Data Integration and Marketing Attribution
Download as PPTX, PDF1 like333 views
Roivenue is an award-winning marketing attribution platform that has experienced significant growth and provides data integration solutions to enhance marketing performance. The platform enables clients to optimize their marketing strategies through insightful data processing and integrates seamlessly with various systems while ensuring data transparency and security via Microsoft Azure. Multiple case studies demonstrate how Roivenue improves marketing ROI by utilizing data-driven approaches to identify underperforming channels and optimize budget allocation.
Data Integration and Marketing Attribution
- 2. ROIVENUE #1 ATTRIBUTION VENDOR IN CEE 420% Y/Y GROWTH AWARD WINNING PLATFORM WELL FUNDED SPIN OFF FROM TOP5 Czech startups 2018 From 400 applicants Performance Diamonds CEE 1st place - Performance Technology of the Year Microsoft Partners Awards 2nd place in the category Marketing and e-tail Marketing Grand Prix 1st place + special rating ROI king
- 4. To ask questions please go to: sli.do #ROIVENUE
- 5. William Edwards Deming Engineer and management consultant best known for developing statistical process control called Plan- Do-Study-Act. His process was used to rebuild Japan from the ashes of World War II to becoming the second largest economy in the world. ORIGINS “It is not enough to do your best, you must first know what to do, then do your best.”
- 6. 1945 2018 THIS IS JAPAN
- 7. THIS IS YOUR MARKETING
- 8. THIS IS YOUR MARKETING Most companies are in this area.
- 9. THIS IS YOUR MARKETING Most companies are in this area. ● Internal bureaucracy ● Skip levels without preparation ● Loss of institutional knowledge ● Lack of financial resources ● Lack of systems and processes ● No consensus on data clarity
- 10. THIS IS YOUR MARKETING ROIVENUE brings them to maximum performance
- 11. THIS IS YOUR MARKETING ROIVENUE brings them to maximum performance
- 12. MAKING THE RIGHT DECISION REQUIRES THE RIGHT PROCESS
- 13. 6 STEPS TO SOLVING PROBLEMS Problem Definition Criteria Selection Alternative Development Analysis Decision Action Plan Creation
- 14. Problem Definition Criteria Selection Alternative Development Analysis Decision Action Plan Creation Analysis Stuck Here 6 STEPS TO SOLVING PROBLEMS
- 15. ● Different criterias are evaluated against alternatives ● Without the right data, subjectivity can skew the scoring ● Without the right data, scoring is prone to human error ● Without the right data, marketing maturity is impossible ● Can lead to Analysis Paralysis ANALYSIS
- 16. THE POWER OF ROIVENUE
- 17. THE POWER YOU ● YOU concentrate on strategy ● ROIVENUE concentrates on insights THE POWER OF YOU
- 18. ACHIEVING THIS IS ONLY POSSIBLE WITH RIGHT DATA INTEGRATION
- 26. ● SaaS solution with simple integration ● Flexible and non vendor locking ● Up to date with latest connector API ● Variable pricing based on volume ● Minimal IT resources required ● Tight security of MS Azure standards ● Time to deployment = 1 week WHY ROIVENUE AND ENTERPRISES
- 27. Advantages: ● Clients can run their own algorithms ● Clients can replace modules from SaaS ● Clients can control their data ● Clients can switch SaaS solutions easily ● Clients can boost SaaS capabilities with easy integration ADVANTAGES OF USING AZURE ROIVENUE believes in full data transparency and security. Azure allows us the flexibility to configure the system to every client’s need and the power to scale any business.
- 29. WHAT IS ATTRIBUTION ● “attribution is the process by which individuals explain the causes of behavior and events.” ● In marketing, attribution models assign a value to each touch point in the path to purchase
- 30. HOW THIS APPLIES TO MARKETING ROIVENUE™ gives you a new dimension to your marketing data so you can see CLV and real orders ● Integrate seamlessly into your CRM ● Track and manage the whole process from the first touchpoint, to generated profit and CLV Transaction / Revenue DoubleClick impression AdWords click Criteo impression Facebook click Deliveries Profit CV
- 31. Adform impression DoubleClick impression DoubleClick impression Facebook click AdWords click Criteo impression Conversion THE WHOLE PATH 100% LAST CLICK ATTRIBUTION 50% LINEAR ATTRIBUTION 54% 46% CLICK BASED ATTRIBUTION 5% 3% 3% 47% 36% 4% IMPRESSION BASED ATTRIBUTION 50% TYPES OF ATTRIBUTION
- 32. HOW TO DETERMINE ATTRIBUTION OF A CHANNEL Facebook click Google Ads click Conversion Facebook click Google Ads click Criteo click Conversion 100 customers 10 Goals (P = 10%) 100 customers 12 Goals (P = 12%) ΔP = 12 - 10 = 2%
- 33. THE POWER YOU ● YOU concentrate on strategy ● ROIVENUE concentrates on insights THE POWER OF YOU
- 34. BUDGET OPTIMIZER Our algorithm identifies oversaturated channels and suggests budget reallocation for the highest ROI.
