Business leaders are living through a strange contradiction.
We have more data than any generation before us—more dashboards, more attribution tools, more channels, more “insights.” And yet, many teams feel less certain than ever about what’s actually working. Customer acquisition costs rise. Platforms shift their rules. Privacy expectations tighten. And every executive is staring at the same question, sometimes out loud and sometimes in the quiet part of the board meeting:
Are we buying growth—or are we buying noise?
This is why I was so excited to sit down with Elena Levi, Director of Product at Voyantis, joining us from Israel on a schedule that required real coordination (and real gratitude on my end). Elena is a product leader who has spent more than a decade doing something many organizations claim they want but struggle to execute: turning complex data into business impact. Her career has included predictive analytics, fraud prevention, and privacy-first product development, and she’s also a mentor and educator—an important point, because the future of AI in business will not be won by the loudest voice. It will be won by the clearest teacher.
Elena’s work sits at the intersection of three forces reshaping leadership right now:
- Data-driven decision-making (real decision-making, not dashboard theater)
- AI and machine learning adoption (with all its trust, change, and measurement challenges)
- Inclusive leadership (because perspective is not a “nice-to-have”; it’s operational leverage)
And the business problem she’s tackling is both simple and profound:
How do you acquire the right customers—at the right time—so that growth is profitable, repeatable, and resilient?
Let’s talk about what that actually means.
Growth Is Not The Goal. Value Is.
The old story of growth is volume: more leads, more users, more installs, more traffic. Bigger numbers. Bigger charts. Bigger quarterly updates.
But mature executives know the trap hidden inside that story: growth without value is just expensive movement.
Elena framed Voyantis’ purpose in language I wish every leadership team would adopt:
“We help our customers focus on the right customers at the right time.”
Not just the customers who will convert today. Not the customers who are easiest to acquire. Not the customers who spike a vanity metric and then disappear.
The customers who will create the most value six months from now, a year from now, and beyond.
That time horizon matters.
Because the real strategic question in performance marketing isn’t “How do we lower cost per acquisition this week?” It’s:
How do we invest acquisition dollars in future profitability?
That is the shift from inside-out thinking (“What do we want to sell?”) to outside-in thinking (“What do customers need, what will they do, and what will they be worth?”). Elena’s work is built on that outside-in lens—and it’s one of the clearest signals of a company operating at a higher level of maturity.
From Predictive To Prescriptive: The Loop That Changes Everything
Most executives have heard of predictive analytics. It’s become a common language: forecast churn, predict conversion, estimate LTV, and identify segments.
What Elena described goes a step further.
Voyantis doesn’t only produce predictions in a dashboard. It also delivers prescriptive analytics, meaning the system uses the predictions to help drive actions—specifically, by optimizing how acquisition and activation are handled across networks like Google and Meta.
Elena used a phrase that should be on every executive’s whiteboard:
“We’re closing the loop.”
That’s the difference between analytics that are interesting and analytics that are operational.
A prediction that sits in a report is a suggestion.
A prediction that feeds optimization systems becomes a lever.
And for organizations spending significant budgets on user acquisition, that lever can change unit economics fast. Elena referenced outcomes like ROAS improvements in the 20–30% range, which, in high-spend environments, isn’t just “better performance.” It can be a strategic unlock: more runway, more margin, more room to invest in product, more resilience against platform volatility.
But here’s what I appreciated even more than the numbers: Elena didn’t pretend this is plug-and-play.
Because it’s not.
The Messy Truth of AI Adoption: The “Moved My Cheese” Moment
Every leader wants better decisions. Every organization says they want to be data-driven.
Then an algorithm shows up and starts recommending something different than what the team has been doing—and suddenly we see the real operating system of the company.
Elena described the human side of this beautifully.
Even when the results are amazing, AI-driven decision-making can create discomfort because it changes:
- How people do their jobs
- What “good performance” looks like
- How teams are measured
- How much control they feel they have
This is the part many AI vendors ignore. But executives can’t. Because the barrier to adoption is rarely the model.
It’s the organization.
One of the most important lines Elena shared was this: champions inside the customer organization may love the product—and still struggle because they are measured by metrics that don’t align with the new approach.
Read that again.
You can deliver a better system and still lose the internal battle if incentives haven’t caught up.
This is classic change management, but with a modern twist: AI accelerates the gap between what leadership wants and what the organization is structured to reward.
If you’re a CEO or a senior leader reading this, you should treat this as a governance question:
- Are we measuring what matters—or what we’ve always measured?
- Are we rewarding short-term efficiency or long-term value?
- Do our teams have permission to change the playbook?
Because AI doesn’t just optimize campaigns. It exposes misalignment.
The Trust Gap: Why “Accuracy” Isn’t A Persuasive Argument
I asked Elena how she sets the stage with customers who are excited about AI—and those who are terrified of it. Her answer was a masterclass in product leadership.
She explained that once the contract is signed and the model is built, the real work begins:
How do you make people trust the outputs enough to act on them?
And here’s the key lesson Elena offered:
Don’t sell trust with data science jargon. Sell trust with the customer’s language and KPIs.
She said something that made me smile because it’s so true in corporate life: when you start talking about specialized model metrics—Cohen’s Kappa, precision/recall, and similar—most non-technical stakeholders won’t interrupt to admit confusion. They’ll nod. They’ll say yes. And they’ll walk away uncertain.
