Project Nexus: B2B SaaS ROAS Up 18% in 2026

In the dynamic realm of digital marketing, understanding the complete customer journey across various devices remains a persistent challenge. Our recent campaign aimed squarely at this, employing advanced cross-device tracking techniques to unify fragmented user profiles and deliver a truly personalized experience. But can even the most sophisticated data aggregation truly paint a holistic picture of consumer behavior?

Campaign Teardown: “Project Nexus” – Unifying the B2B SaaS Journey

I’ve spent years in performance marketing, and one of the biggest headaches has always been the fractured view of the customer. A user might discover a product on their work laptop, research it on their personal tablet, and finally convert on their mobile phone during a commute. Without connecting those dots, every touchpoint feels like a new customer, leading to wasted ad spend and disjointed messaging. That’s precisely what we set out to fix with “Project Nexus,” a three-month campaign for a mid-market B2B SaaS client specializing in project management software.

The Strategic Imperative: Why We Needed Unified Data

Our client, a company based out of the Atlanta Tech Village, had a strong lead generation funnel but struggled with conversion rates for users who engaged across multiple devices. Their existing attribution model was heavily last-click focused, meaning valuable early interactions on different devices were often ignored. We knew that a significant portion of their target audience, IT managers and project leads, used multiple devices throughout their day. Our primary goal was to create a unified data view, allowing us to serve more relevant ads and attribute conversions more accurately. This wasn’t just about efficiency; it was about understanding the customer journey itself, a journey that rarely stays confined to a single screen. We hypothesized that by stitching together these profiles, we could reduce our Cost Per Lead (CPL) and significantly boost our Return on Ad Spend (ROAS).

Budget, Duration, and Core Objectives

Project Nexus ran for 90 days, from January 1st to March 31st, 2026, with a total advertising budget of $250,000. Our core objectives were clear:

• Increase overall conversion rate (free trial sign-ups) by 15%.
• Decrease CPL by 10% through more intelligent retargeting.
• Improve ROAS by 20% by attributing multi-device conversions more accurately.
• Achieve a 70% deterministic match rate for logged-in users.

The Strategy: Deterministic and Probabilistic Matching in Tandem

Our strategy was a blend of deterministic and probabilistic cross-device tracking. For deterministic matching, we worked with the client to implement hashed email addresses from their CRM and website login data. This allowed us to identify users who logged in from different devices with a high degree of certainty. For probabilistic matching, which is inherently less precise but broader in reach, we partnered with a data clean room provider. This provider used various signals like IP addresses, Wi-Fi networks, browser types, and behavioral patterns to infer connections between devices. I’ll admit, the initial integration was complex; it required significant engineering effort on the client’s side to ensure proper data flow and anonymization, particularly concerning compliance with privacy regulations like GDPR and CCPA. We spent almost a month just on the technical setup before launching any ads.

Attribution Model Shift

Crucially, we moved away from a last-click attribution model. We implemented a time-decay attribution model, giving more credit to recent interactions but still acknowledging earlier touchpoints. This allowed us to see the influence of a mobile ad seen in the morning on a desktop conversion later that day.

Creative Approach: Contextual Relevance Across Screens

Our creative strategy hinged on contextual relevance. If a user visited a specific feature page on their desktop, our mobile retargeting ads would highlight that exact feature, perhaps with a testimonial or a limited-time offer related to it. We developed three distinct ad sets:

1. Awareness (Top-of-Funnel): Broad targeting on LinkedIn and Google Display Network, focusing on problem/solution messaging.
2. Consideration (Mid-Funnel): Retargeting users who visited the website but didn’t sign up, using case studies and feature comparisons.
3. Conversion (Bottom-of-Funnel): Highly personalized retargeting for users who started a free trial or added items to a cart but abandoned, offering incentives like extended trial periods or dedicated support.

We used a mix of video ads (short, punchy explainers for mobile), static image ads (demonstrating UI on desktop), and carousel ads (showcasing different features). The ad copy was dynamically tailored based on the user’s last known interaction, a capability unlocked by our unified user profiles.

Targeting: Precision Through Data Unification

With our cross-device capabilities, our targeting became significantly more precise. We could create custom audiences that included users who, for example, clicked on a LinkedIn ad on their work computer but then visited the pricing page on their home tablet. This allowed us to:

• Exclude converted users across all devices, preventing ad fatigue and wasted spend.
• Retarget warm leads with messages relevant to their specific stage in the funnel, regardless of the device they were currently using.
• Build more accurate lookalike audiences based on truly unified profiles, rather than fragmented device IDs.

We specifically targeted IT decision-makers in companies with 50-500 employees, primarily in the Southeast region, focusing on cities like Atlanta, Charlotte, and Nashville. This regional focus allowed us to test the strategy before a broader rollout.

