AI What to AI How - Jason Downie, Making Science
0 likes143 views
Making Science, a global marketing agency with over 1,200 employees and $200M+ in annual revenue, focuses on leveraging data and technology to enhance digital marketing performance for 600+ brands across 16 countries. The document outlines their innovative approaches to overcoming challenges posed by privacy regulations and cookie limitations, using solutions like Google Analytics 4 and first-party data strategies to optimize customer experiences and drive conversions. Notable case studies include significant gains in mobile transactions for an online retailer and improved lead generation for a university, demonstrating the efficacy of their advanced analytics and machine learning models.
AI What to AI How - Jason Downie, Making Science
- 1. KEYNOTE AI What to AI How Jason Downie US CEO MAKING SCIENCE LAS VEGAS, NV ~ OCTOBER 28 - 30, 2024 DIGIMARCONWORLD.COM | #DigiMarConWorld
- 3. MAKING SCIENCE AT GLANCE People Centric 1,200+ employees across the globe leveraging data and technology for more than 600 brands to solve their business problems Technology Driven Data, Analytics, Mar/Adtech, & Cloud expertise to drive digital marketing performance. Our team has 400 Engineers, 500 Google Ads / GMP and 90 GCP certified professionals. Global Experts Making Science has presence in 16 countries to support local & global customers around the clock, in more than 20 languages. Publicly Traded With $200M+ in annual revenue, our growth is driven by the success of our valued customers and collaborating partners. Barcelona Bogotá Chicago Ciudad de México Copenhagen Dublin Gothenburg Hamburg Helsinki Lisbon London Madrid Mallorca Malmö Miami Milán New York Oslo Padua Paris San Francisco Stockholm Valencia
- 4. MAKING SCIENCE AT GLANCE Performance Enhancing Technology Led by talented product managers, our team includes 500+ engineers who ship automation and AI technologies that supercharge Google Ads and GMP. Our award-winning solutions drive media performance and customer success. Data Activation Experts (Ads + Cloud) Our Google Ads, GMP, and Google Cloud teams drive results by helping customers build bridges across internal marketing, analytics, and IT stakeholders. We make 1P data useful for marketers and have the expertise to transform data from site analytics, CRM, and Cloud platforms into assets for media activation. Google DNA We are seasoned Google alumni with 25+ years of Google experience. As a Data AOR, we’re built with media agency collaboration in mind. We do not pursue long-term managed service or media AOR scope. Fully Transparent Operating Model We offer highly competitive GMP rates, clear-cut pricing, and do not require annual minimums for DV360, SA360, or CM360 contracts. As a practice, we never add hidden service fees. All consulting and implementation services are explicitly discussed and scoped with our brand and agency partners.
- 5. MAKING SCIENCE AT GLANCE MEDIA CLOUD SECURITY GMP Sales Partner ANALYTICS & ADVERTISING Incrementality certified DV360 CM360 GA4 360 SA360 Tag Manager CREATIVE Conversion optimization +1000 certifications +500 certifications
- 6. DIGITAL TRENDS In a world where consumer expectations around privacy drive regulatory changes, the digital marketing ecosystem is rapidly evolving. The Digital Markets Act (DMA) and the continued signal loss from third-party cookies are reshaping how we interact with digital spaces. This evolution demands a proactive response. Businesses must prepare, adapt, and evolve to stay ahead. Regulatory Changes Reacting to consumer demands, stringent regulations like the Digital Markets Act (DMA) enforce enhanced data protection and privacy standards, compelling businesses to prioritize user data security. Consumer Expectations In 2024, privacy-conscious consumers demand greater transparency and control over their data, pushing for a harmonious balance between personalization and privacy in digital spaces. Technology Changes Adapting to the dual demands of privacy regulations and consumer expectations, new technologies emerge. A prime example is Google Analytics 4 (GA4), which introduces enhanced privacy features tailored to the modern web
- 7. DIGITAL TRENDS Apple’s Safari browser starts blocking third- party cookies by default, marking a major shift in privacy focus 2009 2017 Netscape Communications creates the cookie 1994 1995 Firefox begins to block third-party cookies by default 2018 Third-party cookies are first deployed, enabling advertisers to track users across different websites EU adopts the Cookie Directive, requiring user consent for storing or retrieving data, a first in addressing cookies' legal implications GDPR goes into effect, impacting data privacy and the use of 3P cookies CCPA is enacted, providing California residents with enhanced privacy rights 2020 2025 Google elevates user choice for turning off 3PC - expected 80% reduction in signal
