Airbnb’s Algorithmic Odyssey: Balancing Automation, Pricing, and User Trust

Over the past few weeks, numerous articles featuring interviews with Brian Chesky, the CEO of Airbnb, have emerged. In these articles, Chesky has made several intriguing statements. One particularly noteworthy comment pertained to the concept of “pillars” within the company.

“‘To use a precise metaphor, it’s kind of like we never fully built the foundation. Like, we had a house and it had four pillars when we needed to have 10’... Math aside, there are three core pillars Chesky says would add up to ‘a really great service’: affordable prices, reliability and proper customer support when things go wrong. But retrofitting a large house isn’t easy. ‘The bigger you are, the more effort it takes to increase quality,’ Chesky says. ‘And that’s what we’ve been really focused on.’”

I find the concept of “pillars” within Airbnb intriguing and very much aligned with a concept that I’ve been “promoting” over the last few years: the idea of Product and Process Market Fit (as opposed to just Product Market Fit). The insights gleaned from Chesky’s comments underline a striking reality: Airbnb appears to have a significant gap in its processes. 

The Issues 

One stark insufficiency lies in their customer support. Many who attempted to cancel or modify a booking during COVID will likely concur that their experience was far from satisfactory. No phone number, just email support, slow responses, and an overwhelming absence of transparency and clarity. It often felt like navigating a maze with no end in sight, which underscored the urgency for Airbnb to address this process inadequacy.

The article reveals another glaring issue: the platform’s inadequate support for guests when hosts cancel. Many guests have lamented the sparse recourse available when faced with last-minute cancellations. The situation grows more perplexing when considering Airbnb’s cancellation policies —the fact that a host can rescind an offer at the last moment and leave guests stranded without adequate compensation strains the platform’s trustworthiness severely. To make matters worse, there are occasions when some apartments don’t even exist! Imagine the disorientation and helplessness of arriving at a destination only to discover that the promised accommodation doesn’t exist or isn’t available. Some may claim that this holds for any platform and that Airbnb doesn’t have control over each and every accommodation (like Marriot does, for example). However, they have many more options for different locations and prices, which a hotel chain cannot match. The idea behind a platform is to “let the market regulate.” 

But markets usually don’t do a good job when it comes to “long tail risk,” which in the context of platform governance refers to the infrequent yet potentially high-impact events or anomalies that can severely impact a platform’s reputation and user trust. While these events may be outliers, occurring far less frequently than regular, smooth transactions, their negative implications can be disproportionate. For platforms like Airbnb, this could translate to rare instances where hosts abruptly cancel on guests or where listings are misleading or non-existent. These atypical occurrences can overshadow hundreds of positive interactions. So while most transactions on a platform may be positive, the few negative ones often get amplified in users’ perceptions and discussions. For platform-based businesses, it’s crucial to understand and manage these long-tail risks effectively, as they can be pivotal in shaping the overall perception of a platform’s reliability and trustworthiness. Therefore, proper platform governance isn’t just about ensuring day-to-day operations run smoothly; it’s also about anticipating, preparing for, and addressing these rare but impactful risks.

Airbnb’s operational framework underscores an essential reality of platform-based businesses: the critical need for clear and effective processes, given their lack of resources (in terms of ownership/control over the listed properties). Consider the intricacies of their pricing structure. Guests frequently encounter additional charges such as service and cleaning fees. Yet, paradoxically, despite being levied a cleaning fee, they might still be tasked with chores like doing the laundry or tidying up before departure. This dichotomy isn’t merely a hiccup in communication; it reveals a broader ambiguity in the platform’s processes.

In the absence of clear guidelines, hosts and guests alike resort to their own strategies or “hacks” to navigate the platform. For instance, a host might capitalize on the system’s flexibility by setting a low initial price, only to add fees later. Conversely, a guest might develop strategies to circumvent or challenge these fees based on past experiences with other hosts. Such self-devised strategies fill the vacuum left by a lack of comprehensive processes, creating a landscape where inconsistencies abound. While these individual adaptations might seem like innovative solutions, unfortunately, they lead to a more fragmented user experience. 

Over time, these inconsistencies can erode trust, emphasizing the necessity for platforms like Airbnb to continually refine and clarify their operational processes. By doing so, they can ensure a more standardized, transparent, and reliable experience for all parties involved.

Product Market Fit

To truly grasp the concept of “product and process market fit,” it is essential to first comprehend the foundational idea of “Product Market Fit.” This term is a central theme in many entrepreneurial texts. For instance, in the book Blitzscaling, Reid Hoffman posits that an absence of Product Market Fit is a primary bottleneck hindering a company’s growth potential. I agree. 

Andy Rachleff, who is credited with coining the term, eloquently defines it as follows:

“To me, product-market fit is when you have proven the value hypothesis. So the value hypothesis is the what, the who and the how. What are you going to build? For whom is it relevant? How’s the business model?... first you have to prove a value hypothesis and only once you’ve proven the value hypothesis should you test a growth hypothesis.”

