SolutioniPinYou

Beijing,.

July,

2016DMPBackground

:Owned

MediaSystemSocial

MediaTagging/

SegmentedInsightsTouch

points/Data

DotBrand

ExposureInsight-Driven

ActivitiesConsumer

Voice3rd

Party

Media

SolutionDSPTradingDesk/

RTBPurchaseGMAProfilingTo

retain,

refine

and

reuse

1st,

2nd

and

access

3rd

partydataacross

mobile,

desktop

and

social

in

a

secure

and

robustmanner.And

also

have

a

tailor-made

platform

for

client’s

dataandinsights

ysis.Solutions

:Owned

MediaSystemSocial

MediaTagging/

SegmentedInsightsTouch

points/

Data

DotBrand

ExposureInsight-Driven

ActivitiesConsumer

Voice3rd

Party

Media

SolutionPurchaseGMAMediaBig

DataDataValue

EnablerProfilingModeling

&AlgorithmIPinYouBig

DataTransSocialGraphicData

Acquisition

&Enri

entMore

EffectiveMedia

BuyingBetter

EngagementPerformanceContinuouslyysis

&Advertising

ServiceOpen

IDDevice

ID/s

Device

BehaviorID/Phone

No./Member

IDUser

IDUniversal

Mc

IDGMASocial

Engagement3rd

Party

dataDevice

IDPhone

No.Open

ID??Open

IDUser

IDUser

IDPhone

No.Device

IDTaxonomyPhone

No.LocationOutputDataManagementWe

Create

Universal

McDonald’s

ID

toward

DMPIntegrationEnrient

Visualized

Insightysis

toolClient

data

management,

tracking,

andCustomer

insightsServe

for

media

buy4.

Demonstrate

reporting

&

output

format,

which

is

easy

to

read

and

understandysis,

which

isPropose

the

way

of

working

&

procedure

to

adapt

modelingcompleted

by

3rd

party

vendorOverall

integration

and

output

frameworkPart

1iPinYou’s

Capability

for

DepictingConsumerPart

2User

Ability

-iPinYou

DMPPart

3Business

RequirementPart

4Technical

RequirementAtta

ent[

Content

]FundamentaliPinYou’s

Capability

forDepictingConsumerWe

believe

following

core

competence

are

required

for

DMPpartnerhas

integrated

&processed

1st

party’smultiple

data

sources.Whether

the

partnerhas

built

a

scientifictaxonomy

andmodeling

algorithmthat

leverage

big

data

and

actual n

datacan

seamlessintegrated

with

3rdparties’

data,

DSP,client’s

othersystem

and

DMP.has

a

proveninfrastructure

andtechnology

thatprocess/clean/refineterabytes

of

data.has

responsive

teamon

the

ground

andexperienced

tosupport

customizedneeds.has

experienceingeneratingvisualizedmeaningful

insightsTotal

management

ofbrandDaily

website

coverageGuests

sDaily

Solute

blogsDaily

exposurechanceDevice

IDServed

DMP

ProjectWe

have

lots

of

experience

in

integrating

with

all

sorts

of

1stparty

&

3rd

party

dataThe

1st

party

data Data

from

Collaborative

partnersComple y

connected:

activates

multi-dimensionalonline/offline

dataiPinYou

Dhina

Mobile

DataAsiaInfoDhina

UnionPay

Data720

millMobile

ID860

mills6000Audiences22millDailyUpdates18millBidded

Imps250

KQPS450

+APPAudiences830

millMobileUsers500k

AppUsageBehavior5

TUser

BehavioralData-

Daily400+UserBehavior900K+Bases1050UserAudiences4

TUserBehavioralData-

Daily400

millMobileUsers210

millBroadbandUsers900

millUnionpayCard

Users45millUnionpayAcceptedStores60millDailyTransactionsGMASocialEngagement3rd

partymediasolutionDMPCross

Channel

IntegrationUniversal

Mc

ID3rd

ePnartty

Data

EnriMcDonaldTouchPointiPinYouself-own

bigdata

iPinYou

Data

PartnersiPinYou

integrates

all

the

data

of

McDonald,

further

enrichingconsumer’s

portrait

by

iPinYou’s

big

data個(gè)人應(yīng)用商品搜索IP地址消費(fèi)設(shè)備類型職業(yè)社交信息記錄LBS位置消費(fèi)地點(diǎn)品牌/機(jī)型教育內(nèi)容偏好偏好商品WiFi熱點(diǎn)消費(fèi)金額使用時(shí)長(zhǎng)互動(dòng)參與性質(zhì)工作地消費(fèi)商戶硬件特性收入類型瀏覽品類居住地消費(fèi)次數(shù)網(wǎng)絡(luò)型態(tài)人生階段搜索瀏覽頻率常去場(chǎng)所消費(fèi)時(shí)間操作系統(tǒng)………………………………Plenty

iPinYou’

s

tag

systemUserInterestPurchaseTrendsDynamicPositionConsumptionEnvironmentIDdataAdded

with

the

big

data

of

mobile

and UnionPay

,

iPinYou

can

update

users’

characters

andunderstand

each

user’s

profile/interests

more

holisticallyOnline/OfflineIntegration

of

Dataforming

6

Major

UserDimensionsAudienceAttributeDAAT

is

our

patented

taxonomy

system

that

proves

effectsfor

hundreds

of

clientInformation

classifiedand

filedWeight

assessment

and

updaabelAll

browsing

data

will

befiled

after

being

classifiedso

as

to

do

furtherresearchBrowsing

FeatureBrowsing

HistoryAd

InteractionTimeGeo…….Normal

browsingAd

interactionWebsitecommunity

actionWebsite

registerhighlowMediumlowSpecialbrowsinghighMediumDynamic

assessmentSearch

Behavior（由于涉及

技術(shù)，部分內(nèi)容不宜詳細(xì)展示，敬請(qǐng)諒解）Update

andsaveDemographicsSpecific

visit

actionIn-market

IntentLong-term

InterestsAdvertising

according

toconfidence

value

of

label

ondifferent s.

Ad

won’tbe

deliver

if

the

label

weightless

than

confidence

value.label

weight

3DynamicadjustmentConfidence

value

1.0label

weight

2.5label

weight

1.5label

weight

0.5TAlabel：“Demographics”:

present

behavior“Long-term

Interests”

long-timeattention“In-market

intent”:

short-termattention

or

purchase

intensionWhat

can

enrich

for

McDonald

’consumer

portrait

by

iPinYou’scooperating

3rd

party個(gè)人關(guān)注個(gè)人應(yīng)用專區(qū)活躍APP……傾向人口屬性位置信息設(shè)備屬性財(cái)經(jīng)金融/數(shù)碼電腦/辦公家用電器個(gè)護(hù)化妝……月收入職業(yè)受教育程度……常去商圈地域分布常去場(chǎng)所……品牌

上市價(jià)機(jī)型設(shè)備品類屏幕尺寸……中國(guó)移動(dòng)大數(shù)據(jù)亞信大數(shù)據(jù)銀聯(lián)智惠大數(shù)據(jù)接入點(diǎn)屬性個(gè)人關(guān)注屬性(家庭、公司、公共wifi)規(guī)模(規(guī)模人口數(shù))地理位置……個(gè)人

體系個(gè)人關(guān)注(基于 的內(nèi)容特征，及人對(duì)各類站點(diǎn)的瀏覽 偏好)購(gòu)物傾向(基于人在各的 行為，及各類商品的類別歸屬)搜索(在各搜索引擎及站點(diǎn)的搜索行為及搜索的)……休閑生活服務(wù)購(gòu)物綜合其他行業(yè)企業(yè)……消費(fèi)消費(fèi)地點(diǎn)消費(fèi)金額消費(fèi)偏好消費(fèi)次數(shù)消費(fèi)消費(fèi)城市消費(fèi)總額 統(tǒng)計(jì)餐飲月消費(fèi)次數(shù)統(tǒng)計(jì)無(wú)業(yè)商圈月金額 統(tǒng)計(jì)住宿消費(fèi)次數(shù)網(wǎng)購(gòu)結(jié)婚地鐵站出行地點(diǎn)高額消費(fèi)消費(fèi)城市金額統(tǒng)計(jì)超市零售出行地點(diǎn)次數(shù)分布消費(fèi)筆數(shù)在本市的學(xué)生海外旅游每月消費(fèi)金額統(tǒng)計(jì)運(yùn)動(dòng)……月收入預(yù)估…………消費(fèi)檔次……母嬰……Build