- 8. DIGITAL TRENDS Lost conversions weighted by revenue 2017 Past Observed (vía cookies) Modeled automatically 0% Lost 2023 Present Observed (vía cookies) Modeled automatically Modeled if features adopted Lost Observed (vía cookies) Modeled automatically Modeled if features adopted 2024+ Future Observed (via Anonymous Identifiers) Precision Era to the Predictive Era = Do More With Less Data
- 9. DIGITAL TRENDS Virtual assistants Siri, Alexa… Online search Google Image and video editing Including automation Social media To show relevant content ECommerce Product recommendations Customer support Automatic responses Facial recognition In smartphones Personalized Ads Based on user behavior Personalized Interfaces In webs and apps
- 10. DIGITAL TRENDS
- 11. DIGITAL MARKETING MATURITY Compliance Centric Customer Centric Privacy Centric Value Centric Business Value
- 12. DIGITAL MARKETING MATURITY Business Value Compliance Centric Customer Centric Privacy Centric Value Centric Privacy/Tag Audits Consent Mode V2 GA4 Source of Truth Offline Data Integration via APIs (Composable) Customer Data Platform / Enterprise Data Warehousing Deployment Customer Match Data Driven Attribution Gen AI for Content Creation Predictive AI (Propensity/pLTV) for VBB & Custom Bidding* Regression-Based Attribution* Implementation & GA4 Migration Enhanced Conversions Value Based Smart Bidding (VBB & Custom Bidding) Consent Banners & Consent Management Platforms 3P Cookie Mitigation Plan PAIR Biddable Conversion Values (data layer/variables) Cleanroom Analysis BigQuery Data Export Reporting & Analysis Rules-Based VBB Measure Engage MS Tech Available * Creative Automation* GA4 Audiences
- 14. DIGITAL MARKETING MATURITY Business Value Compliance Centric Customer Centric Privacy Centric Value Centric Privacy/Tag Audits Consent Mode V2 GA4 Source of Truth Offline Data Integration via APIs (Composable) Customer Data Platform / Enterprise Data Warehousing Deployment Customer Match Data Driven Attribution Gen AI for Content Creation Predictive AI (Propensity/pLTV) for VBB & Custom Bidding* Regression-Based Attribution* Implementation & GA4 Migration Enhanced Conversions Value Based Smart Bidding (VBB & Custom Bidding) Consent Banners & Consent Management Platforms 3P Cookie Mitigation Plan PAIR Biddable Conversion Values (data layer/variables) Cleanroom Analysis BigQuery Data Export Reporting & Analysis Rules-Based VBB Measure Engage MS Tech Available * Creative Automation* GA4 Audiences
- 15. ● ● ● ● ●
- 16. Retail Brand The Challenge An online retailer specialized on branded equipment, clothing, shoes and accessories, faced difficulties in tracking users who started their shopping experience on desktop but completed their purchase on mobile device. This lack of visibility limited a comprehensive understanding of the customer journey, impacting critical decisions regarding user experience and marketing strategy. Additionally, it was observed that many mobile users faced difficulties completing the checkout process. The Results Optimizing the mobile site and implementing User ID tracking led to a substantial increase in mobile purchases, driving both revenue growth and a better customer experience. ● Increase transactions from mobile devices: from August 24 to October 15, 2024, mobile transactions increased by 27.2% compared to the same period in 2023, rising from 384,544 to 489,954 transactions. ● Increase revenue from mobile devices: Thanks to the User ID implementation and mobile optimization, mobile revenue grew by 27.5%. ● Enhanced user experience: GA4 insights enabled us to identify and resolve pain points for mobile users, leading to a seamless experience and higher customer satisfaction. The Approach We implemented Google Analytics 4 (GA4) with a strong focus on User ID to improve cross-device tracking. This setup allowed us to track users across devices, offering clear insights into the customer journey, including when users logged in on one device and completed their purchase on another. Analysis revealed that 55% of transactions occurred on mobile devices, which prompted the decision to optimize the mobile experience by improving the website's design and functionality to make the mobile purchasing process smoother and more effective. The business objective Enhance cross-device visibility and mobile user experience to increase conversions and maximize customer lifetime value.
