We Have More Marketing Data Than Ever. So What Do We Actually Know?

By James Arnold
Rooster Chief Digital Officer

Digital marketing was supposed to solve the measurement problem.

That was part of the promise from the beginning. Unlike traditional media, digital was measurable. We could count impressions. Track clicks. Follow website behavior. Record conversions. Connect campaigns to leads. Eventually, surely, we would be able to follow the entire customer journey and know exactly which marketing investment produced the sale.

We certainly got more data. But we didn’t get the certainty we expected.

Privacy changes have removed some of the individual-level tracking marketers became accustomed to. Google, Meta, retailers and other platforms have plenty of reasons to keep their data inside their own ecosystems. Agricultural buying cycles can stretch for months and involve dealers, retailers, distributors, salespeople and other stops along the way. Even when we have the data, getting our advertising platforms, website analytics, CRM and sales systems to talk to one another can be a project unto itself.

So, I asked Scott Blessman, VP of Data Science and Analytics at Goodway Group: Have we ever been in a worse place for marketing measurement?

His answer? “Yes and no.”

We've never had it better

Let's start with the good news. Marketers have analytical capabilities today that would have been unimaginable not that long ago. Blessman points to media mix modeling as an example. Sophisticated modeling that once might have required an enormous investment and six months of work is increasingly accessible to far more organizations.

“We can understand these things more quickly,” Blessman said. “We can get that data more readily available.”

AI is accelerating that shift even further. Massive datasets can be analyzed more quickly. Relationships that would have taken analysts significant time to identify can surface much faster. Research can be incorporated alongside campaign performance. Modeling can help us understand how channels may be working together instead of evaluating every tactic in isolation.

Bethany Andrew, Executive Marketing Consultant at State of Marketing, sees the same opportunity in day-to-day analytics. We have traditionally spent a tremendous amount of time looking at the things immediately in front of us: keywords, ads, clicks and conversions.

But there is much more available.

“What about the other things?” Andrew asked. “The placements, the time of day, the day of week, all of the segmentation that we can break down.”

AI gives marketers a much more realistic opportunity to explore those larger datasets. So measurement is better. Much better.

We've also never had it worse

Here's the other side. For a while, digital marketing created the illusion that we were approaching perfect attribution. We could identify users. Follow behavior across sites. Connect ad exposure to website activity. Build increasingly complicated multi-touch attribution models designed to distribute credit across the journey. Some of that tracking went too far.

Blessman points out that privacy regulation appropriately pulled back capabilities that allowed marketers to follow individuals at a level consumers understandably found creepy.

At the same time, the digital ecosystem fractured.

“Meta doesn't want to share anything. Google doesn't want to share anything,” Blessman said.

Neither does the retailer that owns the transaction. Or the platform that owns the audience. Or the company that considers the customer relationship its competitive advantage. Why would they? Their data has value. The result is a marketing ecosystem filled with incredibly sophisticated individual systems that don't necessarily have much incentive to work together.

And in agriculture, we add another layer. The person influenced by the marketing may not buy online. They may talk to a dealer. They may buy through a cooperative. They may call a salesperson. The sale might occur months after the first marketing exposure. The person researching the product may not even be the person whose information ultimately appears in the transaction.

Trying to draw a perfectly straight line through that is difficult.

Perfect attribution maybe is the wrong goal

This is where Blessman's perspective gets particularly interesting. His answer isn't to wait for some magical future technology that finally connects every impression, click, website visit, sales conversation and transaction. He thinks we should reconsider what we're trying to accomplish.

“I think where we collectively need to go is more away from a one-to-one, this person saw this, this person saw that and then they converted, to a modeled approach where we're looking at how the channels worked together,” he said.

For years, much of marketing measurement has been built around assigning credit. Paid search gets this much. Social gets this much. Display gets this much. Email gets this much. Eventually all of those percentages are supposed to add up neatly to the sale. Real people don't behave that neatly.

Maybe someone sees a video. A month later they encounter a creator talking about the product. Then they Google the brand. They visit the website. Later they see a retargeting ad. At some point they ask a dealer about it. Three months after that, they buy. Which tactic gets the credit?

We can build a very sophisticated model to answer that question. Or perhaps we can ask a better question: Did the marketing move the business?

Start at the end and work backward

Blessman's approach is simple. Start with the business objective. Then work backward.

If the objective is sales but the transaction can't be observed directly, what signal tells us we're getting closer to a sale? Maybe it's a qualified lead. Maybe it's a dealer-locator action. Maybe it's a phone call. Maybe it's an email capture. Maybe it's increased branded search. Maybe it's a known customer returning to research another product.

“Basically backing into those metrics that actually matter,” Blessman said.

It’s obvious but not often done. Digital platforms have conditioned marketers to begin with what they can measure. We open the dashboard and there are impressions, clicks, CPC, CTR, engagement rates, video views, sessions and dozens of other numbers waiting for us. So we report them.

The availability of a metric somehow becomes justification for its importance. Blessman argues for the opposite approach.