- 35. Case Study 1 E-SHOP INCREASES PROFIT BY EXCLUDING BAD CLIENTS
- 36. ● Some customer misuse return policy ● Packaging and delivery costs reduce ROI ● Let’s not market to unprofitable customers HYPOTHESIS: Can we increase marketing ROI by excluding non-profitable customers?
- 37. ● Identify customers with negative CLV ● Integrate CRM data with Marketing platforms ● Exclude bad customers from marketing campaigns ● Run multi-channel campaign with the same budget and measure mROMI. PROJECT PLAN
- 38. ● Scoring of bad customers was already developed by the client in R studio ● Client wanted to control the scoring algorithm WHY AZURE FOR THIS CASE?
- 39. User information from website, marketing platforms and internal system combined and processed in ROIVENUE.
- 40. Unified and cleansed user activity exported to AZURE
- 41. Millions of rows further transformed and prepared for scoring
- 42. Trained model used to score customers
- 43. User’s predicted CLV read by ROIVENUE
- 44. User’s predicted CLV pushed to marketing platforms Unwanted customers are excluded from the marketing communication
- 46. RESULTS Conclusion: Delivery Ratio Increase by 24%
- 47. RESULTS Conclusion: mROMI increases by 11% from 2.3 to 2.55
- 48. Case Study 2 MULTI NATIONAL INSURANCE COGS PREDICTION
- 49. ● ROIVENUE evaluates marketing effectiveness including the gross profit ● Achieved by integrating the COGS for every transaction with the internal ERP system ● Problem: Precise COGS known only after the end of the contract ● Prediction: COGS algorithms are applied to calculate the profit HYPOTHESIS: Can we see the whole marketing efficiency up to the profit?
- 50. Predict COGS of each transaction from the internal ERP PROJECT PLAN
- 51. ● Client did not want to disclose the algorithm which predicts COGS, yet wanted to manage better their campaigns ● Developing such system internally would take 2 years ● ROIVENUE and Azure produced this to 2 months ● All data privacy concerns were covered WHY AZURE FOR THIS CASE?
- 52. User information from website, marketing platforms are combined and processed in ROIVENUE. Internal system is not connected.
- 53. Aggregated reports (OLAP cubes) were exported to Azure
- 54. Data from internal system is read and stored in a dedicated Azure SQL database
- 55. Data from ROIVENUE and internal system is combined. Analysis Service is used to build aggregations
- 56. Power BI is used to visualize data from the Client’s managed cloud database
- 58. RESULTS Conclusion: Only part of data be visible in ROIVENUE (values censored)
- 59. RESULTS Conclusion: Full marketing efficiency in Power BI (values censored)
- 60. Case Study 3 MULTI NATIONAL BANK CAMPAIGN REACH AND TARGETING OPTIMIZATION
- 61. ● New cookies do not have the historical information, ● we even do not know if those belong to ours clients ● and absolutely no idea if those already purchase the product ● Cookies in marketing platforms are degrading rapidly ● 80% of cookies will not survive longer than one week ● The idea is that continually enriching collected cookies should yield better results HYPOTHESIS: Can we increase performance by enriching historical cookies collected from our web properties with data from newly received cookies?
- 62. ● Match purchased DSP cookies to known first party cookies ● Match first party cookies to user profiles ● Push user profiles to DSP/DMP marketing platform Phase One Get maximum reach by collecting external cookies Phase Two Optimize targeting to get maximum visits Optimize targeting to get best CTR PROJECT PLAN
- 63. ● Marketing systems are producing huge volumes of data ● Azure is an elastic cost effective system for processing ● Cookies were able to be enriched on daily bases ● Cloud system can be decommissioned after the campaign WHY AZURE FOR THIS CASE?
- 64. User information from platforms are combined / processed in ROIVENUE. Internal system is not connected.
- 65. User cookies from website and from RTB platforms are pushed to Azure. Over 100 million events produced every day.