- 17. Led Gen The Challenge A top international university aimed to optimize online student recruitment in a competitive environment with 20% less data due to new regulations (Digital Market Act-DMA, elimination of cookies). The goal was to increase campaign effectiveness for the academic year, targeting a 50% growth in leads and a 30% reduction in cost per lead in its 9 main markets. The Results The cookieless strategy using first-party data proved effective in a data-limited environment. Meaning a 78% increase in pre-enrollment leads, a 3.5x boost in conversions, and a 70% reduction in cost per lead, optimizing the budget and surpassing recruitment campaign effectiveness goals before the academic year began. The Approach A cookieless strategy based on our client first-party data was implemented in three phases. ● GA4 was deployed to unify campaign and web behavior data. ● Privacy solutions like Google's Enhanced Conversions were introduced to address information gaps due to new regulations. ● Server-to-server measurement was established to capture data blocked by other browsers. After refining the data, we activated it through an AI- driven Value-Based Bidding strategy in Search Ads (SA360) campaigns, targeting users identified as valuable in the previous stage.
- 19. DIGITAL MARKETING MATURITY Business Value Compliance Centric Customer Centric Privacy Centric Value Centric Privacy/Tag Audits Consent Mode V2 GA4 Source of Truth Offline Data Integration via APIs (Composable) Customer Data Platform / Enterprise Data Warehousing Deployment Customer Match Data Driven Attribution Gen AI for Content Creation Predictive AI (Propensity/pLTV) for VBB & Custom Bidding* Regression-Based Attribution* Implementation & GA4 Migration Enhanced Conversions Value Based Smart Bidding (VBB & Custom Bidding) Consent Banners & Consent Management Platforms 3P Cookie Mitigation Plan PAIR Biddable Conversion Values (data layer/variables) Cleanroom Analysis BigQuery Data Export Reporting & Analysis Rules-Based VBB Measure Engage MS Tech Available * Creative Automation* GA4 Audiences
- 20. DIGITAL TRENDS
- 21. MAKING SCIENCE SOLUTIONS & PRODUCTS Challenge: Manual management and creation of effective RSA copy and PMax assets is a time and labor intensive process, for businesses that manage multiple brands, diverse SKUs, frequently updated inventory or promos, or multiple languages. Solution: ● Algorithmically analyzes search behavior and identifies top performing message structures to generate tailored ads for each search using Natural Language Processing technology. ● Analyzes consumer behavior and variables that move them from the evaluation to the purchase, creating an ad for each user and each search. ● Generative AI adapts existing assets to streamline low performing PMax creative optimization. Continuous learning technology efficiency increases over time, detecting and adapting to changes in consumer purchase behavior and market nuances. Customer Profile: Google Search RSA and PMax spenders. Rely on product feeds, manage frequent product description, inventory, promo, and/or multi-lingual updates. Verticals include Commerce (Ecom, Retail, Travel), Lead Gen (Finance, Auto) Impact Benchmarks: +11% CTR, +15%, ROAS -10% CPA Data Connection (Access) Journey Analysis Activation Strategy Activation Performance Analysis 2 weeks 3-5 weeks Product feed Prerequisites ● GCP Project ● Product feed or substantial CSV of products ● Online conversion tracking ● Min. conversion threshold ● Diversity of search patterns Available in the
- 22. About Google Marketing Platform: Google Marketing Platform is a unified advertising and analytics platform that helps enterprise marketers make better decisions faster. With Google Marketing Platform, you’re in control of every campaign, so you have the flexibility to adapt to the needs of your business and your customers. Learn more at g.co/marketingplatform © 2023 Google LLC. All rights reserved. Google and the Google logo are trademarks of Google LLC. All other company and product names may be trademarks of the respective companies with which they are associated. MAKING SCIENCE SOLUTIONS & PRODUCTS MediaMarkt is a multinational leader in the distribution sector of consumer electronics and related services. It is part of the MediaMarktSaturn Retail Group, with more than 1,000 stores located in 10+ countries. They view themselves as a partner, daily companion and navigator of their customers, and combine digital products and services to create practical solutions. Their portfolio of brands and offerings enables them to respond flexibly to the needs of different customer groups and different countries. The Challenge Their priority was to seamlessly incorporate their constantly evolving client offers (ie. simultaneous promotions, limited time sales, etc.) into their digital advertising strategy and optimize for performance. Historically, the variety of offers led to a lack of organization, visibility and misattribution of conversions. MediaMarkt was also looking to enable real time product management, where they were able to adjust product promotions and prices based on different