Ondra
Travel discovery has moved onto social, and the accounts that do it well still convert almost none of it. We built Ondra to be the commercial layer behind the post: the page, the capture, and the follow-up that a good social account has never had.
What it is
Ondra is a sales engine for travel sellers that converts a seller’s own photography and a typed offer into a social post, a matching landing page, and a follow-up email sequence, turning social traffic into captured, nurtured demand instead of passing engagement.
Social is now where travel gets discovered, which means every property and advisor is already a demand-generation channel whether they think of themselves that way or not. The good accounts have real audiences and photography better than the OTAs use. What they do not have is anything behind the post: no page about the offer they just advertised, nothing recording who responded, no follow-up for the people who did. So the interest either evaporates or completes somewhere that has built all of that.
Ondra takes one typed offer and generates the whole sequence from the seller’s own photography: the post, a landing page continuing that exact offer, and an email sequence shaped by how the visitor described themselves.
Social became the discovery layer for travel. Nobody built the commercial one.
Discovery already moved, and it is not close
Survey data on this is consistent to the point of being boring.
For a travel business this stopped being a channel decision some time ago. The market forms on social now, which makes presence there an operating requirement rather than a marketing tactic. It is the equivalent of having a phone number in 1990. The interesting part is the next number: roughly 64% of travelers say they would be comfortable booking directly through a social platform. The willingness to transact is already there. It is the mechanism that is missing.
The operational problem is consistency
The difficulty of running a travel social account is not thinking of something to post. It is posting continuously, forever, without a team. The platforms price this explicitly.
Individual content decays fast on top of that. Instagram posts lose most of their reach within four days, TikTok within a week. So the treadmill never stops, and it punishes the exact pattern a small operator falls into: a strong month, then a busy season, then three quiet weeks that cost more than the strong month earned. Consistency is the thing that breaks people, and it breaks them regardless of how good their work is.
The accounts that win at it still do not make money from it
This is the part that convinced us there was a company here. Plenty of travel operators are genuinely good at social. Real followings, real engagement, photography that outclasses anything on an OTA listing page. And their commercial results are close to indistinguishable from operators who post badly, because a great post and a mediocre post both terminate in the same place, which is nowhere.
Here is the sequence as it exists for most sellers, including the successful ones:
Neither audience nor content is the constraint. The missing piece is everything between the post and the transaction, and its absence is why a 40,000-follower boutique property and a 400-follower one can have similar direct-booking numbers.
What the missing layer costs
When the demand a seller created completes on an OTA, it does so at a price:
Cost of acquiring the same guest
Share of reservation value, independent properties
Upper bound of each published range on a shared 0–30% scale. Industry estimates for independent hotels, not measurements from Ondra.
For advisors the same failure shows up as absence rather than expense. An advisor rarely loses a booking to a named competitor; the engagement simply ends, with no lead recorded and no way to tell which post did the work. That version is harder to notice, which is part of why it goes unaddressed for years.
Why this has not been built
Almost every tool aimed at this market assumes the problem is producing content, because that is the visible symptom. So the market filled up with caption writers, image generators and scheduling tools, none of which touch the part that actually leaks.
The harder problem, and the one we went after, is generating a set of pieces that agree with each other. A post, a landing page and an email sequence built from one offer and one photo library, still telling a single coherent story by the time a visitor reaches the third email. Producing any one of those is close to free now. Producing all of them in one pass, consistently, so nothing resets to generic at a handoff, is the actual work. It is a retrieval and matching problem rather than a writing one.
What changed how we built it
Good accounts monetize like bad ones
We expected audience size to correlate with direct bookings. It barely does. Operators with strong followings and genuinely good content convert at rates close to operators with neither.
That told us we were building a commercialization product, not a content product.
Consistency is the constraint
Nobody I spoke to was short of things to post. They were short of the ability to keep posting through a busy season, and the platforms charge for that gap specifically.
So output volume had to be near-free, or the product would fail exactly when its users got busy.
The willingness to buy is already there
Around two thirds of travelers say they would book directly through a social platform. That reframed the problem for us. This is not a demand gap or a trust gap.
The gap is plumbing, and plumbing is a far more tractable thing to build.
Coherence is the hard part
Any single asset is cheap now. A post, a landing page and three emails that still agree with each other, on the same offer with the same photographs for the same audience, is not.
Most of our engineering went into holding the chain consistent rather than into the writing.
The loss happens at the handoffs
A specific post qualifies its audience before anyone clicks. Sending that qualified visitor to an undifferentiated booking page discards the qualification.
People look for funnel problems at the top. In this category they are mostly in the seams.
Generated imagery is the wrong bet in travel
Travel is one of the few categories where the customer physically audits the marketing on arrival. A room that does not exist becomes a refund and a review.
We compose from the seller’s own photographs and generate no imagery, which is harder and, we think, the only durable position.
Telling small sellers to do SEO is bad advice
Search rewarded press coverage and budget, and early evidence suggests AI answer engines reproduce that bias toward flagship brands and OTAs.
A boutique property cannot out-publish that. Social origin with owned follow-up is a fight they can win, so that is where we pointed the product.
One typed offer, one complete sequence
The photo library becomes searchable
The seller uploads their own photographs. Ondra reads and indexes them automatically across property, room, destination, activity, season, light and mood. An archive that was previously a folder can be queried in plain language.
The offer is typed, not configured
No integration to set up, no field mapping, no template to choose. The seller writes the offer the way they would say it to a colleague: South Beach Insider, $189 a night, Labor Day.
The commercial layer generates with it
A post built from the photographs that match the offer, a landing page continuing that same offer rather than a general booking flow, lead capture on that page, and an email sequence shaped by how the visitor described themselves. Generating them in one pass is what keeps them consistent.
The product itself
Moments from the build, recorded as they run.
Photographs come in, get cropped, captioned and confirmed.
The archive reads itself. Every frame carries the tags that make it findable later.
The offer gets described in plain words. No fields, no template.
The public page edits live, so the seller sees what the visitor will see.
What that produces
From one offer and one library, Ondra generates several versions simultaneously. The offer and the photographs are identical across all of them; nothing the visitor reads is.
Identical across all three
What this visitor reads
Your romantic South Beach getaway starts here
Email oneEvenings meant for two in South Beach
What this visitor reads
Your dynamic South Beach adventure awaits
What this visitor reads
Your solo Miami adventure begins here
What we’re building next
At the moment the seller tells Ondra what to promote. The version we’re working toward reads the inventory instead, so the system can see which dates are soft and which rooms are unsold and generate against them without being asked. Marketing a catalog then stops being a task someone schedules and becomes something that runs, including across the long tail of offers that never get attention because nobody has time to reach them.
Who it’s for
The same engine serves four groups at different entry points. The product does not change between them. The way in does.
Where the cost of the missing layer is measurable and the case is arithmetic rather than persuasion. Every booking taken directly avoids a commission on a guest the property already found.
The largest group by count, and self-serve. Their version of the problem is invisibility rather than commission: engagement they cannot attribute and never capture.
Retention is usually healthy and social usually is not. Most have paid an agency or a contractor already and can tell the difference between output and results.
Catalogs with hundreds or thousands of offers, where conversion budget goes to the booking flow and almost none reaches the inventory feeding it. They know the gap exists and will not build this internally.
Working on this problem?
If you run a property, an advisory practice or a portfolio and this is your gap, I want to hear how it shows up for you.
