rob tells us his 11-year old informed him dabbing would also be acceptable. (In other 2017 news, two of our sponsors have included fidget spinners in the attendee swag bag.) And since Twitter grew by no new users this quarter, Rob says your Tweet about #PubForum might make all the difference to that platform.

9:25: Rob is welcoming keynote speaker Anthony Katsur to the stage. Tony is a self-described “ad tech journeyman,” currently SVP of Platforms and Operations at TV network affiliate Nexstar Media Group–“a large media company no one’s ever heard of,” Tony explains. He’s made it to PubForum in spite of a ruptured Achilles tendon from a football accident.

9:28: Ad tech pervades all parts of our lives around the world, 24/7. There’s more to it than delivering ads. The reach, scale, precision and multi-channel pervasiveness of ad tech is greater than it’s ever been. What people in ad tech and ops do matters more than ever, and will continue to increase in influence.

9:30: We’re almost 25 years into the digital industry. Tony was on the scene very early. He says he’s been blown away by the innovation, which hasn’t stopped. But the launch of the ad exchange was an inflection point that really changed the business.

9:31: Header is the latest innovation, but it’s still new. How about group bidding? Integrated content management systems and ad serving–“the true beauty of native,” Tony says? Or the cookieless world, “the Holy Grail, not as ephemeral as a cookie?”

9:32: There have been a few inflection points–the launch of ad serving, ad exchanges, RTB. Header is another point. Sending IOs and working off spreadsheets are no longer the way we do business. Programmatic will surpass all traditional buying channels by 2020. (But don’t say “AI” right now–we’re not quite there yet.)

9:34: Programmatic is driving the growth in ad spend by device, which is “mind-boggling. We’ve been arriving, and now this is our time. We’re here today,” he says of digital.

9:35: Mobile spend is growing at 50% YOY. Video is driving programmatic growth. (Programmatic display has plateaued, to the surprise of some.) “The world is really our oyster, and many of us will be well employed into our retirement. … But my god, what have we done?” We’ve made the media models dysfunctional at the same time. TV, radio and print are easy by comparison. Digital is hard and programmatic added tremendous complexity while it added growth. “With mystery there’s market”–hidden fees, disparate technologies, fraud and other risks that don’t really occur the same way in other channels.

9:39: There’s a vendor imbalance, and the model doesn’t feel like it serves the marketer/media company relationship, or the user experience. Media companies can feel barraged.

9:40 Nexstar looked into the cost of the complexity of the supply chain, and found adding another vendor (whatever it may be) on the node compounds the complexity “to the nth degree.” Point solutions are often necessary, but “dear god, get acquired sometime soon, because I can’t deal with 18 companies for one transaction.” For each partner, add about two full-time employees . You need someone to manage partnerships and drive monetization. The break-even revenue point needs to be $1M per partner annually or greater. Every time you bring on a new partner, there’s lost time for employee training. The opportunity cost per lost time for training is about 25% (that’s per vendor). That’s the cost to media companies.

9:43: There are hard costs to run an ad tech company too. About 60% of a public ad tech company’s OPEX goes into R&D and people costs, or 67% for a private company. You see data servers and cloud computing accounting for about 10% of revenue  on average. That’s 75% of revenue going to those costs. It’s not sneaker parties and foosball tables. Those perks count for less than 1% of OPEX. The money goes to tech and people.

9:46: Ad tech is often compared to the financial industry. Tony’s not buying it. Ad tech is expensive because it happens 27/7. The financial exchanges doesn’t work nights and weekends. This is where the heavy cloud services, fiber networks and engineering skills come into play.

9:48: You have your agreements with vendors, and then you have vendors’ profitability economics. Tech companies have historically taken “innovative” measures (ghost bidding, hidden fees, arbitrage, first-price auctions) to be profitable. It’s not transparent, but it’s their fight for survival.

9:50: The magic number from a take rate perspective is 18%, from Tony’s and Nexstar’s research and calculations. That’s the break-even point, more or less, and it’s not what media companies want to hear. It costs money and takes time to automate your way out. This 18% figure might change over time, but it’s what’s necessary right now to be sustainable. You might want to take this knowledge into your negotiations with partners.