- 66. Internal first party cookie tracking is running on Azure and used by the DSP
- 67. Third party to first party cookie matching runs on daily bases
- 68. Phase One cookies from DSP network with banking affinity loaded into the DMP
- 69. Phase One Campaigns objective is to reach all unknown cookies
- 70. Phase One Reach is monitored in Power BI to decide when to stop
- 71. Phase Two Profile for known cookies is built, enhanced with internal data, then matched if consent was granted
- 72. Phase Two Cookie profile is scored on a daily basis for upsell potential. Two models are executed: one for existing customers and the second for prospects.
- 73. Phase Two Cookies with high probability to convert are pushed to the DMP
- 75. CAMPAIGN RESULTS Conclusion: Visits increase after both enrichments (values censored) Numbershavebeencensored
- 76. CAMPAIGN RESULTS Conclusion: Cost per visit drop after second enrichment (values censored) Numbershavebeencensored
- 77. Case Study 4 TELCO USES ATTRIBUTION TO MAXIMIZE AD PLACEMENT ROI
- 78. ● problem had always been knowing which placements are the most effective in bringing in sales and contracts. ● traditional measurement had not been able to determine which placement is better ● led to a conservative strategy which meant overspending on media placements that didn’t have any proven ROI HYPOTHESIS: We can identify underperforming ad placements by measuring impression based attribution Multi-touch Attribution
- 79. TECHNICAL CHALLENGE Matching the ad impressions from media placements with those from ad platforms and conversions
- 80. PROJECT PLAN ROI = 0,9 10,000 EUR investment => 19,000 revenue ROI = 14 10,000 EUR investment = 150,000 EUR revenue ● Same campaign tested on different media placements
- 81. CAMPAIGN RESULTS: Conclusion: With redistribution of budget, we were able to double sales revenue with the same budget
- 82. CAMPAIGN RESULTS: Conclusion: With redistribution of budget, we were able to double sales revenue with the same budget
- 83. Case Study 5 ACQUISITION CAMPAIGN FOR A MARKETING EVENT
- 84. CAMPAIGN RESULTS: Conclusion: Using the platform allowed for optimal performance of the campaign
- 86. AUTOMATION FREES UP TIME NUMBER OF CLIENTS MONTHLY TIME SAVED ON REPORTING 10 18 25 46 50 92 100 185 300 555
- 87. REVENUE AND SPEND ANALYSIS
- 88. MANAGEMENT DASHBOARD The performance dashboard allows for management to view the all client data and KPIs
- 89. ROIVENUE™ PRODUCTS DATA INTEGRATION Data extraction, cleansing, collation and export for consolidated view of marketing process 1 000 000 sessions 20 000 0000 impressions 3 websites No attribution ATTRIBUTION ROI & basic attribution for e-commerce & performance marketers. 1 000 000 sessions 20 000 0000 impressions 5 websites Click-based attribution ATTRIBUTION + Full cross-channel impression based attribution for brand builders 2 500 000 sessions 80 000 0000 impressions Unlimited websites Impressions-based att. ENTERPRISE Best fit for complex advertising, data and measurement infrastructure Unlimited Sessions Impressions Websites Impressions-based att.
- 90. TRY US FOR FREE All attendees will receive a link to try ROIVENUE for 1 month.
- 91. CONCLUSION “Computers are incredibly fast, accurate, and stupid. Human beings are incredibly slow, inaccurate, and brilliant. Together they are powerful beyond imagination.”
- 92. Thank you for coming
Editor's Notes
- #2: Děkujeme MS, přesouvali jsme venue
- #3: Kombinace digitálních agentur, data scientistsů a data analytiků a vývojářů
- #6: Roivenue je jako W. E. D. Engeneering DNA. Statistician. Developed techniques used for sampling in census - used to date. How he hepled rebuilt post war Japan. A lot of nations and companies try to do their best.
- #7: Japan. From ashes to one of the biggest economies and advanced societies. The Americans from Detroir went to Toyota to learn. Recently creating AI that replaces even creatives. Even been to Japan? Land of robots. Toilet has 10 buttons.
- #8: I was in Japan but I was also in hundreds of companies that do digital marketing.
- #10: Two main reasons: skipping levels and no consensus over what metrics and KPIs to watch. - one client who does it really well - watches the same KPI on weekly meetings.
