Hess Persson Estates

I grew up on a winery property in Napa and spent a good part of my early life in the vineyard and the cellar, where the feedback loop runs about a year. You make a call in spring and find out whether you were right the following fall. Digital marketing gave me the opposite problem. I can see what happened by lunch. The hard part was never getting the number. It's deciding what to do about it.

That gap is where most marketing teams actually live. The dashboards refresh, the numbers are broadly right, everyone nods, and nothing changes on Monday. I've built my share of reports nobody acted on, including some I was proud of.

Turning data into growth opportunities

One small habit changed this for me. Before pulling a report, I write down what I'd do differently depending on what it says. If I can't name two different actions, I don't pull it.

Our email program is the clearest example. It had grown to a lot of sends without anyone asking whether they all still deserved to exist. Rather than write another performance summary, I went through every campaign from a full half-year and gave each one a verdict: recycle, optimize, sunset, or one-time only. The output wasn't a chart. It was a calendar with fewer and better campaigns on it, and a written reason for every cut.

The signals I actually watch

I weigh signals by how well they predict revenue, not by how easy they are to measure.

First-party behavior comes first. Purchase history, club retention and cancellation, how reservations convert in the tasting room, and email engagement broken out by acquisition source. When we dug into list health, the thing that mattered most was discovering that one historical bulk import had skewed our engagement baselines for months. No platform metric was going to surface that.

After that, I look for agreement between systems that have no reason to agree. Our paid search numbers looked strong on CTR and cost per acquisition, but the figure I trusted was the one that lined up with reservation data from our booking platform. A platform reporting its own success is a hypothesis. Two independent systems telling the same story is closer to evidence.

Connecting eCommerce to broader growth

The commerce site is the cheapest research environment a company owns. Every product page, subject line, and offer is a live test of what language makes someone act, and the answer comes back in hours instead of quarters.

“A platform reporting its own success is a hypothesis. Two independent systems telling the same story is closer to evidence.”

We treat those findings as inputs for teams not doing eCommerce at all. Purchase and visitation data have shaped our hospitality outreach strategy. DTC engagement patterns inform which stories the trade side leads with. It runs the other direction, too, and that's easy to forget: wholesale coverage, tasting room capacity, and inventory all limit what eCommerce should be pushing in a given month.

Where translation breaks down

Our eCommerce platform, analytics, email tool, booking system, and every ad network each keep a defensible version of the truth, and they don't match. I gave up on reconciling them. One system is the source of record for revenue, the rest are directional, and I say that out loud so nobody relitigates it.

The constraint I think about most is volume. In a luxury business, statistical significance is often out of reach, so I've learned to trust direction and repeatability instead. Does the pattern hold across three cycles? Does it survive a change in season, price point, and channel? That's a weaker standard than a clean test, and it beats waiting for a sample size that isn't coming.

Underneath all of it, a correct finding with no narrative doesn't travel. Most of my analysis ends in a one-page brief for people who won't open a spreadsheet, and that's the step where an insight turns into budget.

Building a data-driven culture

Agree on definitions before building dashboards. Most arguments about whether a campaign worked are really arguments about what counts as a conversion.

Give every person one number they own. Shared responsibility for a metric turns into nobody's responsibility. Publish what you killed. People start trusting data when they watch it retire someone's favorite tactic, ideally the leader's. That sunset list bought more credibility internally than any report on a win. Protect time for questions nobody assigned. Reporting cadence expands to fill the calendar, and reporting isn't analysis. Almost everything that changed our strategy this year came from an unscheduled question.

Mostly, reward curiosity about the mechanism. Whether a number went up is rarely the interesting part. The teams I've seen get good at this are the ones still asking why long after the number has been reported.