digital ad fraud - bia advisory services...2. 100% of impressions generated between 4a –5a local...
TRANSCRIPT
September 2018 / Page 0marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Digital Ad FraudImpact on Local Ad Spend
September 2018Augustine Fou, PhD.acfou [at] mktsci.com 212. 203 .7239
September 2018 / Page 1marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Agenda• What is Ad Fraud?
• How big is the problem?
• Why is it not caught?
• How it affects Local Markets
• What local publishers can do
• What marketers can do
September 2018 / Page 2marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
What is Ad Fraud?
September 2018 / Page 3marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
What is digital ad fraud ?
Ad Fraud = ad impressions caused by bots, not seen by humans
Impression Fraud(CPM) Fraud
(includes mobile display, video ads)
Click Fraud(CPC) Fraud
(includes mobile search ads)
September 2018 / Page 4marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
How bad guys commit ad fraud1. set up
FAKE SITES2. buy
FAKE TRAFFIC3. sell
FAKE ADS
September 2018 / Page 5marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Why is ad fraud bad?
Advertisers Publishers
Bad Guys
1/2
1/2Ads are not shown to humans, wasted
ad dollars
Ad revenue declines because dollars are stolen by bad guys.
Steal money using fake ads; siphon dollars out
of ecosystem.
September 2018 / Page 6marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
The most profitable criminal activity
2,500 - 4,100% returns
11% returns1% interest
digital ad fraud
stock marketbank interest
“where else can I get multi-thousands percent returns on my money? Right. Nowhere.”
September 2018 / Page 7marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Economics of botnets explained
Source: MIT Tech Review, May 2018
“distributed denial-of-service attacks using a network of 30,000 bots can generate around $26,000 a month. Spam advertising with 10,000 bots generates around $300,000 a month, and bank fraud with 30,000 bots can generate over $18 million per month. But the most profitable undertaking is click fraud, which generates well over $20 million a month of profit.”
Botnets can be used for a variety of things
September 2018 / Page 8marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
How Big is Ad Fraud?
September 2018 / Page 9marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Everyone has an opinion…
Ads fraud is “non-existent” – IAB Australia
“Ad fraud is $6.5 billion or 9% of
display ad spend”-- ANA/WhiteOps
“88% - 98% of clicks are generated by
bots”- Oxford Biochron
September 2018 / Page 10marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
(2015) Display ads …Increased CPM prices
by 800%Decreased impression
volume by 92%
Source: http://adexchanger.com/ad-exchange-news/6-months-after-fraud-cleanup-appnexus-shares-effect-on-its-exchange/
260 billion
20 billion
> $1.60
< 20 cents
September 2018 / Page 11marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
(2016) Video ads …
Source: Dec 2016 WhiteOps Discloses Methbot Research
“Methbot, steals $2 billion annualized;
and it avoided detection for years.”
1. Targeted video ad inventory$13 average CPM, 10X higher than display ads
2. Disguised as good publishersPretending to be good publishers to cover tracks
3. Simulated human actionsActively faked clicks, mouse movements, page scrolling
4. Obfuscated data center originsData center bots pretended to be from residential IP addresses
1 botnet eats 15% of video inventory
September 2018 / Page 12marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
(2017) mobile app install fraudSource: June 2017, Tune
average 20% fraud
100% fraud
> 50% fraud
September 2018 / Page 13marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Examples of fake sites/apps1221e236c3f8703.com62b70ac32d4614b.coma6f845e6c37b2833148.comda60995df247712.comd869381a42af33b.coma1b1ea8f418ca02ad4e.com1de10ecf04779.com2c0dad36bdb9eb859f0.coma6be07586bc4a7.comfe95a992e6afb.com42eed1a0d9c129.comda6fda11b2b0ba.comafa9bdfa63bf7.com739c49a8c68917.combaa2e174884c9c0460e.comd602196786e42d.com153105c2f9564.com8761f9f83613.com20a840a14a0ef7d6.com31a5610ce3a8a2.com5726303d87522d05.com3ac901bf5793b0fccff.comb014381c95cb.com2137dc12f9d8.com
06f09b1008ae993a5a.comfbfd396918c60838.com97ff623306ff4c26996.comb1f6fe5e3f0c3c8ba6.com23205523023daea6.com6068a17eed25.comb1fe8a95ae27823.comf4906b7c15ba.comeac0823ca94e3c07.com1f7de8569ea97f0614.com21c9a53484951.com24ad89fc2690ed9369.comefd3b86a5fbddda.com34c2f22e9503ace.com0926a687679d337e9d.com6a40194bef976cc.com33ae985c0ea917.com02aa19117f396e9.comf8260adbf8558d6.com9376ec23d50b1.compushedwebnews.coma0675c1160de6c6.com0f461325bf56c3e1b9.com850a54dbd2398a2.com
com.dxnxbgj.mkridqxviiqaogwcom.obugniljhe.fptvznqwhmcjmcom.bpo.ksuhpsdkgvbtlswcom.rlcznwgouw.vvtexstbfttngccom.kasbgf.sbzwtgpcbjexicom.bprlgbl.vbzecom.zka.lzhsoueilocom.alxsavx.mizzucnlbcom.jxknvk.lrwfdfirdzpswcom.tvwvqbt.wbshaguqycom.iwnxtpahcu.leyuehdwdbbcom.okf.rhvemtykfibzpxjcom.obpmirzste.ldsjpvcom.zmm.shmxvjxnsagnduicom.nqzwr.leusrmpmsqcom.rced.zcdsglptpdlwpucom.kerms.ehlsgnccom.cmia.iabhheltmcom.skggynmtx.tyyjnwpefvqtllcom.kgdtltnuv.hayvfhobcom.ztzsiqg.dyojlxdscxwscom.xlwuqe.ddrdhsuosbncom.rkrhmzee.wjcoznxucom.ebhzb.hbzvomzpcctovj
Fake sites Fake sites Fake apps
September 2018 / Page 14marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Ad fraud is at all-time highsThere’s $100B in digital ad spend to steal from, year after year
U.S. Digital Ad Spend($ billions)
Actuals Projected
Digital Ad Fraud($ billions)
($300B worldwide)
September 2018 / Page 15marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
How do we know it’s fraud / useless?