- #12: IN mall or Alza, they are just now trying to figure that out
- #14: PD - what are awe actually gonna solve and what is the scope Criteria - tangible metrics Alternatives - buying a house than different neighborhoods. Analysis - some criteria are going to be subjective
- #16: Cannot pull data. Don't agree on the data or don't believe the data. Are reevaluating criteria
- #17: Cost of traffic, no bounces, must be new customers
- #28: Pavel
- #29: Tady přichází síla Azure a MS technologií. Nicméně jsou MArketingové probémy které Roivenue jestě nedokáže vzresit, a nebo není vhodné je v Roivenue resit i kdzy bz to bzlo moyne. Roivenue je samozrejme mozne propojit s jinymy cloud sluzbami, ale o tech se dnes nebudeme bavit. Tim ze Roivenue bezi na Azure je tento ekosystem pro integraci nejblize a zat9m jsme na teto cloud platforme realizovali nejvice enterprise integraci.
- #30: Dnes bzchom vám ukázali několik Marketingových scénářů a jejich následnou integraci do Azure pomoci Roivenue. Zde vidíte pouze nekolik produktů firmy Microsoft, které jsou pro naše scénáře nutné. Těch technologií Azure nabízí více, ale to nechme na jindy.
- #37: Martin Pripravili sme si tri case studies a náš CMO za Roivenue sa chce podeliť o aplíkácií systému pri marketing na tento event. Začíname prvým case, kde bolo cielom zoptimalizovať cielenie marketingových kampaní tak, aby sa sústredili iba na zákazníkov s kladnou zákazníckou hodnotou.
- #38: Je vám možno známy nešvar, kedy niektorí zákazníci majú tendenciu zneužívať možnosti vrátenia zboží a tak vytvárať dodatočné náklady pre váš biznis. Vzniknutá objednávka na webe ešte neznamená, že budete realizovať profit. A zákazníci, ktorí často vracajú tovar neprinášajú potrebný obrat na to, aby pokryly svoje náklady. Preto dáva obrovský zmysel nesústrediť sa na nich v marketingu - naše náklady na nich by boli totiž zbytočné. Radi by sme teda prestali cieliť reklamu na zákazníkov, ktorí sú stratoví. Informácia o tom, ktorí to sú je už napočítaná interne. Ďalším krokom bude výsledky tohoto nápočtu preniesť do marketingových platforiem a naučiť ich lepšiemu cieleniu. Ako k tomu pristúpiť?
- #48: Martin Monitor popis Na výsledky sa pozrieme do Roivenue Performance monitoru, dashboardu ktorý sleduje marketing ako proces od investície, cez vytvorenú a doručenú objednávku až po realizovaný profit. Poskytuje pohľad na okolo 60 metrík, ktoré umožňujú identifikovať úzke hrdlá v tomto procese. V tomto case sme sa zamerali na sekciu doručovania objednávok vytvorených na webe. Na spodnej časovej ose je vidieť moment, kedy sme aplikovali výsledok nápočtu kvality zákazníkov v R. A presne v tomto momente sme zaznamenali skokový nárast miery doručenia
- #49: Pri pohľade na celkový proces je vidieť celkový výsledok nášho snaženia vyjadrený návratnosťou. Na stĺpcoch vidíme investíciu do marketingu a čiarový graf nám ukazuje jej návratnosť vyjadrenú ako mRomi. Táto metrika vyjadruje profitovú návratnosť investície a je primárnou metrikou pre vyhodnotenie efektivity používaniou v Roivenue. Ak máme mRomi rovné 3, za každú korunu investovanú do marketingu realizujeme 3 koruny hrubého profitu. Vo výsledkoch tohoto case sme zaznamenali skokový nárast v návrastnosti mRomi pri znížení investície do marketingu, čo je velmi priaznivý stav.
- #50: O vyhodnocovanie investície pôjde aj v našom druhom case, situácia sa tu bude ale iná. V prvom prípade sme pracovali s klasickým eshopom, u ktorého z interných dat lahko poznáme, aké sú naklady na objednávku. Teraz ale pracujeme s medzinárodnou poistovaciou inštutúciou a ako to v prípade zmluv na poistenie býva, náklady na jednu zmluvu v momente jej uzavretia nepoznáme.
- #51: Dostávame sa tak pred projekt, v ktorom chceme opäť vyhodnocovať marketingové investície až na úroveň vytvoreného profitu. Komplikuje sa nám to ale neznámymi investíciami. Reálne náklady budeme z povahy poistenia poznať až na záver, po uplynutí zmluvy. Preto bude nutné tieto náklady v momente vytovrenia zmluvy nejakým spôsobom predikovať. Poďme sa opäť s Honzom pozrieť na technické riešenie tohto case.
- #52: The project plan was simple. And it would be easy as well if there would not be a problem there?
- #58: Technologically it would be possible to develop such solution entirelly in the internal infrastructure, which was considered at the beginnig. The internal IT was very open minded and all security requirements were met. Unoficilly we even did hear, that some systems are better secured on the cloud than in the internall infrastructure.
