Se lf-Se rvic e Da ta Ma na g e me nt fo r Ana lytic s Use rs a c - PowerPoint PPT Presentation

Se lf-Se rvic e Da ta Ma na g e me nt fo r Ana lytic s Use rs a c ro ss the E nte rprise

Unle a shing the Po te ntia l o f Gre a t Co mpa nie s Busine ss Solutions Pro ve n, inte g ra te d so lutio ns to de live r ra pid re turn-o n-inve stme nt > High-Impac t Busine ss Optimize d SAS Outc ome s Ar c hite c tur e s SAS Analytic E xpe r tise E nte rprise c la ss a rc hite c ture s fo r o ptima l SAS pe rfo rma nc e Se a so ne d da ta sc ie ntists to o n-pre mise a nd in-the -c lo ud pro vide stra te g y a nd c o nsulting se rvic e s

E xa mple o f Hig h I mpa c t Busine ss Outc o me s $ 3.5M 20 % $ 3M $ 1M 200 % ne t pro fit inc re a se fro m inc re a se in sa ve d b y c lo sing g a ps in sa ve d via ide ntifying inc re a se in c usto me r I VR flo w re de sig n c usto me r re te ntio n me mb e r c a re hig h risk c hurne rs spe nd

E xa mple o f Hig h I mpa c t Busine ss Outc o me s $ 6M 28 % 50 % 7M 360 re ve nue inc re a se via pe r da y I nte ra c tive uplift in time sa ving s fo r use rs de g re e re a l-time vie w ne xt b e st o ffe rs lo ya lty tra nsa c tio ns inc re me nta l sa le s wo rking with ra w da ta o f c usto me rs

Spe e d is the K e y to I nc re a sing Da ta a nd Ana lytic Va lue Value of Information Value of the Analytics Time T he va lue o f da ta de c re a se s o ve r time a nd o rg a niza tio ns ne e d to re a c t q uic kly to ma ximize its va lue thro ug h the use o f a na lytic s

Vo lume , Qua lity a nd Qua ntity: T he Ana lytic s Da ta Cha lle ng e E xisting Ana lytic s Pro c e sse s

Simplifie d T e c hno lo g y: Se lf Se rvic e Da ta Ma na g e me nt Adva nc e d Use r Sa nd- Boxe s Be ne fits • F a ste r a c c e ss to da ta • Simplify a c c e ss to e nte rprise da ta • Ab ility to lo a d te st da ta • Allo ws use rs to lo a d ne w da ta • E limina tio n o f duplic a te da ta • I T suppo rt a nd g o ve rna nc e • Re duc e d risk • Da ta q ua lity c he c ks a nd b a la nc e s Re so urc e b a la nc ing • Data L abs Database L ab Gr oup T able

Giving Use rs Ac c e ss to Mo re Da ta

Busine ss Ne e d fo r Ag ile Ana lytic s F le xibility vs. IT Proc e ss • Ana lyze q uic kly • T e st Ne w T he o rie s • Ne w Da ta • Do e s the ne w da ta pro vide a dditio na l insig ht? • Do e s the ne w insig ht c a use a c ha ng e in thinking o r dire c tio n? • T e st F a st • Wa s the the o ry rig ht? (Suc c e ss o r F a ilure ) • Pro duc tio nize wha t wo rks; disc a rd wha t do e sn’ t! • Ad d the ne w a pplic a tio n • Ad d the ne w d a ta • Or d e le te a nd mo ve o n! 9

Do n’ t Just Use Pro duc tio n Da ta – E vo lve I t 3r d Par ty Data • Ofte n re nte d , supplie r a nd / o r fo rma t c a n c ha ng e , va lue ne e d s va lid a tio n, o nly a pplie s to so me pro je c ts T e mpor ar y & Re se ar c h Data • E xplo ra to ry me tric s a nd a g g re g a te s, re q uire me nts no t fully d e fine d , sho rt live d , e a rly sta g e wo rk Pr e - Pr oduc tion Data & Pr ototype s • E ithe r o f the a b o ve c a n tra nsfo rm into this • Pro c e ss is d e fine d a nd pro ve n, the re is inte re st in fo rma lizing it, b ut it o nly e xists in the Da ta L a b 10

T e ra da ta Da ta L a b What is diffe r e nt fr om tr aditional sandboxing? • An a rc hite c ture de sig n tha t e na b le s g o ve rna nc e • I mpro ve d fle xib ility with wo rklo a d ma na g e me nt • Se lf-pro visio ning , ma na g e me nt a nd se rvic e c a pa b ilitie s a t the b usine ss unit le ve l

Diffe re nc e Be twe e n a Sa ndb o x & Da ta L a b s F unc tion Sa ndbox Da ta L a bs Runs Unsuppo rte d Pro duc tio n Apps Ye s No E nviro nme nt Ba c kup & Re c o ve ra b le No Ye s Spe e d o f Pro c e ssing & Prio rity No Ye s DBA Suppo rt (a g re e me nt) No Ye s Use rs c a n impa c t & impa c t o the r use rs Ye s No Spa c e is ne ve r c le a ne d up o r re c la ime d Ye s No Wo rk lo a d ma na g e me nt se t up No Ye s Use rs T ra ine d o n Optima l use No Ye s