up

a

holistic

customer

profilesDeviceIDUser

IDPhone

No.LocationCreate

Universal

McDonald ID

via

identifying

different

data

sourceUniversal

Mc

IDGMAOpenIDIdentity

a

customer

via

different

data

sources，create

Unique

IDDeviceIDOpen

IDUserIDPhone

No.GMAGMAloginloginloginbookingbookingbookingIntegrate

3rd

party

data

to

enrich

customer

profilesThe

3rd

party

data

supplement

that

McDonald

has

Integrated

,including

input/outputtwo

modes

of

data

interaction.DMPBaidu/Tencent/Ali/weibo/operation……Labels

of

3rd

party

data

:Fundamental

information

,

DynamicPosition，Internet

data，purchasedata，department

store

data，O2OdataAPIInputDMPOutputVisualized

insightAudience

reportMedia

buyAdvanced

customer

profiles

based

on

Universal

IDMobile

3rd

party

ProfileGMA

ProfileAPPLoyalty

degreeProduct

interestsPeriodof

purchaseIntension

of

purchaseDeliver

areaSocial

Profile(online)PC

3rd

party

ProfilePCAudience

AttributeUser

InterestPurchase

TrendsLoyalty

degreeSocial

Profile(offline)OfflineLoyalty

degreePeriodof

purchaseWechat/

alipay/posPurchase

areaOnlineLoyalty

degreeProduct

interestsPeriodof

purchaseIntension

of

purchaseDeliver

areaWechat/

alipayDeviceidAudience

AttributeUser

InterestPurchase

TrendsOwned

MediaSystemSocial

MediaTouch

points

/

Data

DotBrand

ExposureConsumer

Voice3rd

PartyPurchaseGMAMediaBigDataProfilingIPinYouBig

DataTransSocialGraphicBehaviorOpen

IDDevice

ID/s：Phone

No.-

12123133212近3個(gè)月消費(fèi)頻次-9次近1個(gè)月消費(fèi)金額-

2393rd

Party：麥樂(lè)送/GMA：商品麥旋風(fēng)，甜筒常用地址-

寧區(qū)***消費(fèi)類別冰品消費(fèi)總金額-

458,2Socialfollow-麥當(dāng)勞PayWhole

picture

about

depicting

consumer

portrait-

支付interactive-關(guān)注”免費(fèi)喝雪碧活動(dòng)”P(pán)ortrait個(gè)人關(guān)注-飲食、旅游傾向-咖啡、甜品移動(dòng)端-休閑 、社交、人口統(tǒng)計(jì)統(tǒng)征屬性-女，25-35關(guān)注商品-服飾、飾品消費(fèi)能力-2000-5000/月Device