- #59: Technologically it would be possible to develop such solution entirelly in the internal infrastructure, which was considered at the beginnig. The internal IT was very open minded and all security requirements were met. Unoficilly we even did hear, that some systems are better secured on the cloud than in the internall infrastructure.
- #60: Po všetkej tej práci v pozadí môžeme časť výsledkov vidieť priamo v Roivenue. Vyznačené časť procesu sa pozerá na data z marketingových platforiem (investícia, impresia, klik) spojené s datami z webu - návštevami a konverziami. To mi ale nedáva plnú informáciu, mňa zaujíma ten profit. Ten je štandardne vyjadrený v poslednej časti funnelu, ktorá je momentálne šedá a tieto data v ROivenue niesú dostupné, pretože sme nenapojili interný feed. Honza už spomínal, že celý dataset je prezentovaný v Power BI, takže pri pohľade tam...
- #61: ...a tam vidíme celý pohľad. Power BI je totiž živené datami uloženými na Azure a prezentuje integrované data z konektorov z Roivenue s výsledkami interného nápočtu klienta. Splnili sme tak zadanie a pozeráme sa na vyhodnotenie marketingových investic na základe mRomi, veľmi podobne ako v prvom príklade. Grafy máme ešte rozdelené podľa zdrojových platforiem, ktoré sú prítomné v datasete Roivenue. Na druhej strane stojí interná databáza klienta s mnohymi dimenziami ako napríklad demografické segmentácie, ktoré je tak isto možné aplikovať na prezentované výsledky. Marketinový manažér tak vďaka jednému dashboardu okamžite vie, aký je úspech jeho kampaní.
- #62: V tretej case sa pozrieme na case silnej banky, ktorá pripravovala akvizičnú kampaň a spojila sa s nami aby túto kampaň exekuovala efektívnejšie, ako tomu bolo vo zvyku. Záujmom nebol teda iba maximálny reach, ale aj efektivita cielenia reklamy v ňom.
- #63: Začiatok bol naplánovaný tradične, nákupom obrovského množstva cookies. Uvedomovali sme si ale známe problémy, ktoré so sebou takéto stratégie nesú. Nákupom veľkého množstva cookies na českom trhu, ktoré sú segmentované do potencionálneho záujmu na českom trhu vzniká tomuto silnému hráčovi očividný problém - aká časť týchto cookies patrí jeho existujúcim zákazníkom? Jedná sa o agresívnu akvizičnú kampaň, takže títo zákazníci sú mimo nášho pola záujmu a necheme na nich cieliť - netreba presviedčať presvedčených. V prvom kroku teda chceme o nakúpených cookies zistiť čo najviac, k tomu sa bude hodiť interná databáza tohoto klienta. A v ďalšiom kroku sa chceme pozrieť na známy problém degradácie cookies, kedy vieme, že väčšina z nich nevydrží viac ako týždeň. A čo bude ešte lepšie je mať možnosť obohacovať tieto cookies o vlastné data a zefektívniť tak komunikáciu s nimi v marketingových systémoch. Budeme tak nielen vedieť s kým sa bavíme, ale dostaneme aj možnosť optimalizovať našu komunikáciu tak, aby priniesla čo najlepšie výsledky. Poďme sa na to pozrieť, čaká nás najkomplikovaniejšia technológia dnešného dňa
- #64: Pilotný projekt, kterým oveříme potenciál
- #77: Výsledky opäť pozorujeme v Roivenue a sledujeme postupnú aplikáciu oboch riešení. V prvom bode vidíme nárast investícií - tu došlo ku nákupu nových cookies. V priebehu následujúcich troch týždňov sme kontrolovali reach v Power BI aby sme sa dostali na požadovanú hranicu. Zároveň sme sa na dostupných cookies tri týždne učili a následne bol aplikovaný krok dva, kedy sme začali na základe výsledkov nápočtu lepšie cieliť. Znamenalo to jednak to, že sme nekomunikovali s cookies, ktorí už boli našimi klientami, takže celková investícia sa zvýšila. Zvýšenú kvalitu zacielenia je potom vidieť na druhom slide...
- #78: Kde vidíme výrazný pokles ceny za návštevu - potrebujeme menej impresií na to, aby sme získali návštevu. To je spôsobené práve vyššiou kvalitou zásahu umožnenou obohacovaním cookies a prispôsobením komunikácie s nimi - použitie bannerov a placementov tak, aby najlepšie odpovedali cielovým segmentom ktoré vznikli z analýzy cookie dat.