That uncertainty becomes hesitation.
And hesitation kills adoption.
So Elena does what excellent leaders do: she meets people where they are. She demonstrates model value using the metrics the organization already cares about:
- ROAS
- CPA
- conversion rates
- downstream value measures
That is not “dumbing it down.” That is translation—one of the most underrated executive skills in modern business.
The Privacy-first Advantage: Doing More Without Holding PII
We cannot discuss AI and growth without addressing privacy. It’s not just regulation; it’s market expectation. The last several years have forced companies to confront a hard truth:
The era of unlimited tracking is over.
Elena emphasized that Voyantis can deliver predictions without holding personally identifiable information (PII). That matters—for compliance, for customer trust, for risk reduction, and frankly for leadership peace of mind.
Instead of “following Lisa around the internet,” the system learns from patterns of behavior, flows, metadata, and actions that indicate likelihood to convert, likelihood to churn, or expected future value.
Executives should appreciate the strategic value here:
Privacy-first design doesn’t only reduce risk—it expands your addressable market. It enables you to operate globally with fewer constraints, it supports GDPR compliance, and it positions you as a company that can innovate responsibly.
And in today’s landscape, responsibility is not a slogan. It’s a differentiator.
Why Predictive is Hard (and why most businesses never get there)
I told Elena something I believe strongly: truly predictive businesses are rare. Many companies are responsive. Some are proactive. Very few are consistently predictive in a way that impacts strategy.
Predictive capability requires:
- Clean enough data
- Stable enough funnels to model (or expertise to model the messy ones)
- Organizational willingness to act on forecasts
- Continuous monitoring so models don’t drift as markets and products change
Elena made a critical point that every executive should internalize:
The barrier to building a model is dropping.
The barrier to operating a model reliably is still high.
This is where product leadership meets infrastructure reality.
Elena described two core areas of value that become more important as model-building commoditizes:
- Orchestration: keeping models live, updated, and aligned as products change, markets shift, and data evolves
- Prescriptive execution: translating model outputs into actions that platforms can actually use—without breaking campaigns
This is the unglamorous work that wins.
Because the cost of a model failing quietly isn’t just a performance dip—it can be budget waste at scale.
The Outlier Trap: Why More Revenue Can Break An Algorithm
Then we got into one of my favorite moments of the conversation: outliers.
Elena gave an example: imagine your best users typically generate a predicted lifetime value of around $200. Then you acquire a user predicted at $10,000—an extreme outlier.
Every business owner’s instinct is immediate:
“Give me more of those.”
Of course. But Elena explained a counterintuitive truth about platform algorithms:
If you feed the network a signal that’s too extreme, it can “lock on” to it in a way that suffocates the campaign. The algorithm starts searching only for $10,000 users—who may be extraordinarily rare—and your acquisition pipeline collapses.
This is where experience becomes strategy.
The goal isn’t just to identify high-value users. The goal is to communicate value to ad networks in a way that improves learning without creating pathological behavior.
Elena described this as a matter of calibration and ranking—still signaling “top user” value, but compressing variation so the network optimizes effectively.
Executives should read that as:
The difference between AI theory and AI performance is operational nuance.
That nuance is where competitive advantage lives.
Inclusive leadership: Why Elena’s Approach Scales Across Organizations
Elena also represents something we need more of in technology leadership: inclusive, educational product thinking.
She doesn’t posture. She explains. She respects that different stakeholders have different languages. And she recognizes that adoption is emotional as much as it is rational.
That matters because AI tools don’t fail in the lab—they fail in the meeting.
They fail when:
- The champion can’t defend the change
- The KPI system punishes experimentation
- The organization doesn’t know how to talk about trust
Elena’s approach isn’t just about building a product. It’s building the conditions for the product to be used.
That is leadership.
Executive Takeaways: What To Do Next
If you lead a business spending meaningfully on acquisition—or you plan to—Elena’s journey points to a few strategic moves:
- Shift from short-term acquisition efficiency to long-term customer value.
Make LTV and retention economics central, not peripheral.
- Treat AI adoption as change management.
Align incentives, dashboards, and expectations to support new decision systems.
- Demand privacy-first architecture.
Reduce risk while increasing global scalability.
- Invest in orchestration, not just modeling.
Models drift; businesses change. Operational excellence is the moat.
- Use the customer’s language, not the data scientist’s.
Trust is built through relevance, not complexity.
Closing: The Future Belongs to Leaders Who Can Close the Loop
What Elena’s journey really highlights is a new standard of leadership: the ability to connect data to decisions and decisions to outcomes—without sacrificing trust, privacy, or people.
Predictive analytics helps you see the future.
Prescriptive analytics helps you act on it.
Privacy-first design keeps you credible while you do it.
And inclusive leadership makes adoption possible.
That’s how modern organizations grow: not by buying more traffic, but by earning better customers—at the right time—for the right reasons.
And in a world where attention is expensive, and trust is fragile, that might be the most sustainable advantage of all.
Listen to the full episode on C-Suite Radio: Disrupt & Innovate | C-Suite Network
Watch the episode: DI 144 The Future of Predictive Analytics.
This article was drafted with the assistance of an AI writing assistant (Abacus.AI’s ChatLLM Teams) and edited by Lisa L. Levy for accuracy, tone, and final content.