September 2018 / Page 16marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Chase: -99% reach, no impact
“JPMorgan had already decided last year to oversee its own programmatic buying operation.
Advertisements for JPMorgan Chase were appearing on about 400,000 websites a month. [But] only 12,000, or 3 percent, led to activity beyond an impression.
[Then, Chase] limited its display ads to about 5,000 websites. We haven’t seen any deterioration on our performance metrics,” Ms. Lemkau said.”
“99% reduction in ‘reach’ … Same Results.”
Source: NYTimes, March 29, 2017
(because it wasn’t real, human reach)
September 2018 / Page 17marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
P&G: cut $200M, no impact
“Once we got transparency, it illuminated what reality was,” said Mr. Pritchard. P&G then took matters into its owns hands and voted with its dollars, he said.”
“As we all chased the Holy Grail of digital, self-included, we were relinquishing too much control—blinded by shiny objects, overwhelmed by big data, and ceding power to algorithms,” Mr. Pritchard said.
Source: WSJ, March 2018
September 2018 / Page 18marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Gross Failures ofFraud Detection Tech
September 2018 / Page 19marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Bad guys easily avoid detectionBlocking of tags, altering measurement to avoid detection
Detection Tag Blocking — analytics tags/fraud detection tags are accidentally blocked or maliciously stripped out
“malicious code manipulated data to ensure that otherwise unviewable ads showed up in measurement systems as valid impressions, which resulted in payment being made for the ad.”
Source: Buzzfeed, March 2018
September 2018 / Page 20marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Despite fraud detection tech… Obvious fraud still gets through; analyze and fight fraud yourself
Launch Week 3 and beyondWeek 2
Initial baseline measurement
Measurement after first optimization
After turning off more “sites that cheat”
Obvious fraud still gets through
September 2018 / Page 21marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
After all flavors of“fraud filters”Obvious fraud still gets through; but we turned off manually early in the campaign
September 2018 / Page 22marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
They miss popunders/redirectsThese forms of fraud typically get by current fraud detection tech
Vendor openly selling 125 billion page redirects (pageviews) per month, at low CPMs)
a.k.a. “zero-click” “pop-under” “forced-view” “auto-nav”
September 2018 / Page 23marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
They miss mobile app fraud“fraud sites’ traffic comes from apps that load hidden webpages”
Openly selling on LinkedIn
September 2018 / Page 24marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Overall ad fraud > bot %Fraud detection looks for IVT(bots); may miss other forms of fraud
% bot + % site + % mobile fraud
% overall fraud = 23%, not 5%
5% 11% 7%
September 2018 / Page 25marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
How Ad Fraud Will Impact Local Ad Spend
September 2018 / Page 26marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Local Media Ad SpendLocal media ad spend to exceed $150 billion in 2018
The digital portion of local ad spend will reach $76 billion in 2021; most of it will be bought and placed programmatically.
