Listen to the podcast episode
Or listen now on Apple Podcasts or Spotify.
Key Takeaways
Platform attribution is built to flatter the platform: Meta counts view-through conversions you can’t fully turn off — an ad served into the desktop right rail counts as ’seen,’ and a booking days later counts as Meta’s win. Pinterest claimed credit a full 30 days after a view, as if anyone remembers a pin from 29 days ago. The data isn’t neutral; every platform is arguing for its own budget.
When credit claims add past 100%, run the turn-off test: Ten bookings come in; the Google agency claims eight, the Facebook agency claims seven — both right, both wrong. The honest gut-check is the nuclear option: turn a channel off and watch what happens to total bookings, seasonality caveats and all. Also notice which declined ad card triggers the same-day panic call — clients shrug when Facebook goes down and scramble when Google does, which tells you where the real pull is.
Never assume metrics you can’t control — re-measure them yearly: Industry-average homeowner retention means nothing; your number might be two years or twelve, and it’s too important to guess. Measure your newest cohort every year instead of coasting on an old assumption. One client lost 30 units in about a year to property sales — owners cashed out at double or triple what they expected the homes to be worth, and no retention program stops that.
Judge your email list on engagement, then purge slowly: A 10,000-address list you never mail is worth nothing, no matter how impressive it sounds on a sales call. But don’t over-correct: a huge list at 10% opens can out-earn a tiny list at 60%, the cohort that opens keeps rotating, and plenty of platforms now block open tracking — the subscriber who looks dead may book next year. The signals that really matter are bounces and unsubscribes, because those wreck deliverability.
Test your fees as hard as you test your rents: Everyone obsesses over pricing tools and nobody A/B tests fees, even though fees touch every single transaction. One operator Conrad talked to moved the reservation fee from $39 to $99, saw almost no drop in conversion, and roughly tripled profit on that line. Try different fee levels across sets of properties and watch what actually happens.
Unproven concepts have no data — taste has to lead: No dataset justified a Scandinavian-inspired micro-resort in Waco or shipping-container cabins in the forest, and both worked anyway. When you introduce inventory a market has never seen, past data can’t help you — you’re skating ahead of the puck on taste, demographics, and your ability to create demand. You generate the data by shipping.
Time on page lies — track intent signals instead: Four minutes on a property page might be a Florida resident who found your cabin on Pinterest admiring the photos with zero intention of traveling to Tennessee. Dates entered on a property page and checkout starts correlate with real bookings far better than any time metric. Set up tracking on the actions only a would-be guest takes.
The ad with the worst click-through rate can be your best ad: For one luxury client, putting prices starting at $2,500 per night right in the headline kills the CTR — and converts two to three times better than the plain version, because the wrong people stop clicking. Filtering is a feature. Judge ads on what the clicks become, not how many you get.
Turning off a ’failing’ channel can tank everything else: Paul shut off Facebook campaigns showing a 0% conversion rate and watched every other channel sag — the omnichannel trickle-down was real even though the dashboard said the channel was dead. Homeowner marketing especially runs on touchpoints that never earn last-click credit, postcards included. If overall performance drops when a channel dies, the data just told you what it was really worth — turn it back on.
What We Cover In This Episode
Conrad and Paul debate data-driven versus gut-feel marketing and land on nuance: when platform data misleads you (attribution, time on page, dead email lists), when you need to collect more (retention cohorts, fee tests, intent tracking), and when there’s no data at all and taste has to make the call.