9:52: The structuring of partnerships matters. First, transparency is important… and be careful what you wish for. In Tony’s intel, publishers give the most negative feedback to/about ad tech companies. That’s because they’ve been backed into a corner. Even Google doesn’t want to disclose fee structures around EBDA. But if you’re not transparent about this stuff with your partners, Google is going to take your market share. It’s not part of a duopoly–it’s a monolith. Getting your SSP down to 6% is an unsustainable option that will help sink that partner and leave you dealing with Google all the time.

9:55: “Buy the tech, not the check.” Some pubs love the tech and value prop, but want to see the money. But you need to look at that relationship. Tony has heard many times, “I want to work with you, but my other partner doesn’t want to.” Well, you, the pub, are the customer. You can tell that other partner to get out of here if you really want to work with who you want to work with.

9:57: Pubs need to audit their own supply. As a vendor, Tony was in favor of “the Pepsi challenge.” If your partner economics sounds too good to be true, audit it.

9:58: So where’s that put you? Three options, first two are not sustainable for most. One is build: custom/bespoke solutions. If you’re doing that, you know the true cost of ad tech. But it takes time to come to market, and it’s not your core competency, probably. Second is buy: But there are excessive valuations and R&D costs. Again, it’s not your core competency. Third is partner. There are foundation options to give you a framework, rich APIs you can plug into. Header bidding has changed the game, and server-to-server is key. If your partner is selling you S2S on match rate, they’re not at the level yet.

10:00: Nexstar has chosen a single lead-tech and service partner. If you’ve remodeled your entire house, you typically work with one general contractor who can bring in the electrician, plumber, etc. You can work with your main service provider on how they outsource and contract. That allows the media to focus on content. Establish a sustainable, transparent rate.

10:02: Platforms do exist. It requires elbow grease and a road map for partners.

10:04: Rob wants to know the questions a media company needs to ask partners in terms of financials, etc. Tony says it’s not necessarily financials. It’s in the data–audit logs are major, and he’s suspicious if the partner does not provide them. It’s also about their willingness to help you onboard other partners and be transparent about the process. And everyone needs to be open about the economic costs of working with the additional partners they want to work with.

10:07: Time for an OpenX sponsor session. Tim Sheets (VP, Monetization) and Steve So (Director of Yield) are taking the stage to talk about keeping up with containers. Tim says OpenX has integrated their tag management container with a great many publishers and have come away with some insights.

10:08: OpenX looked at the count of number of header bidding partners by publisher. 15% had eight header partners, which demonstrates there’s a need for containers to reduce dev work, improve page speed, adhere all bidders to the same timeouts, provide unified reporting, and manage ad quality and the risk of fraud.

10:11: Do you want to go open-source? It might take a lot of dev work and time. But you maintain control. A full-service provider takes the work off your hands, but you sacrifice some control around timing and their existing relationships with other partners. Rob asks the room whether they agree with that. No disagreement that Prebid can be hard.

10:13: A client-side container is more like a tag management system. Server-side containers reduce the number of total calls and can improve page speed.

10:15: Open source client side sequencing: Each exchange takes less time to run their auction the farther you go down the queue of bidders. That doesn’t seem like it’s maximizing revenue from all partners, and it’s worth monitoring if you go down this route. The browser is limited in its ability to track ad tech code. Dictate your order of calls.

10:17: Impact on demand side: If the pub is working with 10 exchanges for four ad units on the page, that’s 40 bid requests. Most OpenX buy-side partners have QPS limits; they can’t handle unlimited volume.

10:18: Rob wants to know whether this is why Rubicon bought nToggle. Tim says it is.

10:20: Sync ad load with content load. Look at the order of your partners and the sequence of their calls, and maybe randomize the order to test. Make sure all demand partners compete on either gross bids or net bids. Optimization techniques like smart flooring can be challenging across multiple demand partners. Monitor discrepancies–pass in all bids from all partners and discount partners with high discrepancies.

10:24: Rob wants to know where real-time guaranteed fits with this. Tim says RTG is not possible without header bidding. It’s a manual process, but it’s in market, and pubs should talk with their partners to see whether it will work for them.