circumstances (ie. stock levels, positioning, priority etc.) The Approach Making Science deployed our proprietary technology, ad- machina, to unlock real time management and automation of Google Ads campaigns, while taking into account a catalogue of over 50,000 products and constantly shifting available offers. ad-machina also performed continuous monitoring of search engine queries, and identified any missing keywords within placements, dynamically incorporating them into creative to maximize relevance and improve conversion performance. The entirety of this process took ~6 weeks, with data analysis and activation occurring within two weeks and result evaluation completing within a month. The Results +45% ROAS +13% Sales +12% CTR of existing ads thanks to their greater visibility +24% CTR of offers that were not included until now
- 23. MAKING SCIENCE SOLUTIONS & PRODUCTS Challenge: Historical user behavior can be used to optimize media performance when converions occur within the next a few days. However, the real-time decision making required to effectively optimize for conversion values that occur beyond 7 days predictive modeling is required. Many advertisers get stuck bidding to conversions or single transaction values and lack the tech and marketing science to develop custom models to shift their bid strategy focus from volume to value. Solution: ● VBB & Custom Bidding acceleration platform helps define predictive values based on propensity to convert, churn, cancel, or predicted Lifetime Value ● Data onboarding to Google Ads, SA360, or DV360 is accomplished in real time using a lossless 1P pixel or Server Side integration to enhance Google AI bidding with smarter conversions Customer Profile: Google Search & Display spenders with ROAS, LTV, or Profit Margin marketing objectives and a need for real-time scoring to improve AI Bidding. May have sales cycles or customer lifecycles that extend beyond 7 days. Verticals include Commerce (Ecom, Retail, Travel, Telco) and Lead Gen (Education, Finance, SaaS). *Dictated by variable customer needs including white glove service, model volume, data source additions, & missing prerequisites Prerequisites ● 1 or more data sources with conversion values ● Historical data (depending on sales cycle) ● Min. conversions volume (~200/month varies by use case) ● GCP project ● Ability to add the Gauss Tag to Google Tag Manager or integrate with Gauss S2S ● Media activation budget (~$50K+/month, varies by use case) Data Capture & Quality Business & KPI alignment Model Training Activation Measuring 3-6 months Monitoring & Retraining smart advertising
- 24. Proprietary + Confidential The Results ● +45% New Customers were acquired or re- activated (non purchases in a year) in the test arm ● -34% Cost per acquiring or re-activating new customers ● +50% LTV ROAS was achieved for new customers in the test arm Carrefour captures +45% more new customers with +37% LTV ROAS in DV360 thanks a to a bespoke ML model that predicts high value customers CASE STUDY The Challenge For omnichannel retail groceries, Lifetime Value (LTV) plays a key role. Measuring Customer loyalty and making marketing decisions based on LTV across channels is essential to acquiring new loyal customers. One of the biggest challenges about LTV is its difficulty to include these signals in DV360 in new customer acquisition campaigns. While for returning customers historical transaction data is available for predicting future LTV, for new acquired customers predicting LTV and using these signals for Bidding is more challenging. The Approach Carrefour, In collaboration with Making Science developed a Machine Learning model that predicts the LTV for new acquired customers and makes the predicted values available for bidding with MatchID. Making Science has developed a Custom Bidding Algorithm that leverages the Machine Learning LTV signal in Prospecting DV360 campaigns (new customer acquisition campaigns). Arena as the Agency and as the responsible for the media activation developed an A/B test in DV360 comparing previous Custom Bidding (using Revenue signals) vs. the new test leveraging the new Customer Bidding model (using predictive LTV signals) The A/B test ran for 6 months New Customer Acquisition LTV ROAS for new customers © 2022 Google LLC. All rights reserved. Google and the Google logo are trademarks of Google LLC. All other company and product names may be trademarks of the respective companies with which they are associated. +50% +45% Carrefour Group is a French multinational specialized in retail that now has stores in more than 30 countries reaching 104 million households across the world each year. They operate thousands of stores and an eCommerce platform that sells a wide variety of products including online groceries. Carrefour has an inspiration to continue growing in the online environment by accelerating its digitization “Thanks to optimising towards high LifeTime Value customer signals in DV360 rather than just transaction value, we are able to truly align marketing campaigns with our internal goal of acquiring high LTV customers” Carrefour Cost of Customer Acquisition -34%