Inside the Blurb · The MarTech Matrix

Why Your Marketing Data Is Lying to You — And How to Fix It

The big picture: Most marketers can’t tell you whether their marketing actually caused a sale — which is exactly the gap incrementality testing strategies are built to close. Two years before Pranav Piyush ever bought anything from True Classic, he was seeing their ads everywhere — YouTube, Instagram, even the mall. He laughed, moved on, and kept shopping at Uniqlo. Then Uniqlo lost him as a customer, and True Classic was the first brand that came to mind.

No last-click attribution model captured that. No multi-touch dashboard told True Classic that their patience had paid off. The click that triggered the purchase looked like it came from an Instagram ad — but the real conversion happened over two years of brand building.

“If you’re the growth marketing manager at True Classic,” Pranav Piyush told us on Inside the Blurb, “you’re optimizing to the wrong signal.”

Why it matters: Piyush is CEO and co-founder of Paramark, a measurement and marketing mix modeling platform he started in 2022 after leading growth at PayPal, Dropbox, Magento, and Build.com. In this episode, he breaks down why last-click and multi-touch attribution both miss the real drivers of growth, what incrementality testing actually measures, and why the dashboard is only 20% of the problem.

The “Spend vs. Invest” Problem

There’s a word Pranav refuses to use: spend. “It’s not about spending. It’s not about better spending. It’s all about investments. This is a capital allocation game.”

When marketers think “ad spend,” they’re thinking about money out the door. When they think “marketing investment,” they’re thinking about return. The shift in framing changes how you measure — because investors don’t just track what they put in, they track what came back and when. Marketing is no different. The problem is most measurement tools only look at the final click, not the full journey that made that click possible.

---

Why Last-Click Attribution Gets It Wrong

Pranav shares another example that makes this concrete. He was in the market for skincare products and turned to Google Gemini for research. Gemini surfaced three options. He chose CeraVe.

But the real reason he chose CeraVe had nothing to do with Gemini. Two years earlier, CeraVe had run a guerrilla Super Bowl campaign featuring Michael Cera — complete with staged store appearances and a viral reveal. That campaign planted a brand memory. When Pranav needed a skincare brand, CeraVe was already top of mind.

“If you’re the growth marketing manager at CeraVe, you’re like: Gemini got me that click. But that wasn’t it at all.”

The buying journey is not linear. Consumers enter and exit markets over months and years. Brand impressions accumulate silently. Last-click attribution captures none of this, and multi-touch attribution still assumes a clean journey that rarely exists in the real world.

---

Incrementality Testing Strategies

Here’s the most common misconception in measurement: incrementality testing equals holdout tests. Many marketers assume that means voluntarily leaving money on the table — withholding ads from an audience to create a control group, while the business is in “go, go, go” mode. Pranav pushes back hard on this. Every brand needs Incrementality Testing Strategies.

“Incrementality testing strategies are not just about holdouts. That is one very, very small part of what incrementality testing is.”

Done right, incrementality testing asks a more fundamental question: when I invest in this channel, do my sales actually go up? When I pull back, do they drop? The goal is to isolate the true causal relationship between marketing investment and revenue — not just correlation. That’s the difference between knowing what you spent and knowing what drove results.

---

The 80/20 of Measurement — It’s Not the Dashboard

This is Paramark’s clearest point of differentiation. Most vendors sell you a dashboard. Paramark argues that the dashboard is only 20% of the problem.

“The 80% of the problem is the ongoing engagement, education, change management, process management, and critical thinking that’s going to get you to your growth goals within your CAC efficiency guardrails.”

That’s the rationale behind Paramark’s Growth Advisory Services — a human-in-the-loop layer where their team works directly with customers on what tests to run, how to interpret results, and what to do next. It’s the difference between having data and knowing what to do with it. As Pranav puts it: “If you’re just a dashboard company, it’s not going to be very good for you in the long term.”