11:55: The first round of attendee breakouts has just wrapped–can’t tell you what happened behind closed doors unless you’d registered for PubForum, so if the FOMO is too much for you, better sign up for the next one in November in Nashville. Anyway: We’re back in the main room for Tout’s sponsor session on “Accessing Quality Audience Beyond Your Own Property.” Daniel Ahiakpor, VP of Strategic Partnerships at Tout, and Shira Brown, Director of Partnerships, Video, at Billboard and Hollywood Reporter, are on the stage. Daniel explains that when Tout looks at accessing audience beyond O&Os, they’re often thinking about video inventory. Billboard and Tout have been partners since Nov. 2016, serving quality video and driving audience.

12:01: Shira says Hollywood Reporter had a great brand, but didn’t have much video scale in November, and wanted to grow beyond owned and operated sites. They worked with Tout to syndicate video across their network. As the partnership has expanded, they started to use Tout’s tech on their O&Os as well. This required close work with the Billboard/Hollywood Reporter product team on customizations.

12:05 Billboard had acquired Spin Media, which had several sites but no video content. They also had little bandwidth to scan Billboard/HR sites for video and place them on Spin sites. Tout has worked to solve those problems.

12:06: In one example, Billboard had done several videos with DJ Khaled, matched contextually with articles via Tout’s publisher network, that generated huge numbers of click-to-play views on mobile.

12:09: Shira says the growth in syndication views has been “a pleasant surprise” that has allowed Billboard/HR to grow their syndication business. It’s opened up incremental revenue opportunities.

1:24: We’re back from lunch and it’s time for the State of Ad Ops: The results of last night’s work groups, when each of the groups came up with an idea for an ad tech startup. We need to spin through these quickly, so let’s see how fast I can type as we flip through the slides… First up is MediaMize–a mediation team between content and sales to handle custom campaigns.

1:26: Beacon, measurement for custom content and video. There’s no way to track real-time events to track on top of video, is the problem.

1:27: Dog Catcher (in the logo, the dog should look like an ad… somehow). Catches bad creative by filtering creative in real time via machine learning.

1:28: YG13, solving the problem of data aggregation. Human Exchange would be “the TaskRabbit for ad ops and sales ops,” turning around trafficking quickly and providing reporting.

1:29: The Ring, One SDK to Rule them All, for app publishers. SDK is expensive (no standards) has bad UX.

1:30: Consortium of Ad Phishing Extermination (C.A.P.E.), fighting for ad justice and financial fairness. C.A.P.E. would be a nonprofit group that enforces standards and holds vendors and certifiers accountable–the problem being that no one is nonprofit now, including certification groups.

1:31: Data Cop, for enforcing data standards as well.

1:31: Adbar, “Where Everyone Knows Your Pain,” because your friends just don’t understand your problems. Which is what AdMonsters is for, right? Or it could be an addition to AdMonsters–partnering you with the exact person you need to talk to in that moment over your beverage of choice. RealIntent aggregates data on user intent and decides how likely the user is to generate quality video views.

1:33: Blood Oath–ensuring ad quality by imposing penalties via “the cabal.” How? Well, you start with lawyers. You get into the contracts an impose penalties for bad ad quality. Pubs are about to start paying the real price, like Google removing all the ads on your site, or Facebook diminishing or entirely blocking your reach. Pubs are losing traffic and there needs to be some other solution.

1:36: AdCheck, “the OPEC of Pubs.” Solving third-party discrepancies by standardizing pricing and measurement.

1:37: “So basically we’re all going to work for the same company,” Rob says. Tony Katsur has been charged with picking a winner. Blood Oath seems to be solving a real problem, he says, but it’s kind of the world of IAB. AdCheck is compelling, he says, and could be fleshed out more. There’s room for an intermediary to help with cash flow and he’s heard the idea floating around the VC community. RealIntent seems interesting, but kind of like what Tout is doing. Data Cop, unfortunately, doesn’t stand a chance, he says.

1:40: We’ll have to wait for Tony to declare a clear winner, but we’ll revisit it later today.

1:41: Time for FreeWheel’s sponsor session on OTT. We’re here with Katie Back, VP, Enterprise Sales, and Mike Lawlor, SVP, Client Services. Premium video publishers need a lot of customization in ad campaigns, especially because of direct sold deals.

