The ROI-maximizing power of marketing attribution continues to drive interest and the proliferation of martech solutions. However, marketers are left to determine the true value of new platform adoption. What is the likely ROI for a brand's specific attribution improvements?
Expected Value of Perfect Information (EVPI) is a financial framework that quantifies the maximum value a company could create by completely eliminating uncertainties around a key business decision.
Read More:
Acme Co is a B2B software business with an omnichannel marketing mix. The brand has multiple but limited attribution systems tied to channels. As a result, some channels use last-click attribution, some use linear modeling, and some are "black-box" with no model visibility.
Without a comprehensive full-funnel approach, leadership has determined the optimal budget split of 40% search, 30% social, 20% email, and 10% events. With $100,000 monthly spend, the total return is $400,000, or a ROAS of 4:1. Based on current monthly spend and ROAS, annual ROI is $3.6M.
Acme’s CMO knows the attribution model has inaccuracies. Due to last-click bias and linear model’s lack of precision, some marketing touchpoints are overweighted while others are underestimated.
But Acme lacks visibility into the expected gain from improving its attribution capabilities.
Brands without full-path attribution visibility base marketing spend allocation on limited data. Applying an EVPI framework can quantify the value of perfect attribution to guide investment tradeoffs. In the example above, current marketing allocations are based on historical spending, revenue data, partial attribution insights, and leadership intuition.
Now, a hypothetical perfect information scenario can be modeled. What would the optimized channel budget be with attribution data that more perfectly quantified each channel’s influence?
If Acme had visibility to all customer-brand interactions across all channels, and could calculate from the currently missing data, this perfect information would drive allocation decisions that create greater value.
For example, what if the business knew the exact path of every revenue producing conversion
With this additional information, Acme can see the discrete value of each marketing element: channel, landing page, ad, creative, content, sequence and combinations.
With a full-path attribution solution, Acme can view all converting paths and identify those marketing elements that create revenue.
By viewing total page paths that begin with paid search, Acme sees that over 400 paths exist. Filtering by “Revenue Generating” reveals that only 20 of the paths generate revenue. In other words, 5% of paid search is effective, and 95% inefficient. In addition, Acme now knows which ads are effective, and which are not producing revenue. Acme also knows which landing pages are effective within paid search paths.
Armed with more complete information, Acme can reallocate the majority of channel spend and repurpose landing page content most effective at driving revenue.
If the monthly paid search return was $160,000 from a $40,000 spend, eliminating the 95% wasted spend creates monthly savings of $38,000, or over $450,000 annually. Moreover, reallocating the 95% non-producing spend would multiply the return. The current channel value is based on a 5% efficiency rate. If the efficiency were to merely reach 50%, the return increases by a factor of 10.
In just one channel, the Expected Value of Perfect Information (EVPI) is a minimum of $450,000/year, and a maximum exceeding $10M. Armed with this quantification, Acme's CMO can evaluate investments to improve attribution:
EVPI provides a ceiling for how much spend can be justified to measurably improve attribution capabilities. Applying an EVPI framework brings greater analytical detail to evaluate attribution investments and maximize their strategic business value.
Finally, this example is limited to one channel (paid search) and two dimensions (ad spend savings and reallocation). In any contemporary complex marketing mix, the sources of EVPI are multiplied by channels and dimensions. Eliminating uncertainty drives increased value in each organic and paid channel by reducing waste, reallocating revenue-producing marketing elements, increasing Quality Scores, and optimizing paths for greatest return.
While the hypothetical perfect attribution scenario may not be attainable, EVPI analysis can guide marketers to focus investments on improving capabilities related to the largest sources of uncertainty. By methodically addressing these high-EVPI areas, attribution insights get incrementally closer to the ideal.
Challenge: A fundamental challenge with attribution is collecting comprehensive customer touchpoint data. Without complete journey data, models have intrinsic blindspots. Expanding data capture to additional media channels should be an EVPI investment priority.
Solution: An ideal attribution solution must use a universal tracking pixel independent of both 1) third-party cookies which can be blocked by browsers, and 2) ad platform APIs which limit individual data and add bias.
Challenge: Many attribution models rely on small subsets of observed consumer journeys to infer channel influence. But small samples increase uncertainty regarding true impact. Investing to increase data volume and representation could significantly improve confidence.
Solution: A robust attribution platform should use the full historical data spectrum and not limit or sample the data before algorithmic processing or visualization. By using anonymized individual journey data, calculations are based on precise consumer activity.
