Pivoting a company: from Luckey to Airbnb Co-Hosting

The story behind

Yesterday, on October 17, 2024, Airbnb announced the launch of its new Co-Host Network. Hosts can now find someone on the platform to manage their place on their behalf. This can include cleaning, check-in, guest management, pricing—everything related to managing a home on Airbnb.

This announcement concerns the international deployment of the service along with some redesign, but the service has actually been available since 2021 in France and Spain. Before that, it was something else: Luckey, which was acquired by Airbnb in December 2018.

I joined Luckey in January 2017, prior to the acquisition. At that time, and for another two years, we provided homeowners with an all-inclusive short-term rental management service on Airbnb and other platforms. When Airbnb bought Luckey in December 2018, we didn’t know what we were about to become, and we continued for three years with the same model before we decided to pivot.

That’s the story I’m going to tell now.

The old and new business models

Old Model

Luckey provided a comprehensive management service to hosts (meaning that hosts had nothing to do except collect their revenue) based on two pillars:

1. Offline Operations: Luckey partnered with on-the-ground “city managers” in each location, who handled every physical aspect of the service: cleaning, check-ins, key management, maintenance (repairs, linen replacement, etc.). In most cases, once they managed more than 10 homes, they would hire people for each task. They benefited from tools we created to make their job much easier (task assignment, planning, reminders, etc.).

2. Online Operations were managed internally by the parent company based in Paris, and included listing creation, pricing, guest support, and host support.

City managers were like “franchisees” benefiting from the tools and the brand. 50% of the revenue went to them, 50% to us.

New Model

Airbnb Co-Hosting provides a marketplace where hosts can find someone (an individual or a company) to manage their place. Airbnb does not directly manage anything for the hosts but provides some tools to assist the co-hosts on the platform.

What was going wrong?

In theory, the old model was promising: city managers could easily launch a business with our support, benefit from leads and customers immediately without investing in advertising, and earn money from day one. This actually convinced many of them to join us. For us, it was an opportunity to scale the company rapidly because we didn’t have to inject investments to launch a new city; we just had to find a partner there.

In practice, though, our business model struggled to scale. Despite Airbnb's significant investment in advertising, acquisition was like a leaky bucket. The churn rate was almost as high as the acquisition rate, and we couldn’t scale beyond 3,000 hosts.

The reasons were multiple, but here are a few hints:

• Inefficient operations: Short-term rentals involve managing the unexpected, which happened almost daily. When a guest or host had an issue, they’d contact online support, who often redirected them to city managers because they had no information about on-site happenings. Thus, Luckey’s promise to city managers was limited. Guests and hosts were frustrated, feeling like we were passing the buck (which we were), and the quality of service was declining.

• Unclear responsibilities: City managers still had to handle online tasks despite these not being part of their mission. Also, the most loyal customers (hosts) were in cities where partners built strong relationships with them, without it being in their job description. Yet again, the online team efforts were insufficient, despite their competence.

• Insufficient revenue for partners: For all the work city managers had to do, 50% was not enough to pay their employees and make a living. They hoped managing a certain number of homes would bring profitability, but that rarely happened because expenses increased almost as much as revenue.

• Disengagement: At some point, some city managers didn’t want to go the extra mile, leading to reduced service quality. The job of a city manager was hard: they woke up early to ensure tasks were well-assigned and handled unassigned tasks. They stayed awake until late in the evening in case a guest had an issue.
  Another reason for disengagement was that city managers didn’t own the contract with customers (hosts). They didn’t value business assets and goodwill despite their significant efforts as entrepreneurs. Consequently, they felt all this work wasn’t paying off much—not just through revenue but also valuation.

• Hard to scale: We were dealing with real homes that partners had to onboard, and you can’t just launch a city and manage 50 properties in one go. A city manager would take a year to get to 20. You might think we could hire several partners in the same city, but it was challenging because we’d have two direct competitors under the same brand, like competitive franchisees. It’s like opening a McDonald’s next to another McDonald’s: it doesn’t work. And city managers wouldn’t accept that even when overwhelmed with properties; we tried it in Lyon, and it was a failure.