1:46: Cord-cutting is accelerating, but user appetite is very high. OTTs are proliferating. Video digital video ad spend is increasing, and OTT is a prime place for growth.

1:47: Desktop viewing has decreased, being cannibalized, while OTT is seeing enormous growth. Engagement is not different between SVOD and AVOD.

1:49: Device penetration among OTTs in the US is heavily swayed toward attached devices. They came early; gaming device makers realized slightly they had similar OTT capabilities.

1:52: Globally, we’re seeing declining linear ratings as device options proliferate. But people are not abandoning the content–the viewing is not going toward YouTube or other non-premium content. Also, OTT viewing peaks in prime time. It tracks very similarly to linear. They’re not the same viewers watching linear, they’re different audiences watching in the same manner, and they’re watching long-form content, including sports and simulcasted video content. They’re also younger (31 on average for OTT vs 54 for linear) and have more income.

1:56: 98% of ads played on OTT are completed. It’s hard to channel surf during an ad break, among other things. Hulu and Kantar Milward Brown found OTT viewers also had high brand favorability versus desktop and mobile.

1:58: There are challenges. Measurement is hard and not standardized. This limits buyers’ investment. Creative specs and endpoint requirements are also complex–VPAID doesn’t work, which is limiting. There’s also a lack of sales channel clarity and buyer awareness. Publishers have to be very focused on endpoints. They need to be creative about measurement to prove the value of their audiences. They try to make sure sales teams are equipped with this knowledge to make a compelling story.

2:02: Rob says there are a lot of good things about ad stitching except for the fact that it’s stitching. Mike says it’s good for sports, for example, because it’s live. But it does create problems with targeting, and there’s no universal answer to it yet.

2:06: Publishers are encouraged to be everywhere, on every device, but in video it’s important to make sure you’re invested, because being everywhere creates a lot of problems to solve.

3:40: The second round of attendee breakouts for the day is done, and we’re onto the Digital Media Leadership Awards. We’re recognizing Joe Barone (Managing Partner, Brand Safety Americas at GroupM), Bryan Moffett (COO, National Public Media) and Sandra Baez (SVP, Advertising Solutions, Univision) for their contributions to the ad ops and digital media community.

3:43: Rob says Sandra is the person whose name has come up very, often for many years, as an expert in such-and-such a niche. Rob says Joe, as an agency person, has always been candid and open in conversations with publishers, which helps ops feel like it’s not in its own bubble. He’s also been a great resource for pointing out when Rob’s assumptions have been wrong, according to Rob. Rob says Bryan has been willing to lead sessions, particularly around audio in recent years. And in his ascension to the C-suite, Bryan is one of those people who’s so good at ops he’s managed to get out of ops.

3:48: Talking backgrounds of the DMLAs. Sandra was in law school, doing an internship with an environmental lawyer, and got a temp job with a publisher for a change of scenery. The pub asked her to figure out what was going on in digital, where she ran credit checks on buyers and found a lot of buyers didn’t have any money–and that the pub was selling impressions they didn’t have. Those experiences committed her to managing yield and working with advertisers who were present, not theoretical.

3:52: Joe says he’s in favor for creating your own job title. When he got into solving discrepancies, that meant putting them in a drawer until the problematic company went out of business. His focus has evolved to address a number of agency pain points.

3:53: Bryan started in editorial and became interested in the business side. NPR’s digital footprint was very small about 12 years ago. He found Google had a program called Google Grants for promoting nonprofits. He was up to $250K per month when Google called and cut them off. In that time, they generated a massive audience and worked on leveraging it for sponsorships. The business has evolved from there.

3:56: Joe is interested in publisher sourcing disclosure agreements. There’s a crisis in confidence between agencies and clients, but there’s an understanding agencies can find them quality audiences. When you get registered with TAG, you have to disclose where your audience comes from, says Rob, so this will remain a meaningful focus.

3:59: Sandra is interested in discrepancies right now. When revenue decreases not because you have fewer eyeballs but because advertisers are trying to find value across all their publisher clients and hindering individual publishers’ ability to deliver the goods, that’s a problem. Your margins are constantly chipped away.