ID/Phone

No./

Member

ID

User

IDUniversal

Mc

IDUser

AbilityiPinYou

DMPOwned

MediaSystemSocial

MediaTagging/

Segmented

InsightsTouch

points

/

Data

DotBrand

ExposureConsumer

Voice3rd

Party

Media

SolutionPurchaseGMAMediaBigDataProfilingModeling

&

AlgorithmIPinYouBig

DataTransSocialGraphicBehaviorData

Integration

&

Enri

entTurn

Data

Into

Actionable

InsightUniversal

Mc

IDOpen

IDDevice

ID/sDevice

ID/Phone

No./Member

IDUser

IDTurn

data

into

actionable

insight

-

iPinYou

DMP,

DUOMINUO4

dimensions

of

standard

reporting

moduleReal-timefeedbackChannel

&conversionTAmanagementTAreporting@copyright品友互動(dòng)We

also

allow

very

customized

tagging

and

dollectionFeature

sample：

Create

all

customized

users

at

the

same

timeAudience

builder-customized

audience

segmentationiPinYou

develops

proven

lookalike

model

to

expandaudience

&

integrate

with

DSP

for

McDonald“l(fā)ook

alike”,

”ad

serving”Live

Demo

–

iPinYou

DMP,

DOUMINUOServiceBusiness

RequirementOwned

MediaSystemSocial

MediaTagging/

Segmented

InsightsTouch

points

/

Data

DotBrand

ExposureConsumer

Voice3rd

Party

Media

SolutionPurchaseGMAMediaBig

DataProfilingModeling

&

AlgorithmBig

DataTransSocialGraphicBehaviorInsight-Driven

ActivitiesData

Integration

&

Enri

entTurn

Data

Into

Actionable

InsightInsight

–

Driven

ActivitiesData

Value

EnablerUniversal

Mc

IDOpen

IDDevice

ID/

IPinYousDeviceID/Phone

No./Member

IDUser

IDData

Acquisition

&Enri

entMore

EffectiveMedia

BuyingBetter

EngagementPerformanceiPinYou

generates

visualized

insights

to

support

strategicthinkingServe

forMediaBuyInsight-drivenreport

&engagementBase

on

insights

tosupport

most

optimalmedia

buyingEffectively

driveengagement

&

businessinsight

relatedDMPUniversal

Mc

IDBusiness

Insight

Driven

–

jacking

up

APP

conversion

&stickinessitem

/

set

meal

mendationShow

product

that

purchase

recentlyAccording

to

therecent

product，

mend

““hot”

set

meal

and

product

that

purchase

recenpromote

purchase

further.Individualized

coupon

pushmendationmend1. According

toof

single

product

,set

meal

forward.2.mend

product

choosefrom“new

”

to

”

hot”n

mendationBy

offline

activities

,

improve

theparticipation

and

acknowledge

ofcustomers.By

using

of

data

of

GEO,

fix

thelocation

of

customer

，offer

”distanceof

store”

to

enrich

the

purchase.GMA熱點(diǎn)推薦離您最近

：550

米越秀區(qū)萬(wàn)達(dá)廣場(chǎng)店1樓推薦理由：加班多？全麥?zhǔn)称吩鰪?qiáng)免疫力！近期嘗嘗鮮添加Business

Insight

Driven

–

establish

the

linkage

people

toproductcoupon：20%off1.

By yzing

of

history

data

,

acquire

the

interests

ofXX

customersabout

foods.2.

Keep

the”

food

label

“of

customers,

naturallyrealize

the

reasonable mend

of

“food

.coupon”P(pán)rofile：User

Interest-food、travelPurchase

trends-coffee、dessert……Food

Profile：Main

Ingredient：beefcake：-whole

wheatAppreciation：-DoubleDiscount：Coupon

preferMain

Ingredients：whole

wheatbeefIngredients：gingelilettucetomatoappreciation：DoublebeefGenerate

customer

insight

and

serve

for

media

buyWhat

kind

of

onlineis

more

to

the

APP’s

ordering

consumers

liking

?Generate

customer

insight

and

serve

for

media

buyUniqueIDNameGenderPhone

No.Open

IDUser

IDDeviceIDPurchasetimeGoodsAmountStoreTakeawayAddressDownloadedAppLocationPhoneModelInterestSpending******************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************Transaction

DataOnline&Offline

DataKFC?2.

What

kind

of do

thecustomers

that

order

food

at

nightwatch?4.

What

kind

of

star

do

the

loyalcustomers

like?Insight:1.Among

the

customers

of

McDonald,

3.