In short, the relationship was slightly unbalanced: city managers were offering more than they received, and cracks quickly appeared.

All of this led to the service quality dropping below 4.4 stars during certain periods, and we couldn’t exceed 3,000 customers despite considerable marketing efforts. Acquisition didn’t compensate for the churn.

How did we decide to pivot?

In spring 2021, leadership gathered, and we decided to test a new model. We didn’t know exactly what at the moment, but we knew we had to redistribute responsibilities to the city managers (along with a bigger commission share).

We had two main options:

1. Let go of every task and give all responsibilities to city managers—only keeping acquisition.

2. Let go of everything except acquisition, listing creation, and pricing management.

We decided to test the first option.

Virtually, this meant we would become a marketplace of independent property managers since we weren’t offering anything to hosts except selecting people for them. To city managers, we’d become lead providers. This was a natural choice: running a marketplace is in Airbnb’s DNA.

Decisions:

• No commission rate taken by the platform, based on the fact that the business model would generate incremental listings and therefore incremental bookings and revenue for Airbnb, which should compensate for our operations (basically, engineers and account managers to maintain the program and tools). Since Airbnb already takes a 15% fee upfront, the operation was supposed to be profitable. This means that the partners went from taking 50% to 100% of the revenue.

• City managers could decide their own commission rate, and not keep 20% like we used to. Upon the model transition, they could also decide to continue working with a host or not. In short, they had full control over their business.

• New Terms of Conditions applied between the three parties (Airbnb, the host, and the partner). The partner and the host had an external contract between them, meaning the partner could value their business.

• New quality conditions:

  • To enter the program, a candidate partner had to have an average rating above 4.8 stars.

  • To continue being listed on the platform, partners had to maintain a rating above 4.6.

  • Under 4.4, partners were removed from the program entirely, and their hosts were offered the possibility of matching with another partner (if they desired).

• Apart from being featured on the platform, partners also benefited from:

  • Dedicated support

  • Training programs

  • A community of co-hosts

How did we transition?

Test phase

We tested the new model with four partners during summer 2021. They operated in very different markets (a big city, a coastal city, and a rural area) with very different portfolios (3 properties, 10 properties, 30 properties, 80 properties). The only thing they had in common was their high motivation and quality. 

This was an opportunity for us to determine:

1. If the model was applicable everywhere in every condition.

2. If the transition was possible in all cases, and within a two-month timeframe.

The test was a success. We learned many things:

• The additional time investment from partners was quite low, much lower than envisioned. It represented an additional hour per day on average, which wasn’t much considering the extra 50% revenue they earned.

• They even gained time in the evenings and weekends because they could more easily anticipate issues. Since they were in direct discussions with guests online, they could adapt check-ins and operations accordingly, without third-party intervention.

• Their quality improved from 4.7 to 4.8, for the same reasons.

• They were happy to gain control over their portfolio. Some decided to let go of customers with whom earnings were insufficient or relationships complicated.

• The training they received ahead of the transition was helpful. They highlighted some missing information, allowing me to add it for future city managers.

• Most importantly, they were very satisfied with the model change. I received only one call throughout the entire summer, even though it was the peak season in most places!

Transition phase

In September 2021, we were ready to migrate all partners to the new model. This needed to be done smoothly. Here’s how we decided to proceed:

• We held a first meeting to present the new model, announcing the test results, with the four testing partners present to testify. We answered questions about why we decided to change the model and more practical questions about what it meant. This was well-received. Of course, it was mixed: some partners were very motivated, some more doubtful, but we knew we had a good buy-in.

• We decided to do three batches of migration (so-called “waves of migration”), one every 1.5 months. This had several benefits:

  • First, it was easier to manage. We had 60 partners to transition, and doing all of them at once would have been both painful and risky. Painful because we could hardly support all of them during the process, and risky because if something went wrong—despite the test allowing us to fine-tune operations—all partners would be impacted, risking significant churn.

  • Some partners needed more time to prepare, either by hiring someone or reviewing the training documents.