T e ra da ta Da ta L a b Hie ra rc hy Da ta L a b hie ra rc hy to ma na g e use r g ro ups, spa c e , a nd wo rklo a d Database Data Labs Table L ab Gr oup Da ta b a se whe re the Wo rkspa c e s a llo c a te d fo r Da ta b a se ta b le to Wo rkspa c e a llo c a te d fo r a g ro up la b g ro up re side s o f use rs to c re a te the ir o wn da ta a na lysis sto re the da ta  No rma l T e ra da ta   la b s. Ca n b e fo r a sing le use r Use r c a n c re a te use r- da ta b a se o r X numb e r o f use rs ta b le a nd lo a d Gro ups c a n b e a rra ng e d b y  Da ta L a b s e xpire da ta de pa rtme nt o r pro je c t  Da ta L a b s a re a llo c a te d Gro ups c a n b e ma de priva te with a fix size , b ut a re L a b Gro up is a fixe d size tha t’ s e la stic sha re d b y use rs.

Sa mple L a b Gro up Hie ra rc hy Vie wpo int Teradata Database DW 1 Marketing Sales Data Scientist Lab Group Lab Group Lab Group Campaign Demand Curve Risk Analytics Lab Lab Lab Promotion Sales Customer Segmentation Lab Lab Forecast Lab Hierarchy Personal Lab Lab

Da ta L a b Po C / ROI Me tric s Be fo re & Afte r Before After Gains Core Process /ROI Modeling Tools Measure Tools Measure Difference Improvement Data Aggregation Base SAS / 1200 Minutes SQL / SAS DI / 2 -1198 59900% SPDS In-DB Minutes Model Execution Base SAS / 1800 Minute s SQL / SAS / 30 -1770 5900% SPDS In-DB Minute s Model Fit/QC Base SAS / 1200 Minute s SQL / SAS / 240 -960 400% SPDS In-DB Minute s Manual QC Excel/SAS 3600 Minute s Data Lab / SAS 10 / Excel Minute s Total Time 130 Hours 5 Hours -125 2768% 3 1 -2 200% FTE's Brands 5 5 (18 Possible)

T he Va lue o f I n-Da ta b a se Ana lytic s Efficiencies gained in core process and tactical projects could be funneled into doing more strategic projects (Strategic Tactical Core process 10%) (20%) (70%) Steps to be taken to reduce time in the core processes: Steps for expanding • Delivery process excellence the work-stream: • Accelerators • Analytics toolkit • Active focus on identifying projects • Large scales standardization: of Strategic value e.g in the ROI and Marketing mix process • Make more resources dedicated to this work- high level of automation and stream standardization has been achieved Strategic Tactical Core process (60%) (10%) (30%)

Justifying Ana lytic Improve me nt Cha ng ing b e ha vio r re q uire s mo re tha n a n ide a , it re q uire s a b usine ss va lue

2016 L e xisNe xis - T he T rue Co st o f F ra ud Study $30,076 Pe r Mo nth ($1002 pe r da y) On a ve ra g e , US me rc ha nts re po rte d a n 8% inc re a se in the c o st pe r do lla r o f fra ud lo sse s, fro m $2.23 to $2.40. T his me a ns tha t fo r e ve ry do lla r o f lo sse s, me rc ha nts a re lo sing $2.40 b a se d o n c ha rg e b a c ks, fe e s a nd me rc ha ndise re pla c e me nt. T he a ve ra g e numb e r o f mo nthly fra ud a tte mpts ha s spike d b y 33% (2015 – 2016).

Va lue fro m Re duc ing Mo de l De ve lo pme nt T ime F aste r analytic syste ms allow use r s to build, te st and imple me nt ne w mode ls mor e quic kly, c r e ating additional value for the or ganization. Time to Develop Model Value Creation Current Process Improved Process Additional Value Created T his va lue is c o mpo und e d b y the # o f mo d e ls intro d uc e d o r upd a te d e a c h ye a r. $1000 9 20 $180,000 Ave ra g e Da ily E xtra Da ys # o f Mo de ls Additio na l Va lue Va lue Cre a te d o f Usa g e Cre a te d Cre a te d

Va lue fro m I mpro ve d Mo de l Pe rfo rma nc e Be ing able to do mor e mode l te sting and to update e xisting mode ls to ac hie ve optimal pe r for manc e c an add signific ant value ove r time $500k $216,000 $400k Additional Value Cre ate d Pe r Ye ar $300k $200K Improve d Mode ls $100k E xisting Mode ls $0 Mo nth Mo nth Mo nth Mo nth Mo nth Mo nth 2 4 6 8 10 12 $30,000 3% 20 $18,000 Ave ra g e Mo nthly E xtra Ga in in # o f Mo de ls Additio na l Va lue Va lue Cre a te d Pe rfo rma nc e Cre a te d Cre a te d Pe r Mo nth

Busine ss Solutions T ha nk You! > High-Impac t Optimize d SAS Busine ss F or more informa tion, Ar c hite c tur e s Outc ome s Visit: www.T e ra da ta .c om/ SAS E ma il: SAS@T e ra da ta .c om SAS Analytic E xpe r tise