What Worked: The Power of Seamless Journeys

The results were compelling. Our deterministic matching, using hashed login data, achieved a 72% match rate for users who logged into the client’s platform across two or more devices. This was slightly above our 70% target, which was fantastic. This high match rate gave us confidence in our unified profiles.

The most significant win was the substantial increase in ROAS, jumping from 2.8x to 3.3x, an 18% improvement. Our CPL also saw a healthy reduction of over 14%, dropping from $32.00 to $27.50. This wasn’t just incremental; it was a fundamental shift in efficiency. I had a client last year who was struggling with similar issues, and their initial skepticism about the complexity of cross-device tracking was palpable. Seeing these numbers, especially the CPL reduction, usually makes even the most cautious CMO sit up and take notice.

The improved CTR (2.3% from 1.8%) indicated that our more relevant, context-aware ads were resonating better with the audience. People respond when they feel understood, not just targeted. One specific example involved a user who viewed a demo video on their work desktop but didn’t sign up. Two days later, they received a mobile ad offering a “quick start guide” to the software, emphasizing its ease of use on the go. That user converted within hours. This kind of seamless journey was simply impossible with our old, siloed approach.

According to a recent eMarketer report, global digital ad spending is projected to reach over $700 billion by 2026, making efficient targeting and attribution more critical than ever. Our results align with the growing need for sophisticated, data-driven strategies.

What Didn’t Work: The Limitations of Probabilistic Matching and Privacy Concerns

While deterministic matching performed admirably, our probabilistic matching component was less successful than anticipated. It contributed to broader reach but had a lower confidence score for individual user connections, leading to some instances of irrelevant ad serving. We saw a higher rate of negative feedback on ads served via probabilistic matches, likely due to less accurate profile stitching. It’s a trade-off: precision versus scale. For B2B, where purchase cycles are longer and decisions more considered, I’ve found that precision almost always trumps sheer volume.

Another challenge was managing user consent. As we expanded our data collection for cross-device identification, we had to be incredibly transparent with our privacy policies and consent banners. This led to a slight dip in initial website engagement from users who opted out of certain tracking cookies, a necessary evil in today’s privacy-conscious environment. We had to iterate on our consent language multiple times to find the right balance between clarity and user experience. This is one of those “nobody tells you” moments in marketing: the technical challenge is only half the battle; the legal and ethical considerations are just as, if not more, complex.

Optimization Steps Taken

Mid-campaign, we made several critical adjustments:

1. Refined Probabilistic Signals: We tightened the parameters for our probabilistic matching, focusing on stronger signals like device models and operating systems rather than just IP addresses. This reduced the volume of matches but increased their accuracy.
2. A/B Testing Consent Banners: We continuously A/B tested different consent banner designs and wording to improve opt-in rates while maintaining transparency. We found that clearly explaining the benefits of personalized ads (e.g., “help us show you more relevant content”) significantly improved user acceptance.
3. Increased Spend on High-Performing Segments: We reallocated 15% of the budget from broad awareness campaigns to our high-performing retargeting segments, particularly those identified through deterministic matching.
4. Adjusted Frequency Capping: Initially, we had a relatively high frequency cap. We reduced it to prevent ad fatigue, especially for users who were identified across multiple devices. No one wants to see the same ad five times a day, regardless of the device.

These optimizations, implemented during the second month of the campaign, were instrumental in pushing our CPL down further and boosting the overall ROAS. We used Google Ads Performance Max and LinkedIn Campaign Manager to implement these changes dynamically, leveraging their automated optimization features where possible.

Looking back, the success of Project Nexus wasn’t just about the technology; it was about the strategic shift in how we viewed the customer. It’s no longer about devices; it’s about people moving fluidly between them. Ignoring this reality means leaving money on the table and delivering a subpar user experience. Any marketer who isn’t grappling with cross-device tracking and the unification of user profiles right now is already behind the curve.

Conclusion

Achieving a truly unified view of the customer journey across devices is no longer a luxury; it’s a necessity for competitive digital marketing. By strategically combining deterministic and probabilistic matching, advertisers can significantly improve campaign efficiency and ROAS. Focus relentlessly on data privacy and transparent consent while embracing robust attribution models beyond last-click to unlock the full potential of your marketing efforts.

Derek Nichols

Principal Marketing Scientist M.Sc., Data Science, Carnegie Mellon University; Google Analytics Certified

Derek Nichols is a Principal Marketing Scientist at Stratagem Insights, bringing over 14 years of experience in leveraging data to drive strategic marketing decisions. Her expertise lies in advanced predictive modeling for customer lifetime value and churn prevention. Previously, she spearheaded the marketing analytics division at AuraTech Solutions, where her team developed a proprietary attribution model that increased ROI by 18%. She is a recognized thought leader, frequently contributing to industry publications on the future of AI in marketing measurement

Credentials 14+ years experience