In essence, Rachleff emphasizes the initial need to validate a product’s inherent value in the market. Only once this foundational value is ascertained should one move on to scaling or growth strategies. Building on this definition, when we introduce “process” into the equation, we’re emphasizing the synergy between the product’s inherent value and the operational systems in place to deliver, support, and scale it effectively in the market.

Another, slightly more controversial, perspective comes from Marc Andreessen, who offers a more pragmatic viewpoint:

“The product doesn’t need to be great; it just has to basically work. And, the market doesn’t care how good the team is, as long as the team can produce that viable product. In short, customers are knocking down your door to get the product; the main goal is to actually answer the phone and respond to all the emails from people who want to buy.”

Andreessen’s insight underscores a vital aspect: at its core, Product Market Fit is about meeting market demand with a functional product. It’s less about perfection and more about practical viability. This perspective implies that for many firms, the foremost challenge is not just crafting a stellar product, but ensuring it aligns with tangible market needs. Once that alignment is achieved, one can argue that the most significant risk has been mitigated.

Product and Process Market Fit

So, when speaking of Product Market Fit, we’re addressing a firm’s ability to effectively meet the market’s demand with its product offering. However, adding the dimension of “process” to this fit emphasizes the importance of having a valuable product and the necessary mechanisms and operations to deliver and support it at scale.

As a company evolves, its growth invariably means interacting with a broader, more diverse customer base, each with their own set of expectations and needs. Thus, while achieving Product Market Fit is paramount, maintaining and adapting it during scaling becomes equally crucial.

As companies mature and expand, they encounter an increasingly intricate web of challenges. They must navigate the waters of burgeoning competition, adapt to shifting regulations, and remain agile amidst fluctuating economic conditions. Moreover, as their customer base grows and diversifies, maintaining their value proposition clearly becomes critical.

Andreessen’s observation regarding a viable product meeting market demand is only the starting point. As companies like Airbnb scale, the mere functionality of a product or service isn’t enough. The value proposition must be fine-tuned, requiring the integration of more robust processes. I’m not talking about standard operating procedures or delineating exact work methodologies. I’m highlighting the fundamental need for tasks and responsibilities to be executed efficiently and effectively. 

For example, a common oversight, as evidenced by many firms, including giants like Airbnb and Uber, is relegating customer service to the back burner. Instead of viewing it as a core component of their value proposition, it’s often treated as an ancillary function, a cost center. The misconception that customer service can be an afterthought in the growth journey undermines the very essence of building and maintaining trust with a growing user base. In many cases, it’s not about the absence of a customer service function but the palpable void of genuine, responsive, and effective customer engagement. 

So, a potential definition of Product and Process Market Fit would be:

The alignment and synergy between a product’s value proposition (Product Fit) and the operational, production, and delivery mechanisms (Process Fit) that support and scale the product, such that both resonate strongly with market demands and expectations.

Product Fit: This is the traditional Product-Market Fit. The product itself — its features, benefits, and design — addresses a pressing need or desire in the market.

Process Fit: Even if a product resonates with the market, the company’s ability to produce, deliver, and support that product at scale is equally important. This includes manufacturing, distribution, customer support, and any other process that might be crucial to satisfying and retaining customers. 

For startups and businesses, achieving both aspects of this fit would mean having a great product, and efficient and scalable systems in place to support its growth in the market.

Achieving and maintaining “product and process market fit” requires understanding a company’s evolution, its current standing, and the value proposition it brings to the table. It delves into ensuring that every supporting component, from the pricing structure to the resources at hand, aligns seamlessly with the firm’s stage of growth and its overarching mission.

The key questions I ask firms, given the evolution of their scale, customer base, technology, brand, and competition, are: what should they start doing, what should they keep doing, and what should they stop doing? Certain practices, like intensive customer onboarding, may be necessary in the early stages. However, as the firm matures, such practices might become untenable or even counterproductive. Conversely, new challenges may necessitate fresh strategies that weren’t previously in the playbook. And, of course, there are core practices that remain vital throughout a company’s journey. 

In Airbnb’s case, listening to user feedback suggests that reliability and affordability stand out as primary concerns. This sheds light on the inherent tension between hosts wanting to maximize revenue and guests seeking value. Customers understandably prioritize transparency in pricing, affordability, and reliability. They want the assurance of knowing what’s available and when. On the flip side, hosts or property owners are driven by their own set of priorities. They aim to maximize their returns, maintain flexibility in their offerings, and minimize the resources they commit to prepping their spaces for guests. 

