Airbnb Hospitality by Voice – Petar Ojdrovic, Home Service – Voice Tech Podcast ep.041

Petar Ojdrovic Home Service

Episode description

Petar Ojdrovic is the founder of Home Service, a voice-based solution for vacation rental management that Improves the guest experience while helping hosts save time and drive 5 star reviews. I’m also joined by my co-host, the inimitable Charlie Cadbury, CEO of Say It Now, who elevates the discussion with his vast experience of building apps for the travel and hospitality industries.

In this episode we learn about the typical problems vacation rental hosts and guests face during their stays in vacation rentals like Airbnb, and how hosts can use voice to tackle these. We cover designing a human into the loop, the ways multimodal should be used, how to improve intent resolution.

Charlie introduces us to the concept of a crap wizard, and how you can avoid having your voice app users feel like one, and Petar shares advice for budding voice developers and startup founders. This in-depth conversation really showcases the potential of voice to gather data and offer value to users.

Highlights from the show

  • What are the typical problems vacation rental hosts face?
    • They have to answer lots of questions, between 7 and 14 separate communications per guest, which takes a lot of time.
  • How you can use voice to answer guests’ common questions for you?
    • The Home Service app comprises a SaaS web interface for host to provide custom information about the property and area, and a voice interface for guests to access that information in the property.
  • Where does the information come from?
    • Most questions are prepopulated, as most guests ask the same questions.
    • The answers to many questions come from the hosts themselves, and pulls information from Google and Tripadvisor.
    • The system also pulls information from rental services like AirBNB to choose which questions to use
  • What’s a crap wizard?
    • A wizard that knows the spells exist but can’t remember the words to invoke them. Voice interfaces are often like this, containing functionality that’s inaccessible due to users being unaware of it.
    • To combat this, simplicity of design combined with visual prompts can be used to onboard users. Multimodal can help too, showing options or a cheatsheet on the screen.
  • Why use SaaS and not voice for the hosts too?
    • It’s a mistake to try and use voice for all use cases – it’s additive, not all powerful. SaaS is better for form filling.
  • How can bringing human into the loop help?
    • You should bring people in when needed, but automate otherwise.
    • You must also make it clear when humans enter the conversation e.g. In Home Service, when moving from voice to chatbot, a notice appears saying that you’re talking to a human.
  • How much does the system cost Airbnb hosts?
    • It’s just $10/mth, and there’s a 2 week free trial.
  • How does Homeservice measure guest satisfaction?
    • It answers guests’ questions immediately, rather than making them wait for the host who is often too busy to reply immediately.
    • This helps increase and improve 5 star ratings on the platforms
    • Home Service also identifies happy customers and continue to engage with them after they check out, in order to obtain more postive reviews for the host.
  • What are he challenges faced by the hospitality industry, and how can voice help?
    • Homeservice provides analytics on guests, which hosts can use to improve their service.
    • Complex business structures and messy integration with multiple legacy property management systems (PMS) prevents a lot of innovation in hotels, and dissuades startups from working with them.
  • Who is your main competition in the space?
    • Volara, who are focused on custom skills for brands, and Virtual Conciege Service.
  • What technical approach do you take for intent resolution?
    • Home Service trains their models with lots of examples for relatively few intents, leading to very accurate intent resolution.
    • They use Amazon intent resolution for STT, and add their own layer for the questions.
    • Each question asked by a guest is assigned a quality score indicating how likely it belongs to an existing intent. The low scoring intents are manually reviewed and assigned, improving the system for future users.
  • What advice do you have for voice developers?
    • Start simple and find a very specific niche – don’t tackle too broad a problem, as intent resolution is too difficult.
    • e.g. it’s much easier to build a voice assistant for hospitality than for sales, as the search space of intents is much smaller.
  • Do you consider AirBNB to be a threat or a potential acquirer?
    • An acquirer, as AirBNB only want to release things they can charge for, and they would likely offer Home Service as a free feature. Therefore they’re unlikely to develop their own feature, and would acquire.
    • In the coming years, AirBNB will need to compete with rising competition from hotels getting into the vacation rental space – they will need to gather more data, and offer more value to guests and hosts. Home Service does both.

Links from the show

Subscribe to get future episodes

Join the discussion

Support the Voice Tech Podcast

Share this article

What do you think?

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Related Posts

Speech Analytics Companies
Petar Ojdrovic Home Service
Top Podcasts

Get notified about new articles