From Intuition to MVP: our not-so-common approach to evolving users’ experience

At Casavo, we strive to innovate the status quo and provide a better experience for everyone that is ready for their next chapter in life.

In this article, I want to share our approach to product development, which helps us to drive innovation and solve complex problems in the real estate industry. Rather than focusing on the details of a specific feature, I will highlight the process we follow and the reasoning that supports our approach.

Intuition to drive innovation

At Casavo, we are passionate about simplifying the home-selling and home-buying experience. We believe that real estate is an extremely complex business, characterized by a lot of uncertainties for sellers and an exhausting experience for buyers. Our vision has always been to leverage technology to simplify this journey and create a delightful experience for our customers.

To achieve this, we started with the idea of simplifying the selling process by being the perfect buyer for every seller. We purchased homes directly, renovated them, and resold them on the market (Sell to Casavo). Our disruptive value proposition allowed us to aggregate supply over time. Later on, to serve more customers, we evolved our value proposition and enabled sellers to choose between Casavo as a buyer or to entrust us to find the best buyer for them on the market (Sell with Casavo). However, this expansion posed a question to our product team: How can we ensure that our experience truly solves sellers’ needs and is differentiating in the market?

To answer this question, we relied on intuition, which is the ability to foresee how a future product will likely work, without already framing a solution in your head. We believe intuition is a key part of the product development journey and can be used to align a team towards a common objective, with the awareness that things may not look like how you have imagined them, at the very end.

This idea of “intuition to drive innovation” was also mentioned by Yuhky Yamashita, CPO @ Figma, at a recent talk — he phrased it as “starting from solution to identify problems”.

In our case, intuition came directly from our CEO, Giorgio. We always had the vision for Casavo to be a platform to connect sellers and buyers, reducing the friction to execute transactions. During a Leadership Team meeting, Giorgio shared his view of how the selling and buying experience would evolve into a match: an instant experience in which sellers can have certainty about finding the right buyer on Casavo, and buyers are guided to discover their next home, accessing properties that are not on sale yet.

Back-it up with discovery

We back-up intuition with strong discovery, as it helps to clearly identify how intuition can really solve customers' problems. An alternative path to starting from discovery could be to validate intuition “in production”, however:

• Intuition is normally about a long-term vision (and sometimes too far in the future with respect to the users’ needs)
• Discovery provides the context needed to move from a vision (intuition) to a solution (product increments to reach a business outcome)

To validate this intuition, we conducted a discovery process to clearly identify how this idea can solve customer problems. We combined quantitative surveys with customer interviews on both sides of the transaction and came up with a clear view of why customers were choosing traditional brokers or selling on their own. One insight stood out, “When I need to sell, I choose a local agency in the area because they probably have the perfect buyer for me in their database. As a buyer, I would reach out to a local agency to find a house in a specific area.”

This discovery process validated our intuition about how crucial it is to reduce uncertainty for sellers by providing them with certainty that someone in our buyer base is interested in buying their property. Would it not be irresponsible for a seller to choose someone else to sell their home if we can already provide them with the perfect buyer, even before they start the process?

Discovery, in this case, helped us to frame our intuition with the customers' eyes — for them, the long-term vision is not relevant unless it adds value to their journey and it solves the uncertainty previously mentioned. Discovery is the process that enables us to reduce the execution risk and gives us a more precise framework to solve the problem.

Validate solutions with the smallest iteration

And here comes the fun part (and what I love doing as a product person), finally. With the clear framework provided by discovery, it’s time to validate our assumptions with users. At Casavo, we believe that the most efficient way to test our assumptions is to ship the smallest iteration that can prove or disprove our hypothesis. User testing is key but, in our opinion, getting feedback from real users, in production, is the best way to learn.

In this case, we sat together with the team and defined the hypothesis to test — specifically, we wanted to validate the idea that providing visibility on buyers' dynamics would push sellers to choose Casavo to sell their homes.

We had different options to test this assumption and we identified a touchpoint that was the right one for this new concept: it was the page where we present to sellers the value of their home and our selling solutions. This touchpoint was not only aligned in terms of storytelling to what we wanted to add to the experience, but it also has a self-conversion CTA. This enables us to quickly measure the impact of the test, using this conversion rate as the key indicator of success.

A snapshot of the board we used to refine the problem with the team. We had different options and we tried to understand what was the best solution to maximize impact and learnings.

To prove to sellers that we had the right buyer for them in our database, we took the “show me the money” approach. We decided to immediately display the number of buyers that matched the home characteristics in our buyer base. To do so, we built a Machine Learning Model to infer which buyers in our database would match with a target property. We decided to build a proper model and not to display “random” data because building such a model was not extremely complex, it had other internal use cases and we wanted to be fully transparent with sellers. With respect to the Machine Learning Model, we adopted an iterative approach as well: we started with something more similar to a simple query rather than a proper ML model and we evolved it over time (but this is another story).

Thanks to this new model, we were able to instantly match a new property being valuated on the website with our buyers and show this info to sellers. We decided to display not only the number of matched buyers but also some details for each of them, to make it look more real.

The first version of the instant match that we would show to sellers at the end of our initial valuation flow. We later evolved the experience based on the insights we received after the release.

We released this new feature across our markets and monitored the impact in two ways: analyzing the self-conversion trend (click on the CTA) and organizing interviews with users that have gone through the experience. Combining quantitative analysis with qualitative insights enables you to measure impact and understand its drivers. In our case, users' interviews were key to understanding how to evolve the rough experience we initially had, prioritizing the different feedback we received.

The quantitative analysis of the impact, on the other hand, gave us answers to our assumptions — showing buyers to sellers drove higher conversion. It was the validation we needed to move from a single feature to an innovative selling experience itself.