  • Most importantly, it was their choice to make. Of course, they had to transition eventually, but it felt less imposed. It was a mark of engagement on their part.

• We held a second meeting to present the waves of migration, let them choose which they wanted to join, and answer new questions. We also reviewed the training, and all Luckey employees were present to reassure them about the support we’d provide.

• I personally called all partners to directly address their questions, ask which batch they decided to join, and ensure their engagement. This was very useful for establishing a personal relationship and reassuring them.

The partners who participated in the test were also available by phone to provide feedback and advice to other city managers who called them. We didn’t give any guidelines; we didn’t want other partners to think they were under our control. They were satisfied, so we knew their advice would be genuine. This helped significantly, and I’m very thankful for it.

The waves of migration were well-structured, and we didn’t relinquish all responsibilities at once. These were the steps:

• Transfer Guest Support (2 weeks): This was the hardest part. As mentioned earlier, most city managers were partly handling guest support and had access to the same tool as our team, so they knew how to do it, but it was a significant increase in time investment. They had to prepare in advance and decide on time schedules to handle them better. They also needed guidelines on sending automatic or templated messages to avoid spending too much time on each interaction.

• Transfer Listing Creation, Pricing, and Host Support (2 weeks): This was sensitive. Some city managers didn’t have any connection with customers, so we needed to send precise communication to avoid alarming customers about the change and make it as smooth and seamless as possible. We sent two communications to hosts, including their partner’s contact details (with them in copy). We recommended partners call them directly when they had the opportunity to establish a relationship or reassure them.
  Pricing and listing creation were technical parts: Listing creation wasn’t so hard but time-consuming, and we needed to give city managers best practices to ensure listings remain attractive and get bookings. Pricing is the topic we received the most inquiries on from partners during migration; this was crucial to their revenue.

• Transfer Sales (2 weeks): This was more seamless. Partners were already handling on-the-ground meetings with customers, so this was just about managing the “inside sales part.” The most important thing was reactivity: bringing an answer to leads as soon as possible. To be honest, partners weren’t super effective on this part because they were handling daily operations simultaneously. But at least, this was their responsibility.

By February 2022, the migration was complete.

What were the results? 

In a few words:

• We went from a 4.45 to a 4.7 average rating in a matter of months. There are several reasons for this:

  • Operations were indeed more effective (and qualitative) when individuals were in charge of the entire process. They were more “human.”

  • The conditions to remain in the program (above 4.6 to stay featured on the marketplace, above 4.4 to remain in the program) ensured that leads were attributed to qualitative partners and acted as a good incentive to do a good job.

• Only 2 (small) partners out of 60 quit the program. The rest were satisfied with the change, even those who heavily relied on us.

• Sales, nevertheless, declined slightly but were compensated by higher retention. As mentioned earlier, we knew partners would be less reactive than us and more selective. But the good news was that we could hire more partners in the same location. So hosts had the opportunity to contact several of them, increasing the chance of conversion. Partners were okay with having competitors since they were clearly independent.

What happened after that?

• We were relying on third-party service providers for guest support and inside sales, which we cut off and reduce costs.

• Some employees changed roles within Luckey. Those who supervised guest support, host support, and sales became account managers, now supporting partners.

• The tech roadmap changed significantly after the pivot. Since we became a marketplace, all efforts targeted that direction.

• After a few months, the name ‘Luckey’ disappeared completely, and we started referring to the ‘Co-Hosting Program’. Partners were now called professional “Co-Hosts”.

• We grew significantly: after two years, we had more than 600 co-hosts and 10,000 properties in the program. The quality remained stable at 4.75.

Closing thoughts

It’s interesting how this was presented more of an operational change than a full-blown vision pivot- which it was in reality. As mentioned in “​​How Did We Decide to Pivot?”, we could have chosen to keep some responsibilities internally at that time. Even though top management didn’t really have a vision for the future, we made the right choice.

I’m very proud of the work we did and proud to have led it. I’m thankful to all the partners and employees who accompanied us. The Co-Host Network wouldn’t be what it is today without them. ■