Value-based optimisation (VBO) is the use of ad buying platform bidders to generate business value above and beyond standard conversion volume. When activated, it creates a stronger through-line between media campaigns and business outcomes, allowing marketers to replace customer acquisition cost (CAC) or cost per acquisition (CPA) models with more meaningful business goals. This can include:
- Maximising long-term KPIs such as LTV, retention, loyalty etc.
- Acquiring and growing specific engagement or revenue cohorts
- Optimising to value across fragmented customer touchpoints (i.e. omnichannel retail, offline-to-online journeys)
This shift from conversion quantity to conversion quality is illustrated in Figure A below.
Figure A – VBO makes media platforms work harder to acquire customers from desired value cohorts
More accessible, versatile, powerful
In the last 1-2 years, the combination of first party data ingestion features and customisable bidding models has made value-based optimisation across the open web and especially walled gardens (Google Ads, Meta, TikTok etc) more accessible, versatile and potent.
Many embedded media platform features, from value-based bid adjustments to site/app analytics integrations, allow marketers to enable VBO strategies without writing code. On the other hand, high-code options allow more technical marketers to deploy custom value events that come from offline touchpoints and/or predictive models. This in turn unlocks fine-grained campaign optimisation.
More resilient attribution supported by first party data
The crucial component that makes this all work in a signal-impoverished world that (still) heavily relies on third party cookies and mobile device IDs is first party data. From hashed emails to clickIDs, various first party customer data parameters serve as join keys that allow advertisers to share business values and events with media platforms.
Sharing these customer parameters has become easier for marketers, thanks to the availability of tag-based1 and/or API-based2 mechanisms. The former involves capturing hashed parameters across digital touchpoints with embedded code snippets, and the latter involves advertisers preparing and sharing relevant customer data from their servers or via integrations with other martech tools.
Two paths: Bring your value data or Build your own algorithm
The bigger the walled garden, the larger its logged-in customer base, and more often than not, the more advanced its optimisation algorithms. While there are gripes about increasingly opaque incentives (see Blackbox Gravity), advertisers who choose to share or bring their value data to these ad platforms effectively create a larger pool of durable, observable business outcomes for attribution, measurement and optimisation.
On the other hand, advertisers who want absolute, fine-grained control over ad buying while minimising exposure of their business data may consider building their own algorithm. However, this entails evaluating the business value of the many available ad impression levers (geo, url, attention, language, etc.), and ensuring the media platforms follow these evaluations when buying ad impressions in real-time.
Four methods of bidding to value
Now let’s further explore each path and their distinct methods:
Bring your value data
Business value
This is the most direct path available: Sharing actualised (recent transactions) or predicted business values (such as 90 day revenue or retention) with ad platforms for optimisation. Actualised conversion values are typically shared via tag-based solutions that capture real-time values such as e-commerce checkout revenue. Custom values such as profit margins, offline revenue, or predictive business values can also be provided via Event APIs or direct CRM/CDP connections.
Proxy events
Build your own algorithm
Bid adjustments
Bid adjustments allow advertisers to assign different values to different ad impression levers e.g.time of day, device, language, OS, etc. These rules allow advertisers to incorporate business context that isn’t shared with ad platforms. For example, they might bid 2x higher for impressions in Sydney vs Adelaide, reflecting the relative value of business generated in each city.
Bid adjustments is one of the oldest tricks in the media optimisation playbook, but it has many new names. These include bid factors, dimensional bidding and conversion value rules – just to name a few. Despite the different names, they ultimately do the same thing, which is to automate optimisations across multiple ad impression levers simultaneously.
Custom algorithms
While a rules-based approach typically starts with a marketer’s intuition (as in the example above, where location signals were identified), a ML-driven approach leverages machine learning to identify and evaluate a vast array of ad impression levers. Data preparation here involves log-level data from the bidstream, which contains relevant variables for machine learning analysis. In addition, data pipelines are often set up to maintain and update the weights of these ad impression levers based on feedback from in-flight outcomes.
Making value optimisation robust
So far, we’ve covered technical complexity and levels of exposure to business data, both crucial factors in selecting the right VBO solution. Now, let’s discuss two more important considerations.
- Measurement – Any real-time results observed through standard attribution methods (last-click or data-driven) are at best correlational. As such, this makes it difficult to determine whether VBO has genuinely increased business value compared to existing optimisation methods. To accurately express business impact, marketers should conduct incrementality experiments to isolate and measure the impact of their chosen VBO strategy.
- Fidelity – This involves evaluating the quality of the optimisation signal. Are we optimising to actual business value or a proxy? If it’s the latter, how much does it influence the target business KPI? Proxy events such as trial sign-ups or content downloads have a more direct, causal effect on your business compared to ad impression levers, which are indirect and noisy signals. The lower the influence of the optimisation signal, the lower the fidelity, requiring more caution when creating VBO solutions with it.
Finally, this all starts with a clear data strategy, where a solid understanding of your business and customers lays the foundation for operationalising your company’s growth model via VBO. Watch our two-part video series, where we delve deeper into this topic and explore a five-step framework to maximise success with your value-based optimisation or value-based bidding strategies.
2Ex: Google’s offline events, TikTok and Meta’s conversion / event APIs, and relevant server-side tagging offerings