4:01: You need one measurement provider and do a billion dollars of business over that, says Joe, but TV happened decades ago. Now everyone wants to be the new Nielsen, including Nielsen. You end up with so many people chasing discrepancies. Finance people will kick back IOs for discrepancies of less than $2. So is the problem getting worse in addition to just not getting better? That’s Joe’s question for Sandra. Sandra says it’s worse because of viewability metrics, IVT and other factors that complicate measurement.

4:04: Rob asks Joe if he uses his own numbers when he buys a lottery ticket or lets the lottery generate the numbers. Joe says he makes sure he has a relationship with the lottery, makes sure the lottery is MRC certified, and uses his own numbers.

4:05: Bryan says Jonah Peretti noticed the business model at Huffington Post had little to do with its core product. That’s where Buzzfeed came from, fusing the business model and its product. Being a generalist pub like NPR, rather than a niche product, can be complicated and sometimes discouraging.

4:10: Sandra says a lot of media buys want to do something that’s never been done before, and it’s intrusive. She needs to educate the buyer that a more subtle campaign can be more valuable, and she says we need to think about how we define engagement.

4:11: ANA has come out and endorsed TrustX, which Joe says is the only news he picked up on while he was on vacation. This is a controversial move–it’s not an open marketplace, as it only has a few dozen publishers. Joe says the ANA positioned the endorsement as suggesting using TrustX instead of other exchanges, and he’s curious about publishers’ perspective and experiences.


TUESDAY, 9:07 a.m.: We’re back in the main room on Tuesday morning following breakfast, some of us still dragging from whatever various groups of attendees and sponsors got up to last night. (I ran into a cluster of folks heading out to do karaoke around the time I was making my way up to my room, after midnight.) The coffee is kicking in. Rob is introducing AdMonsters Editorial Director Gavin Dunaway, who’s going to catch us up on “Industry Buzz.”

9:12: Let’s talk about AI, says Gavin. Right now, when we talk about AI, we’re mostly talking about machine learning. AI is more like a benign child than a robot gone awry. It’s just not smart enough to go awry yet.

9:14: Machine learning is real–it’s not like what some in the industry were touting a few years ago. They weren’t there yet. It’s now possible to use machine learning toward content and ad personalization, multivariate creative testing and other methods. Imagine getting away from manual reporting and leaning on a machine friend. That’s what AI is, not the robot apocalypse.

9:16: Facebook is an ad network–kind of a step backwards. It’s a black box. Hopefully Trade Desk news means mobile programmatic is resurgent.

9:17: Are we excited about hearing about blockchain? Sure. Do we really know what it is? It’s complex and boring, and hard to explain. Is blockchain a solution in search of a problem? Gavin has heard this question in a lot of ad tech discussions. Kind of a leading question. In this case, blockchain has a real purpose for verification. The problem is scale. A lot of the major players need to buy in, and they haven’t. In the meantime, other solutions will come up trying to push blockchain out of the way.

9:18: Is blockchain going the way of header bidding? Hard to say. Header created almost instant revenue for publishers. Blockchain will take a while to rev up.

9:19: Six-second video ads are fine. You can get your message in in six seconds. Problem is getting advertisers to buy in and work with the creatives. What will it do for CPMs?

9:21: There’s been a lot of consolidation among DSPs. That space is feeling the pain, which is something we wondered might happen when header came up. Turns out spray and pray is not the best method with header. Will we see similar waves in SSPs? Possibly. But the end effect will be better for pubs–fewer partners, more clarity.

9:22: S2S is fast, but it doesn’t give you the same transparency as “traditional” header.

9:23: Cookie-matching is real in programmatic. You need a lot of partners, and in S2S you don’t have so many, so it’s hard to scale. This leads us into our Tuesday keynote, Ken Zachmann, Principal Advisor at Diverge Digital Media.

9:24: Here’s Ken. He explains he came up through Datalogix, helping publishers with data matching. Datalogix became Oracle, and then Ken went to V12. In pursuit of better data matching, he knew he needed a company that had emails. He’s been burned by cookies. He went out looking for a better way than using cookies as a proxy.

9:26: The old programmatic promise was about “the right people, the right place, the right time.” It sounded great. Then the data started coming back. Ken found campaigns were hitting entirely the wrong audiences. He was left explaining this to advertisers and trying to figure out with publishers what went wrong.