---

How Paramark Works

Paramark’s incrementality testing strategies & workflows follows a virtuous cycle that moves from experiment to model to output and back to experiment again.

1. Get your data right. Most brands don’t have all their data in one place. Amazon data, retail flat files, Meta campaigns split into five sub-buckets — before any model can work, all of it needs to be normalized, cleaned, and validated. Paramark says this step alone typically takes weeks and is the most underestimated part of the process.
2. Design and run incrementality experiments. Paramark uses ML models to identify the right geographies, budgets, and time windows for each test. A TV test, for example, might pulse up investment in select regions while keeping others at baseline — then measure lift not just during the test but weeks after, because many marketing effects are lagged.
3. Feed results into a Marketing Mix Model (MMM). Experiment results calibrate an MMM that shows how each channel and sub-channel contributes to incremental vs. baseline sales. Baseline reflects built-up brand equity and organic demand — the stuff that would have happened anyway.
4. Scenario plan and reallocate. If one channel shows a CAC of $86 and another shows $2,000, that’s the signal to reallocate. Run a test to confirm. Rinse and repeat every month, every quarter, every year.

---

What AI Actually Does (and Doesn’t Do) for Measurement

There’s a lot of AI hype in the measurement space right now. Pranav cuts through it with a useful frame. “LLMs are word prediction machines. They’re really good at writing code and writing prose. But they’re not necessarily good at writing math.”

Incrementality testing and MMM are mathematical disciplines. AI won’t replace the underlying models — the math is still the math. What AI will do, and is already starting to, is make those models more accessible. In the next 12 months, Pranav expects every incrementality platform to have natural language interfaces: ask a question about your MMM output in plain English, get an answer. Agentic workflows will follow, automating test setup and insight surfacing without requiring a human to pull every lever.

But the models themselves remain the source of truth. “You can build a dashboard with Codex and Plotly in a weekend. That doesn’t mean you know how to read the data.”

---

What Good Measurement Vendors Look Like

Pranav was a buyer before he was a founder. His framework for evaluating vendors in this space:

• Speed of response is a proxy for customer success. “If your vendor isn’t responding fast during the sales process, it’s going to be 10 times worse after.” A slow SDR is a preview of what customer success looks like post-contract.
• Demand pricing transparency. “I don’t know why we’ve conditioned the whole MarTech world to never reveal prices.” If a vendor won’t give you a ballpark before a sales call, that’s a signal — and not a good one.
• Lean on your network. Most buying decisions in this space get made in Slack communities and group chats. Before signing, talk to someone who’s actually used the platform in production.

And on Paramark’s side: transparency about the limits of measurement. “We never tell customers that you can get to double-digit decimal precision on an incrementality test result. Anyone saying that is promoting a falsehood.”

---

The Future: When Measurement Meets Media Buying

Pranav sees measurement and media buying converging in the next two years. Right now, the coordination overhead between an incrementality platform, an agency, and an in-house team is painful. He sees that friction disappearing.

Not fully autonomous — Pranav is explicit on this. “There’s still going to be a human in the loop, just like there’s a human in the loop in any Claude interaction.” But the administrative layer — aligning on the next test, setting it up, reading results, briefing creative teams — will get automated. The brands building measurement infrastructure now will be positioned for that future. The ones still running on last-click attribution will be starting from scratch.

---

About Pranav Piyush and Paramark

Pranav Piyush is the CEO and co-founder of Paramark. Before starting the company in 2022, he spent his career leading growth at PayPal, Dropbox, Magento, and Build.com — where he came to believe that marketing was “highly misunderstood, underlevered, underresourced, and understaffed” across the industry.

Paramark serves advanced, high-growth brands spending millions per month across complex channel mixes — DTC, Amazon, retail — where incrementality testing and measurement are hardest and the stakes are highest.

If you’re evaluating a measurement or attribution partner, check out Paramark’s profile on Blurbs for case studies, vendor Q&As, and the key questions you should be asking before you take a demo call.

---

More Conversations on The MarTech Matrix