At first glance, it might seem that these sets of desires are at odds. However, there’s a sweet spot where both parties’ needs converge. Demand surges when there’s a robust alignment, especially at the right price point. In this zone of mutual benefit, customers and hosts find their needs met and value maximized. In essence, when the equilibrium is achieved, everyone stands to gain. However, achieving this equilibrium can’t be done just by allowing the market to “self-regulate.” Additional processes are needed. 

Algorithms and Processes

For Airbnb, many of these processes rely heavily on technology. This is true when it comes to pricing as well as verification. 

As technology evolves, questions regarding the role of algorithms in determining pricing and other aspects of the service naturally arise. Airbnb has long implemented dynamic pricing strategies, assisting hosts in setting competitive rates. However, this approach has stirred debates with some critics wondering if such automated pricing mechanisms could inadvertently lead to price collusion or stifle the organic interplay of supply and demand. It’s a delicate balance, blending human intuition with algorithmic precision and ensuring that fairness will remain intact.

Dynamic pricing algorithms, powered by modern machine learning and AI tools, have become prevalent among sellers on online platforms and have replaced manual price settings. While platforms like airlines and hotels often employ dynamic pricing, concerns arise about its impact on consumer welfare and the overall structure of marketplaces. Notably, the founder of Poster Revolution faced prosecution under antitrust laws for using algorithms to gather competitors’ pricing data with the intent to coordinate pricing on Amazon Marketplace. The court’s decision deemed price-fixing cartels illegal, regardless of their execution means. However, challenges emerge when there’s no clear evidence of collusion, especially since some algorithms can unintentionally coordinate to set higher prices.

Together with Ken Moon and Amandeep Singh, we analyze data from Airbnb to understand the effects of hosts using algorithms on the marketplace’s outcomes. Airbnb introduced the “Smart Pricing” feature in November 2015. This feature automatically adjusts listing prices in response to demand fluctuations. Many hosts have since adopted this automated pricing tool. Additionally, several third-party firms, like Beyond Pricing, have emerged, offering dynamic price management services for hosts. Consequently, hosts can choose between managing their prices manually, using Airbnb’s Smart Pricing, or employing a third-party service. In our research, we construct and evaluate a model that considers both customer preferences and the cost structures of hosts.

In our preliminary analysis (the paper is not yet available), we demonstrate that for a significant number of markets, the adoption of smart pricing algorithms by some hosts leads to a significant increase in the price of the overall market, as demonstrated by the graph below:

In this case, a good process (helping hosts price) is tilting the market into an equilibrium that is much less favorable to consumers. 

Airbnb is also diving deeper into technology by introducing a new verification system to ascertain the existence of listed apartments. While this seems like a step forward, there’s concern that the platform may encounter unforeseen challenges. At scale, companies often grapple with issues related to algorithmic biases or what we term algorithmic anxieties. For Airbnb, the algorithms’ primary function isn’t the listing’s creation or its “consumption” (you really must stay there yourself as AI is, unfortunately, unable to vacation for you at this point). The main function of Airbnb is matching, which is done via curation.

The research paper Algorithmic Anxiety and Coping Strategies of Airbnb Hosts, delves into the experiences and perceptions of Airbnb hosts concerning the platform’s algorithmic evaluation and reveals several interesting findings:

Algorithmic Anxiety: This anxiety stems from the lack of clarity regarding how the hosts’ actions impact their profile representation on the website. The authors emphasize the need for designers to create systems that not only provide seamless experiences but also address the anxiety the hosts feel. They suggest that exposing more about algorithmic evaluation systems can potentially decrease user anxiety.

Alerting Workers on Damaging Actions: The paper suggests that platforms should alert users about the function of algorithms during their routine use rather than relegating this information to help pages. Automated mechanisms can be used to detect when a worker’s actions significantly affect their evaluation, either positively or negatively. This can offer users more agency and incentivize them to continue providing good service.

It’s evident that Airbnb has faced challenges with curation. Their latest move toward generative AI could inadvertently heighten these algorithmic anxieties. A solution may lie in incorporating more human intervention. While automated processes excel at scale, a deeper dive into competitive advantage might require more human-centric support. This void in Airbnb’s service delivery is why companies like Vacasa have emerged. Recognizing the processes Airbnb lacks, these companies offer value-added services that could potentially harm Airbnb. These firms enable hosts to list on other platforms like VRBO and Booking.com, potentially causing a “market leakage” - a situation where value flows away from Airbnb. This scenario benefits hosts and customers but might not be in Airbnb’s best interest. As we move forward, it’s crucial to evaluate if some of these challenges are more structural in nature.

Personally, I’m a fan of Airbnb and I hope they continue to evolve and resolve their issues. Some of my best vacations were in Airbnb accommodations, particularly in Europe. However, over the past few years, the pricing, reliability, and additional inconveniences (the fact that you can’t leave your luggage after you check out, for example) have nudged me back toward hotels (or more localized websites). Airbnb may have built a global village, but ultimately, it’s the strength of its foundational pillars that will determine whether guests feel at home.