9:28: Problem is that cookies plus probabilistic equals the wrong person.Too much tech, too many Venn diagrams, in probabilistic. The cookie proxy wasn’t finding the right people and it was giving data a bad name. Talking about “data” made it harder for Ken to get meetings with business leaders.

9:29: The landscape was fragmented and Ken knew it. It was a mess. The line that we were “pretty sure” we were finding the right people is BS–a waste of advertising dollars.

9:30: The market is moving more toward identity and people-based approaches. Oreos have a shelf life. So do cookies. A household changes. You have four people and a lot of variables in play. You could have 19 cookies attached to four emails. And what if that email was from 1998? We see old data attached to individuals being used for targeting right now. That messes up the measurement post-campaign, makes it hard to decipher who you’ve reached and how the campaign has performed. And it’s not connected to mobile. We lose a ton of insights when you lean on cookies and people are spending 86% of their time on mobile in apps. This was a “come to Jesus moment.”

9:33: Identity-based targeting is a new day, ideally. Unfortunately, it makes no sense. Explaining how it works sounds like a ton of jargon. A “unified experience and transparent measurement?” That explanation doesn’t even sound transparent!

9:34: There are problems with identity graphs hitting the right people. Well, the graph looks great and purports to connect to all kinds of devices and platforms. It makes sense on paper. But one piece is missing: Who is this ID, this person, driving these matches? We know a lot about her and what she consumes online. That sounds useful, right? We have all these data points on “she.” But Nielsen reports 11% of marketers have high confidence in the audience they’re targeting. Reach was really squeezed.

9:37: We’re asking the wrong questions. We need to know who “she” is, not her devices and habits. It’s hard to get a straight answer. Tech platforms will tell you about their algorithms and how they use deterministic and probabilistic matching. What does that mean, though?

9:38: Deterministic matching is tied to some truth set, like Facebook or an offline PII set. When a company mentions “deterministic,” you need to ask how they define that. Are your tech partners getting deterministic data from probabilistic partners? If so, you’re still going to have problems honing in on addressability. Saying “deterministic” can be sales pitch language, and it should be a red flag.

9:41: Probabilistic has come a long way. You have a lot more touch points against PII. But dig in if the scale looks too wide. You want to know what goes into a company’s probabilistic matching. Some will get much looser about precise targeting it if they want to deliver scale. But will you see value in broad scale and a high margin of error in reaching who you want?

9:44: You need a PII truth set to know you’re reaching a person, not just an ID point. The solution is to start with specific individuals and not cookie data. The old approach was technology connecting devices that were all speaking different languages. You don’t make a solid connection. Ideally, the common language is the specific individual.

9:45: How’s this work with privacy? You start with subscriber files, user logins/registration data. Work with data providers to understand your users and clean up your files. Is this the right email they’re using now, if you’re using email to reach people? A good provider will have tiers of emails. You see a fall-off in emails. Sometimes people will use the same email address to log in years after they’ve created an account on such and such a site.

9:48: We’re not talking about DMPs. Their core competency is merging and blending data and building audiences, not verifying PII.

9:50: Publishers have great data. Data companies can provide a good backbone.

9:51: Mobile numbers can be a very strong persistent ID (this is usually what “mobile-first” companies are using), tied to email, and verified against everything else. Problem is the telecoms have those numbers, so it’ll be interesting how that works in the marketplace.

9:52: Why’s this matter to pubs? Well, the duopoly’s market share. Their domination will only get worse. They have all of this login data and PII, which is a massive advantage in people-based marketing. Publishers need to have a very sound data and onboarding strategy to retain their market share.

9:53: So how do you build a data strategy? Ask partners whether the ID graph is backed by a truth set. “Traffic” and “unique views” don’t cut it; they’re imprecise and don’t signify unique individuals (or how robust their graph is).

9:54: As what the truth set is and how it’s verified. Are they grading their own homework or verifying it against an external source?

9:55: What’s the recency of the PII? This is a hot-button question now. People want less than 30-day recency for desktop, and they want it nearly to-the-minute in mobile.

9:56: Who are my readers and buyers? We’re not talking about traffic or proxies, but individuals. And is your user ID tracked and matched with individuals against PII in a privacy compliant way? Most above-board companies are, but do your do diligence that your partners have the proper opt-ins and outs.