Which

type

of

customers

usuallywhich

kind

of

people

coincide

with work

overtime?Generate

customer

insight

and

serve

for

media

buyDeliver

addressAiming

at

discovering

and

identifying

ofcustomers

in

main

areas.Using

deliver

address

and

times

to

find

outmain

customersgroup

in

smallareas.Combining

with

mealtime

,purchaseamount

,

buildings

around

to

identify

thistype

of

customers

as

“

work

overtime

“group.Aiming

at

charactersof

thistype

ofcustomers

,

push”coffee

set

meal”

toimprove

the

order

efficiently.Generate

customer

insight

and

serve

for

businessHot

area

aboutdeliveryBased

on

the

geo

data，getsome

discovery

on

mapaboutthe

hot

area

that

McD

deliveryis

busy.Items

associationSales

trend

&

predictBased

on

theclick

data

on

麥樂(lè)送app/PC

site/ad，build

thepredictmodle,whichcouldpredict

sales

trend

aboutspecific

item,

and

compare

with

the

real

salsetrend

,shown

as

below.It

is

about

the

associationbetween

items

,for

examplesidesand

dessert.

Thisindicates

the

customer’sprefer.Generate

customer

insight

and

serve

for

media

buyRe ing

from

n/site/taxonomy

.ServiceTechnical

RequirementProven

hardware

infrastructure

to

support

DMPWe

have

sophisticated

DMP

software

stack

tomanage

multi-billion

dataDollectionT1

T2

T3

...

TnReportOLAPDashboardDSSGanglia、Nagiosysis

on

real

time

logMedia

strategyAudience

strategyIntegrationAPIWe

have

proven

Disaster

Recovery

systemScope:

the

disaster

recovery

of

the

DMP

is

only

focus

on

the

DMP

itself;

doesn’t

cover

the

data

producer/consumer

side

out

of

theDMP

service;and

data

and

system

would

be

sync

to

cloud(hadoop

is

necessary).Thehardware

require

less

thanIDC

server。Goal:

when

a

disaster

struck,

minimize

thehazard,

reduce

the

equity

loss,restore

the

businessoperation

in

theshortest

time

dependson

the

disaster

type

and

damageThreats:

the

disaster

recover

of

DMP

will

try

to

face

nature

disasters,

e.g.

floods,

hurricanes,

tornadoes,

earthquakes;

or

man

madedisasters,

e.g.,

hazardous

material

spills,

infrastructure

failure,

terrorism,

disastrous

IT

bugs,

etc.According

to

the

recovery

time

objective

(RTO)

and

recovery

point

objective

(RPO)

of

DMP

disaster

recovery,

the

disasters

could

be

classified

to

4types:Disaster

typeDescriptionCritical

Response

ActionsRTORPOReachabilitydisasterThe

whole

infrastructure

is

keep

running,

onlythenetwork

access

to

the

DMP

service

is

lost

due

to

ISPissues

or

network

attacks(e.g.

DDOS)Contact

the

ISP

or

IDC

to

resolve

thenetwork

problemDepends

onthe

SLA

withIDC0Ha

isastersSome

or

all

critical

devices

stop

working,

but

thereisno

data

loss,

e.g.

power

disruptionReboot

halted

devices

or

replacedwith

new

devices10minutes0IDC

disastersThe

whole

IDC

isdestroyed,

e.g.

earthquakeRestore

the

service

by

cloud

provider1hour4hoursCatastropheBoth

IDC

and

DMP

online

backup

are

destroyed,

e.g.the

IDC

and

online

backup

are

burnt

at

same

timeRestore

from

offline

backup1day1

weekDisaster

Response

ProcessNotificationReachability

disasters

and

Ha

isasters

will

be

discovered

by

DMP

monitor

systems.

Operation

team

will

be

notified

by

text

message

andautomaticallySigned

agreement

with

IDC

and

will

be

got

notified

for

IDC

disasters

or

catastrophesby

mobile

phone

andIn

worst

case,

if

the

IDC

crews

are

all

killed

or

unconscious

in

the

disaster,

DMP

team

willget

informationby

news

or

be

notified

by

DMP

consumersDamage

assessment:

it

will

include

but

not

limited

to:Type

of

the

disaster,

originof

the

emergency

or

disruption,

potentialfor

additional

disruptions

or

damage,

status

of

the

infrastructure,

itemsshouldbe

replaced,

list

of

systems

and

services

that

need

to

be

restored,

sequence

of

restoration,

instructions

for

reporting

failures

to

the

team

leadsExecute

recovery

stepsaccording

to

the

disaster

typesFor

example

to

restorefrom

an

IDC

disaster,

the

steps

are:

make

sure

the

online

backup

are

still

alive;

allocate

devices

from

cloud

provider;

allocateand

boot

database

cluster;

allocate

and

boot

hadoop

cluster;

allocate

and

boot

cache

cluster;

allocate

and

boot

report

cluster;

allocateand

boot

datacollection

cluster;

allocate

and

boot

load

balancer

and

other

network

devicesReconstitutionMake

sure

the

DMP

service

is

restored

and

back

into

work

by

continuously

monitoringShutdown

the

contingency

systems

when

new

devices

or

a

new

DMPservice

are

established

once

againWe

use

the

most

advanced

cluster

to

secureefficient

data

storage數(shù)據(jù)倉(cāng)庫(kù)配置信息閾值信息用戶找回報(bào)表系統(tǒng)兩個(gè)集群雙備We

developed

complete

modules

toplish

DMPAd

server

ModuleBehavior

ofOfficial

SiteOthers/APIInternal

System

ModuleLog

Collection

ModuleMapModuleDleaningData

Integration

ModuleData

Modeling

ModuleThe

Multi-DimensionalReport

ModuleReal-time

Report

ModuleSupport

AdvertisingModuleWe

have

developed

proprietary

system

tomonitor

the

performanceiPinYou

has

proven

data

security

system

and

schemeTo

process

the

contents

of

the

database

according

to

the

actual

situation

ofUnilever,

and

it’sconvenient

for

the

administrator

tomanage.Distributed

store

data,

physically

dispersed

in

different

devices

and

keep

multiple

copies.

Andwithdual

fault-tolerant

system,

automaticbackup

to

avoid

any

data

losses

caused

by

physical

damage.10000+

auto-monitor

projects,

7*24

constantly

guard.Basedon

the

permissiondesignconcept

on

the

Role-Based

Access

Control,

RBAC

model,

build

upkernel

object

model

with

permission

control

system

to

managethe

data

with

reasonablepermissionsassignment.Object

model

includes:

users,

user

groups,

characters,

control

object,

visit

pattern

and

operation.AvoiddatalossPrevent

data

leakingService

Level

AgreementService

Update>99.9%Data

Stability>99.9999%Data

DestroyableData

will

be

removed

from

the

hard

drive

uponcustomer's

requestiPinYou

has

proven

data

security

system

and

schemeIDC

use

BGP.

If

apply

double（multiple）IDC

infrastructure

，fiber

connection

among

IDC

；if

IDC

&

cloud

solution,fiber/ connect

IDC

and

cloud

。[IDC]hardware:（綠盟…）or

firewall

（iptables）;[cloiud]

azure

and

Alibaba

Cloud

Computing

has

internal

network

security

solution.Such

as

Alibaba

DDOS

or

WAF.Middle

ware

update

in

time

and

WAF

above.NetworkMiddlewareStrict

IT

procedure

control

，include

data

security

,

authority

control

rule,

informationmanagement

standard,

remote

access

management

rule,changes

management

，which

areaccord

with

Sarbanes-Oxley

Act

404.）About

linux

system,

activate

iptables，and

upgrade

OSontime

；keep

one

eye

on

0day

leak，and

fix

it

in

time.OSInformationThe

speed

toresponse

client’s

requestserious

system

issue:

solve

within

3

hoursnon-serious

system

issue

:

solve

within

24hoursThe

way

/

process

to

solve

or

address

urgent

issue

/

requestreport

to

McDonald’s

related

departmentcheck

the

problems

7*24 and

delivering

solutionproblem

checking：identify problem

through

monitoring

&

warning

logensure

every

operation

can

be

restoredrecord

all

the

processing

steps

and

solutionback

up

before

eachoperationclarify

what

issue

andkeep

itin

record

forpreventing

if

thiscase

is

solved

within

30

rm

&

report

McDonald’s

related

department

about

this

issuethat

might

cause

what

results。iPinYou

quickly

send

experts

for

on-site

support

if

this

case

cannot

be

solved

onlineMulti-dimension

to

ensure

the

stability

of

DMPELK/GraphitestormdruidprestohiveMRsparkhadoopsurosqooprocketmqbid

server

clustertracking

sever

clusterredis/ssdbLVS/HAPROXYiPinYou

has

developed

most

robust

local

infrastructure

to

secure

99.99%

stabilityRobust

Infrastructure

supports

1

million

QPSWell

designed

extendable

cluster,

onlinebackup

to

prevent

anyonline

shortage50000+

monitoring

measurement

to

track

systemhealthinessUtilize

the

most

advanced

open

source

solution

to

provide

real

timebidding,ysis,

data

mining,

modeling

and

other

services

with

high

efficiency

and

great

stability.