Challenge: Rules-based attribution models (whether pre-built or custom) make inherent assumptions in their algorithms and weighting rules that may misrepresent influence. Evaluating alternative modeling techniques via sensitivity analysis can reveal assumptions with large EVPI to prioritize improving.
Solution: A more tuned (and accurate) modeling approach uses ML-driven algorithms that are continually trained on current data. As a result, the model continuously learns and adjusts algorithmic formulae to capture changes in customer behavior that preclude outdated assumptions.
Challenge: Many models take a “one-size fits all” approach rather than customizing algorithms for different customer segments. Tailoring model logic based on known differences in journeys could better reflect true channel influence. For example, a linear model or even custom algorithm may accurately assess touchpoints in a specific industry email marketing channel and customer segment. However, the same algorithm may include assumptions that fail when used in organic search or with other segments.
Solution: Flexible attribution modeling allows marketers to view customer journeys across multiple models. By using a model-agnostic approach, segment-specific insights are surfaced as buyer paths are compared through different model visualizations.
Challenge: Shorter customer journeys tend to have more complete data. But overlooking longer decision cycles omits key influences. Often ad platforms and less sophisticated solutions enforce a short conversion window (e.g., 90 days).
Solution: B2B brands (or long-cycle B2C) with windows of 4+ months should use an unlimited window to ensure early touchpoints are not truncated from analyzed customer journeys.
Challenge: Isolating the true incremental impact of marketing amidst simultaneous changes in pricing, product, and other factors remains challenging.
Solution: Statistical enhancements like uplift modeling can help address this source of uncertainty. Large-volume businesses may combine attribution with marketing mix modeling (MMM) or incrementality testing for a holistic marketing measurement framework.
EVPI analysis allows marketers to frame a systematic data-driven plan, moving from areas of greater uncertainty to more nuanced enhancements. With a portfolio approach managing cost and complexity, the EVPI of full-path attribution can keep rising.
The quantification of EVPI provides a data-backed framework for marketers to evaluate potential investments in improving attribution insights. By revealing the maximum financially justifiable spending to enhance capabilities, EVPI analysis can methodically guide tradeoff decisions to maximize strategic value. Some key ways marketers can leverage EVPI:
If EVPI analysis reveals significant potential upside from perfect information, it helps justify expenditures to address current data limitations. For example, high EVPI may warrant investments in additional touchpoint tracking, improved CRM data, larger sample sizes, or tools to capture offline interactions. The greater the EVPI, the more spend can be motivated to improve suboptimal data inputs degrading model accuracy.
EVPI quantifies the impact of uncertainties, allowing modelers to focus improvements on assumptions, biases, and analytics techniques with the largest identified EVPI. For instance, if EVPI shows advanced modeling in Search has a $500K annual value, it may be addressed first. But if offline event tracking has just a $50K annual EVPI, it would be a lower priority for improvement. EVPI guides attention to issues that matter most.
Each potential attribution improvement initiative, whether enhancing data, algorithms, or tools, can be weighed against its cost and the EVPI to determine if it warrants investment. If the EVPI is $250K, acquiring a new $100K dataset is likely money well spent. But implementing a new $500K machine learning system would need further justification. EVPI provides an analytical hurdle rate for attribution investments.
Since achieving perfect attribution is not realistic, marketers should take an iterative approach to incrementally improve capabilities over time guided by EVPI impact analysis. Quick wins tackling high EVPI areas like additional tracking can provide returns to justify further expansions. EVPI is re-quantified as new initiatives are implemented, creating a virtuous cycle of incremental improvements through good investment decisions.
In today's complex omnichannel environment, marketing attribution is indispensable yet inherently imperfect. However, expected value of perfect information (EVPI) analysis provides a strategic framework for marketers to quantify the monetary value of achieving perfect attribution with no uncertainties. This data-driven approach supports justified investment in enhanced capabilities by revealing:
While perfect attribution is not attainable, EVPI analysis enables marketers to methodically improve cross-channel measurement and budget allocation effectiveness guided by data.
Take Galileo for a test drive. Arcalea’s revenue attribution platform passes the EVPI test with mechanisms that ensure your brand investment multiplies. By tangibly tying spending to a quantifiable ROI expectation, it builds a defendable business case for justifiable attribution investments. In navigating the complexities of today's customer journeys, EVPI transforms attribution into an asset with immense strategic value.