9:58: The LiveRamps and Neustars of the world are getting there. Ken thought companies like that would lead the way. They’re responding to the market trends, but tracing back to “what they think” is an individual is not enough. They would need to append email to it.

10:01: Holding companies are building data-centric identity solutions. CRM data can be one point in this method. They’re trying to become the onboarding truth set, and they can be helpful in replacing bad email matches. The holding companies have been able to bring in-house all of these companies in LUMAscape that promised they could do this stuff.

10:06: The individual should be the foundation, and you build a graph off of that, branching out to different devices and other points. Companies that are successful at this start with people, combine in data points–think of it as a bicycle wheel, as opposed to the old method, which was like a tire with no spokes. For pubs, you get the ability to monetize your audience and support brands’ campaigns. For marketers, it’s a more consistent experience and easier to analyze and optimize.

10:10: For publishers, you need to unify users down to single identities, optimize content and creative based on people instead of cookies, and drive relevant content to them.

10:15: Rob points out that as we automate, so will malvertisers. Alex Calic, CRO of The Media Trust, is coming up onto the stage to explain what TMT is doing to combat malware.

10:17: Advertising used to be about direct, face-to-face relationships, Alex says. We’ve gone from Mad Men to mad machines. The tech is seen as impersonal and cold. But automation is efficient. You get faster delivery of both good and bad content/ads. Technology is a great enabler, but a low barrier of entry creates chaos. Did you get the Google email telling pubs they were running bad ads on their sites, per surveys from the Coalition for Better Ads? Ironically Google has kind of enabled the delivery of those bad ads.

10:20: Pubs can use tech for quality control. Programmatic equals problematic. It’s not that programmatic is bad, it’s that there are so many nodes involved in delivering ads. Failures are plentiful. Now we have ad blocking, increased regulation, and damage to the publisher brand. Those factors impact monetization. Now we also have industry standards. But it’s time-consuming to get certified.

10:23: If you have your own ad policies, great. If not, work on them or talk with a security vendor.

10:24: Publishers need buy-in on ad quality from the highest level. Ops teams are constantly putting out fires. They need CISO, legal, privacy officer, IT and sales support to create and drive policies. You also need to connect with your upstream policies. If you see consistent problems from one partner, you can look into the source of those problems. It might be one third party that you can get the primary partner to turn off if you have a relationship.

10:26: Transparency goes both ways. Everyone’s relying on machines, and you need to have conversations with your partners. One phone call can solve some problems. You can use technology to create these relationships, too. What we need now is risk management, which relies on direct relationships, and an expectation that a company’s main point of contact will get this phone call.

10:33: Dave Marquand, Integral Ad Science VP, Publisher and Analytic Products, is onstage talking about viewability now. Pretty much everyone is seeing IOs with viewability or brand safety requirements right now. They’re legitimate goals, but publishers are not equipped to deal with them efficiently. Take Publicis’ requirements, for example. There’s no “brand safe” button at DFP. So how do you make sure you don’t underdeliver? You see huge amounts of waste to meet those guarantees.

10:35: IAS is automating this process, pushing data from the advertiser into the ad server to serve the right number of impressions, optimized for viewability or brand safety expectations. Solving the “buffering” problem saves impressions that can generate more revenue for pubs.

10:38: Dave has invited Business Insider SVP, Sales Operations and Client Services Marc Boswell onto the stage to talk about the publisher’s perspective. Marc says they’ve been seeing IVT and brand safety guarantees show up on the programmatic side, after seeing them on the direct side for years. This means BI’s client services team spends a lot of time looking at reports. Automation and optimization helps. The ongoing question is finding out why impressions are not viewability. There are ongoing problems with discrepancies.

10:41: Marc wants to make sure his site is as viewable as possible, but still needs to hold advertisers accountable. Advertisers may want to drop trackers that add weight. These conversations are getting easier because a lot of pubs are pushing back. But we still need more transparency around these issues.

10:45: The right rail and “above-the-fold” really aren’t a thing anymore, Marc says. People scroll right past them. BI needs to keep that in mind, along with site speed, while re-designing their sites.