Source: BIA/Kelsey
September 2018 / Page 27marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Budgets go to geotargeted ads
Not Normal – in both campaigns1. 100% mobile apps; 100% Android; same top 15 apps in both markets2. 100% of impressions generated between 4a – 5a local time3. 100% fake devices; 15 unique devices generated top 95% impressions4. 100% data center traffic, randomized through residential proxies
September 2018 / Page 28marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
How? Mobile emulatorsMobile emulators can simulate actions pass fake sensor data (GPS)
install, launch, interact
September 2018 / Page 29marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Programmatic GeotargetingStudy on programmatic geotargeting shows low quality
Sources: BIA/Kelsey, Marketing Science Consulting Group, September 2017
September 2018 / Page 30marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Local Media – real humansIn contrast, there is a lot more “confirmed humans” in local media
Seattle, WA
San Francisco, CA
Los Angeles, CA
Las Vegas, NV
Bozeman, MT
Denver, CO
Dallas, TX
Houston, TX
Atlanta, GA
Chicago, IL
Miami, FL
Washington, DC
New York, NY
Philadelphia, PA
Boston, MA
Bangor, ME
September 2018 / Page 31marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Consumers Trust Local MediaConsumers who live there trust local media, distrust “any website”
Source: http://www.newsmediaworks.com.au/adtrust/
September 2018 / Page 32marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Local Media vs GeotargetingValue of local media significantly outweighs programmatic
Local Media Geotargeted Programmatic
AudienceLocal human residents
Device-reported location• Fake devices• Fake location• Fake users (bots)
Trust Consumers trust Consumers DIStrust
Fraud Risk Low High
ROI High Low
September 2018 / Page 33marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Ad fraud in OTT?Of course, where there is ad spend, there will be con artists
Real OTT Fake OTT
Real humans watching content
Fake devices generating fake ad inventory
Real ESPN inventory, authenticated devices
Fake Roku inventoryFake Hulu inventoryFake Netflix inventory
Fraud risk - low Fraud risk - high
September 2018 / Page 34marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
What Local Publishers should do …
September 2018 / Page 35marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Don’t “source traffic”Sourcing traffic (not digital marketing) exposes you to fraud
Choose Your “Traffic Quality Level”
“Valid traffic” costs more
September 2018 / Page 36marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Local TV websitesGreat consistency in the data over long periods of time
September 2018 / Page 37marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Local radio websitesGreat consistency in the data over long periods of time
September 2018 / Page 38marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Magazine websitesGreat consistency in the data over long periods of time
September 2018 / Page 39marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Filter data centers and botsOn-Site measurement, bots are still coming
In-Ad measurement, bots and data centers filtered
10% red
-7% (filter data
centers, bots)
3% red
September 2018 / Page 40marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
What Local Buyers should do …
September 2018 / Page 41marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Measure your campaigns
In-Ad(ad tag)
On-Site(embed code)
Measure
AdsMeasure
ArrivalsMeasure
Conversions
1743 156
A
more humans (blue)good publishers
On-Site(embed code)
social
nativevideo
display
affiliate
September 2018 / Page 42marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Progressively clean campaignsAnalyze and turn off the fraud yourself
Launch Week 3 and beyondWeek 2
Initial baseline measurement
Measurement after first optimization
After turning off more “sites that cheat”
September 2018 / Page 43marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Shift budgets to higher qualitySome channels have more humans, shift budget to these
• Blue means humans• Red means bots
Where to find this?Your own bots versus humans measurement, with detailed and complete supporting data.
What to look for?Be sure to measure for both bots and humans; rank order channels, sources, or sites by highest humans first.
What to do?Some channels have more humans (blue) and others have more bots (red); shift budgets during campaign to more human channels.
September 2018 / Page 44marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Display 4
2,036 humanshuman conversion rate
Compare actual outcomesSite Traffic Conversions
8,482 818
4,216 humans
5%human conversion rate
14,539 193
225 humans
9%human conversion rate
2,248 23
168 humans
5%human conversion rate
1,527 9
Display 3
Display 2
Display 1
Humans
40%
September 2018 / Page 45marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
“buyer beware”and
“you get what you pay for”
September 2018 / Page 46marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
About the Author
September 2018
Augustine Fou, PhD.
acfou [@] mktsci.com
212. 203 .7239
September 2018 / Page 47marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Dr. Augustine Fou – Independent Ad Fraud Researcher
2013
2014
Published slide decks and posts:http://www.slideshare.net/augustinefou/presentationshttps://www.linkedin.com/today/author/augustinefou
2016
2015
2017
September 2018 / Page 48marketing.scienceconsulting group, inc.
linkedin.com/in/augustinefou
Harvard Business Review
Excerpt:
Hunting the Bots
Fou, a prodigy who earned a Ph.D. from MIT at 23, belongs to the generation that witnessed the rise of digital marketers, having crafted his trade at American Express, one of the most successful American consumer brands, and at Omnicom, one of the largest global advertising agencies. Eventually stepping away from corporate life, Fou started his own practice, focusing on digital marketing fraud investigation.
Fou’s experiment proved that fake traffic is unproductive traffic. The fake visitors inflated the traffic statistics but contributed nothing to conversions, which stayed steady even after the traffic plummeted (bottom chart). Fake traffic is generated by “bad-guy bots.” A bot is computer code that runs automated tasks.