Value-Based Bidding: Start bidding on the customers who impact your bottom line

Last week, I was chatting with one of our clients in retail and realized how Value-Based Bidding (VBB) is often seen as a mysterious solution: everyone has heard of it, but many wonder, “Isn’t this what Google Ads already does?”

Today, my teammate Javier Pozueco and I will clarify what VBB is and help answer the key question: is it worth pursuing?

The starting point: Not all customers are equal

You already know it, customer value varies. But this insight often stays locked where it belongs: in your CRM database. Sometimes you activate CRM audiences, but rarely do you leverage this data directly on media platforms. VBB lets you target based on the value each customer or conversion brings, aligned to your specific goals—whether that’s maximizing revenue, boosting profitability, avoiding cancellations/returns, or acquiring long-term customers. Whether your focus is online sales, offline sales, or lead generation, VBB is the right approach.

The philosophy of VBB

Once your objectives are clear, VBB lets you share the right data with Google Ads and weight conversions to optimize bidding. Instead of using the actual value of each conversion, you adjust conversion values to prioritize those most impactful for your business.

Take a real example: for our client Vodafone, bundle customers (home and mobile) are the most valuable. Using first-party data (CRM and on-site behavior), we identified signals that lead to bundle conversions—like time spent per page and specific offers viewed. With these insights, we developed a predictive model that assigns a custom value to each user who clicks on a Google ad: 1000 for likely bundle customers, 10 for other conversions, and 0 otherwise. This tailored value replaces Google’s automatic values, aligning the bidding strategy with Vodafone’s top priorities.

From VBB to predictive VBB

To give you a better view of the methodology to build the predictive model, I asked Javier, our data science lead, a few questions.

What data or resources do you need to set up a VBB project?

Javier: “We start by assessing your first-party data—typically CRM and analytics data. The goal is to identify what you know about your customers and how they interact with your brand. We also need a steady data feed to maintain the predictive model in Google Ads, ensuring it’s reliable over time. From there, we look at past conversion data to understand which behaviors and characteristics correlate with high-value actions. And of course, close collaboration with your media team helps refine targeting and make it actionable.”

Can you explain the process of building the model? How do you identify and combine the right signals to predict conversion value?

Javier: “Our process begins with data exploration to uncover patterns—such as how the user interacts with certain pages or which product categories lead to higher-value conversions. We look at metrics like visited pages, time spent on site, and even demographic data if available. Once we understand the behavior of high-value customers, we build a model that assigns a score to each interaction to predict these high-value customers. We use machine learning algorithms to combine signals associated with the user behavior patterns and we then test and fine-tune the model for accuracy. Each score reflects the probability that a user will take a high-value action, which we send back to Google Ads to drive more precise bidding.”

Are there any prerequisites to ensure the project will yield high impact for my business?

Javier: “There are two main prerequisites: a well-structured CRM or data warehouse that holds reliable customer data and sufficient conversion volume. Without a good volume of conversions, it’s hard for the model to learn accurately. We also need alignment with your business objectives—if you’re clear about your target outcomes (e.g., long-term value or profitability), we can tailor the model to prioritize those. We also recommend a testing phase to validate the model in real conditions before scaling up.”

Finally, is this model scalable across markets or product ranges, and how do you maintain it long-term?

Javier:  “Yes, scalability is one of the strengths of VBB. Once we have a successful model, it can be adapted across different markets or product ranges by adjusting the parameters to reflect new customer behaviors or goals. Long-term maintenance involves monitoring the model’s performance, especially if there are major changes in customer behavior or new product launches. Typically, we schedule regular retraining, which keeps the model accurate and responsive to trends. Plus, our team can fine-tune it with any updated customer data to keep it fresh.”

Predictive model is not the only way  

There are many ways to implement VBB, and it is not mandatory to create a complex predictive model or dynamic conversion values. There are more ‘artisanal’ approaches you can start with. In any case, the VBB requires three basics: (1) share better data, (2) assign relevant value data, and (3) optimize bids for outcomes that matter to your business. The method can be adjusted to your resources (set up budget, skills and goals).

So, should you try VBB?

What VBB can bring to your business depends on your objectives. Whether you’re aiming for volume, cost efficiency, or high-quality acquisitions, VBB is worth a test. According to Google, advertisers who switched from Target CPA to Target ROAS saw a 14% increase in conversion value. This shift in focus allows you to target higher-value users, not just those who click, but those who truly contribute to your bottom line.

If you need more proof, here’s how Jellyfish used VBB to deliver results for Vodafone. In targeting high-value conversions, we saw a 79% increase in activations, a 17% boost in ROAS, and a 23% reduction in CPA. Stefania Serafini, Senior Paid Search Expert, and Emiliano Bozzi, Digital Data Lead at Vodafone, said, “Google and Jellyfish collaborated with our Digital team to prioritize customer value, expanding beyond volume while enhancing RoAS and reducing CPA.”

In another case, we guided a major fashion retailer’s use of VBB to optimize profitability by predicting returns, leading to an 8% increase in gross profit. 

Ready to start?

Begin by identifying your objectives and assembling a capable team (hello!).

Co-written by: 

Vanessa Chaouat - Cloud Strategy Director, Paris
Javier Pozueco - Data Science Director, London