Test the entire experience with an MVP

Our initial objective was not just about introducing a new feature to increase conversion but to innovate the selling process itself. We experimented with the “instant match” displayed at the end of our valuation flow, which proved to be successful with our customers. We then aimed to simplify the selling process by leveraging our ability to match sellers and buyers and built an MVP to test this idea. At Casavo, we believe that an MVP must cover an end-to-end process, and the most complete an MVP is, even with some “manual” parts, the more we can learn. Former Opendoor CTO Ian Wong deep dives into this concept in his interview with InDepth.

In our case, we scaled the concept of “show me the buyer” by going one step forward: not only displaying an initial instant match made by our model but giving sellers feedback from the market before they would commit to selling. In particular, we adapted the process we use to assess the value of a home and prepare an offer, integrating the buyers’ feedback.

Users that start their selling journey with Casavo, after an online valuation and the initial match with buyers, can share with us some pics of their home and the floorplan, to receive a refined valuation and our selling solutions. They can manage this entire process on their own from our App. Within this experience, we thought about one more step: after sharing the information with Casavo, the seller could now also share them with the matched buyers to receive initial feedback from the market.

The assumptions we aimed at testing with this MVP were two:

• would sellers be willing to share their home details with buyers?
• would having positive feedback from the market (buyers) convince sellers in choosing Casavo to sell?

To design the MVP, we adopted the User Story Mapping format. This framework is particularly useful when you have a clear idea of the main steps of the journey and you wanna deep-dive into the different elements behind it. It also helps with visualizing the “complete” flow and defining what is core and where you need to invest to deliver your MVP.

We quickly realized that the experience had some key steps:

• asking sellers' consent to share the info: we could have bypassed this step by sharing the property by default but we could have not validated the traction of the concept toward sellers
• sharing the property with buyers: we realized that there could have been some issues with the content shared (how could we make sure about the content of the pictures? How could we make sure we never share the property address but only the area in which the property is located?)
• collecting the buyers’ feedback and sharing them with the sellers: we had a lot of uncertainty at this stage, also because we had no idea if there would have been any positive feedback

To build a complete but “quick” MVP we decided to:

• asking sellers' consent to share the info was key: we needed to test our assumptions so we wanted to make it “as perfect as we could”. We decided to start asking for consent from a subset of users who would upload all the info we needed, directly on the App
• sharing the property with buyers would have been very complex to fully automatize (there was technical complexity, a lot of corner cases, and unknowns): we decided to do it manually. Every morning I would manually send out the email to the matched buyers with the properties details
• collecting buyers’ feedback was also done manually: we would check the number of buyers that clicked on the email CTA “I’m interested in visiting the property” and add this info manually in the internal system we use to manage the transaction. The sales team would then pitch this info to sellers when presenting our offer.

After sharing the information with Casavo, sellers have the opportunity to share them with the matched buyers, directly from our App

As described above, we tested the entire process, but invested in automating only the first step — we managed the rest of the process manually.

The first part of our MVP was fully automatized — sellers on the App could give consent to share their home details with buyers. The second part was handled manually — I manually created the email and shared it with the matched buyers. I would also manually report in our internal system the number of buyers that clicked on the CTA.

However, we decided to invest in another key part of the experience: image anonymization. We knew that to fully automatize the process, at some point, we would have introduced some sort of “photo check” to avoid NSFW pics or sharing of personal details (including the address, to avoid buyers reaching out directly to sellers). We decided to invest in it since the beginning, as we could use the initial manual phase as a way to test our models “in production”. Instead of releasing the model once you have a large volume of properties being shared and a fully automatized process, introducing the pics anonymization at this stage meant that I could give some feedback to the team every day, based on the emails I was sending out.

When building an MVP, often data and ML elements need to be taken into account before other parts of the experience. This is because ML often requires a feedback loop that you may wanna anticipate or test on a smaller scale. More details on how we anonymize pictures can be found in the article by our colleague Alberto.

Our MVP not only validated our assumptions but also provided us with new insights to scale the experience. We reviewed data regarding the full-funnel experience every week to spot any optimization opportunities and learnings.

Scale the experience and evolve your value proposition

As we validated the assumption and the overall model, we decided to continue investing in the experience with less uncertainty. The power of MVP lies in providing a clear overview of the potential of a feature or experience with limited upfront investment.

When scaling the experience, we leveraged the learnings we had in the previous phases — we knew that the first version of the scaled and automatized experience could be a little worse than the manual one and we tried step by step to adjust it. We automated the email dispatching and enable buyers to access all the property details in an optimized way. We also collected the buyers’ feedback in an automatic way and informed our commercial team of their interest in real-time within our transactional platform.

This new experience was then reflected in our overall value proposition toward sellers and buyers. It was not necessary to wait for the scaling of the full experience to start communicating it more and more across our touchpoints: as the MVP was already a complete flow, once we saw traction, we started to work alongside our marketing team to adapt our communication and messaging. In this case, product and customer experience were fully aligned with the long-term vision of our company and so to our brand: we made a step toward it.

Final disclaimer

As mentioned at the beginning, the scope of this article was not sharing a success story (as we are still trying to make this experience a success, we are in the middle of scaling it) but more to share some lights on the way we tackled a problem and came up with a solution. There are probably other ways to reach the same results or even better ones, but we thought it made sense to share ours. If you wanna share your approach to building disruptive products, we are curious to hear it!

(A mega thank-you to the amazing team that I’m lucky to work with. Shipping this was only possible because of our PMs, engineers, designers, product marketers, and product ops people! ❤ )