Keep

contributing

to

open

sourcecommunity,

help

tokeep

these

technologies

keep

evolvingWe

have

set

up

standard

to

secure

datanormalizationBuild

up

standard

for

logs:TheDMP

logstandardneeds

to

be

accepted

by

bothparties.

Theintegration

willbe

built

on

thestandard.Each

d ompanies

and

3rd

party

advertising

companies

followsthe

same

log

rule

and

provides

data.

DMP

process

logs

in

thesame

format

and

then

solve

the

problems

caused

by

formatdifferences

from

different

channelsDMP

log

sampleWe

have

proven

algorithm

that

insures

the

bestd

leaningManage

the

complexity

multiple

connections

between

DMP

and

data

partners.Continuously

updated

anti-cheating

database

(with

cheating ,

IP

andURL

lists)Eliminate

the

suspicious

dataaccording

to

indicators

such

as

stay

time

etc.Dleaning

is

a

key

process

to

keep

the

high

quality

of

the

integrated

data.Service

operationA

whole advanced

service

operation

support

DMP

running

smoothly.Communication:communicate

with

3rd

party

in

the

below

ways：concall

meetingonline

by

some

socailwareholdan

on-the-spot

meetingSystem

changechange

definition

：system

upgrade

,bug

fix,

report

developmentchange

progress:Submit

requirements

and

acceptRequirements

developmentCheck

and

acceptpublishService

operationA

whole advanced

service

operation

support

DMP

running

smoothly.Help

desk400

support

Help

Desk

provides

an

efficient

way

to

provide

immediate

support

to

ourcustomers.

It’s

convenient

and

fast.

We’ve

found

that

the

vast

majority

of

our

customer’stechnical

problems

can

be

resolved

over

the

phone

or

with

an

online

chat.Marketing

operationBased

on

many

DSP

ad

deliver

project

experience

,

build

up

a

mature

marketinginsight

capability.And

PM

could

be

the

interface

to

accept

any

requirement

about

marketing.Client

ServiceProduct

DesignProject

LeaderMark

XieiPinYou

COO

&

FounderProjectOperationStrategyStrategy

ConsultantDerek

DaiDirectorStrategy

ConsultantXiaoni

ShaSenior

ManagerProject

OperationLay

LiuProject

DirectorP:roject

OperationZhipeng

ChouProject

ManagerProduct

DesignJibo

WangSenior

DMP

PMProduct

DesignFan

Liu

DMP

PMProject

OperationsTracy

LuGMTech

SupportTech

SupportDi

PaiAlgorithm

ArchitectsStrategy

ConsultantGuo

XuGMProject

ManagerAllan

Du

ProductVPYeXuClient

ServiceDeputy

GMClient

ServiceLu

ShanDirectorImplementation

Teamhand

experienceandCore

members

all

withextensive

experience.Real

time

and

in-depth

support.Tech

SupportZhao

LinAlgorithm

VPTech

SupportFan

LiuAlgorithm

DirectorQuotationITEMDetailsCost

(CNY)DMP

InfrastructureSoftware

infrastructureD

leaningLook

AlikeMcDonalds

Audience

ProfilingMcDonald's

Data

integration

andprocessingAudience

SegmentationAudience

ReportAudience

BuilderCN￥1,250,000.00Social

media

engagement

transactiondata

integrationIntegrationwith

Mcdonald's

GMAand

other

datasource

fortransaction

data

and

other

socialengagementCN￥200,000.00Integration

with

3rd

party

data

venderAPI

integration

with