“It's not just because you can measure it, measure it, it'll tell a good story later,” he said. “We need to be really intentional and think through what we're trying to achieve.”

Just because we can measure something doesn't mean it matters.

Platform performance isn't necessarily business performance

YJ Kim, Strategy Director at AI Digital, raises another measurement problem that becomes increasingly important as platforms automate more of our marketing. Google and other platforms are very good at optimizing toward the outcomes we give them. Sometimes too good.

Kim points to the differences among branded, non-branded, competitor and research-oriented searches. They represent different levels of intent, different costs and different expectations for performance. Blend them together and the overall campaign can appear stronger than it really is. AI optimization adds another wrinkle.

Platforms naturally move investment toward users most likely to convert. That can produce terrific-looking return metrics while concentrating spend on lower-funnel users who might have converted anyway. In other words, the platform can become very efficient at taking credit for demand rather than creating it.

“Metrics like CPC, CTR, CPA, ROAS, they still matter, but they just don't tell the whole story,” Kim said.

A dashboard can be completely accurate and still lead us toward the wrong conclusion.

Attribution isn't the same thing as incrementality

This may be where marketing measurement needs to mature next. Instead of obsessing over who gets credit for a sale, we should spend more time asking what happened because of the marketing.

Blessman describes looking at relationships between channels and business outcomes. When certain channels were in market, did we see an overall lift? When another tactic was added, did behavior change? Did website traffic increase? Did paid search activity rise? Did another meaningful signal move?

Those aren't always as satisfying as a report that says Campaign A produced exactly 42 sales. They may be more truthful.

Blessman also sees sophisticated identity-based attribution as useful, particularly when strong first-party data is available. But he increasingly views that type of analysis as validation rather than the only source of truth. Use the data you can observe. Build a hypothesis about what is working. Model relationships where appropriate. Test. Then use additional data to validate whether the story holds together.

That's a very different philosophy from believing one attribution platform is eventually going to tell us everything.

“There’s no silver bullet,” Blessman said.

AI can help put the Tinker Toys together

None of this means we should accept a permanently fragmented measurement environment. Technology is getting better at stitching pieces together.

Blessman described today's ecosystem as being broken into parts that marketers then have to pull together. CRM data is here. Platform data is there. Sales information is somewhere else. Website behavior lives in another system.

AI has obvious potential to help. Schultz sees that already happening through faster analysis. Even when marketers can't see every optimization decision happening inside an advertising platform, AI-assisted reporting can help identify performance trends sooner. That gives humans a chance to intervene. The system might notice the change. The marketer may understand the reason.

That's an important combination, particularly in agriculture, where weather, geography, crop conditions, dealer activity and dozens of other external factors can influence performance without showing up neatly in an advertising dataset.

The goal shouldn't be to make AI the final authority. It should be to use AI to make the information we already have more usable.

Your CRM may matter more than your dashboard

If there is one practical implication of all of this, it may be the growing importance of first-party data. Schultz believes marketers who haven't prioritized CRM are already behind. That isn't because CRM magically solves attribution. It doesn't.

But it gives marketers a business-owned source of information that can be connected to the signals we're seeing elsewhere. As platforms become more automated and third-party tracking becomes less reliable, that owned data becomes increasingly valuable.

The better we understand our customers, leads, sales activity and meaningful conversion points, the better we can tell platforms what we actually want.

That improves targeting, optimization and measurement.

So the next advancement is not a new dashboarding system but organizing the data we already have.

Plan measurement before you plan media

“First and foremost, it's that part about going in with intention. We need to plan to measure from the beginning,” Blessman said.

Not after the campaign launches. Not when the quarterly report is due. Not after someone asks whether the marketing worked. Before the media plan.

Start with the behavior or business outcome we're trying to create. Determine what signals would tell us whether that's happening. Make sure those signals can actually be measured. Then select the channels and tactics most likely to create them. And once you've decided what to measure, decide what you're going to test.

“When you plan to measure, you tend to just by nature plan to test,” Blessman said.

That's how measurement becomes more than reporting. It becomes part of the strategy.

So, what do we actually know?

Probably less than some dashboards suggest. And more than the pessimists think.

We may never achieve the perfect attribution model digital marketing once seemed to promise. The ecosystem is too fragmented. Platforms have too much incentive to protect their data. Consumer privacy matters. And real customer journeys are considerably messier than the diagrams we put into PowerPoint presentations. But that doesn't mean we're flying blind.

We have better analytical tools. Better modeling. Better first-party data opportunities. AI can analyze information at a speed and scale we never could. We can design tests. We can identify meaningful signals. We can see whether the business moved.

Maybe the problem isn't that we don't have enough data. Maybe we've spent too much time asking our data to give us certainty it was never capable of providing.

The next era of marketing analytics shouldn't be about measuring everything. It should be about knowing what matters, measuring that well, testing what we believe and being comfortable admitting what we still don't know.

The goal should never have been perfect attribution. The goal should be better marketing decisions.