Again lots of acquisitions but also fundraising news in December especially in the voice analytics space. As always, I will share my point of view on those and what this could mean for the Voice industry as well as dig deeper into Alexa’s issues around discovery and monetization. Here is your monthly digest.

No slowing down for acquisitions and investments in December

  • CallMiner raises $75m from Goldman Sachs

Callminer is a US-based company that analyzes interactions over voice or written messages between customers and businesses and then surfaces insights that businesses can act on. They have been around for some time (founded in 2012), employ about 250 people and just announced a $75m financing round, bringing the total amount raised to $140m.

The first thing that comes to mind is that this is a big amount, what the industry calls late stage. This also implies a high valuation which I would assume (no information has been leaked) to be around $300m. This makes sense for a company that hasn’t crossed the $100m ARR (Annual Recurring Revenue) yet. The second thing that comes to mind is that that big amount has been raised from only ONE investor. True, Goldman Sachs is an investor with very deep pockets, but usually, those types of deals are done by several investors, which let me wonder about the relationship between the two companies and the strategic interest for the bank.

  • Observe.ai raises $26m

Another voice analytics company also raised money this month: Observe.ai. The company founded in 2017 and that offers a Customer Service Voice Analytics platform raised 26m$. The investors were Scale Venture Partners (lead investor), Nexus Venture Partners, Steadview Capital, 01 Advisors, and Emergent Ventures. The voice analytics space is developing quickly as companies realize the value that voice recognition and AI could bring to their operations. There is plenty of competition though with the likes of Cogito, Verint, Nice Technologies as well as I2X here in Europe.

  • Hugging Face pivots and raises 13.5m€

French-American startup Hugging Face announced a pivot as well as having closed a 13m€ round from Lux Capital. Previous investors A.Capital Ventures, Betaworks, and business angels with among them basketball player Kevin Durant also pitched in. Hugging Face was founded in 2016 and was originally planned as an AI-based chatbot for teenagers with the goal of entertaining rather than helping them. This explains the name. Now, the company pivoted and is aiming at creating an open-source library for Natural Language Processing. Let’s see how the project evolves.

  • Majelan raises an extra 6m€

The Paris-based platform for production and distribution of podcasts raised 6m€ from existing investors (Idinvest, Xavier Niel) as well as BPI France. That brings the total amount raised to 10m€, which is impressive for a service that is only available in French so far.  Majelan is a mix of Netflix and Spotify: users can access thousands of podcasts for free (those podcasts also being available on other platforms) or opt for the premium version (4.99€/month) which offers exclusive content.

  • Speechly raises 2m€

Speechly out of Helsinki raised 2m€ from Berlin-based Cherry Ventures and business angels. The funding will be used to further develop its Natural Language Understanding API. This API is targeted to companies who want to offer Voice as a user interface to its clients but who do not want to live in the Alexa, Google or Siri ecosystems and have to share their data. My take on this is that while the offer could resonate with some clients, there is quite some competition in the space and distribution (sales!) will be a challenge.

  • Replica Studios raises $2.5m

Replica Studios, a company offering synthetic voices for artificial intelligence secured $2.5m in a funding round led by the Venture Reality Fund. Synthetic voices allow developers to offer “different” voices for text-to-speech than then usual voices of Alexa or the Google Assistant. There are plenty of similar solutions out there from Amazon’s own Polly to France-based Voxygen (also rumored to be closing a financing round) as the market is expanding. Brands have understood the necessity to be present on voice assistants and the need not to use the classic voice of Alexa for example.

  • Coca Cola invests in a voice startup out of New Zealand

Amatil X, the investment arm of Coca Cola invested in Aider, a company from New Zealand that has developed an assistant that collects and analyzes different KPIs about your business and delivers insights. Voice is a natural interface for questions and answers with the assistant. The company raised a total of NZ$ 750’000 or approx. 450k€. Now why did Coca Cola do such an investment and why does the startup that offers a SaaS solution chose Coca Cola as an investor remains a mystery for me.

  • Sonos closes down Snips

Following its much-debated acquisition last month, Sonos is not wasting any time and is pretty shutting down Snips. The company decided to end the local voice control (ie not connected to the cloud) which was Snips flagship product. As previously highlighted and regardless of the official communication, the deal resembles more and more to technology and acqui-hire. Put in layman’s terms, Sonos bought the machine and the people to run it and won’t sell anything the machine has or will produce.

Monetization of Alexa Skills

While voice assistants are a clear success with users, it seems that when it comes to monetization there is still some way to go. A (paid) article published by the magazine The Information revealed some numbers about Alexa’s revenue targets and results. Remember that just like Apple with the App Store, Amazon makes money when consumers purchase on the store via its In-Skill-Purchase (ISP) by taking a commission of 30%.

Well, Alexa missed its target big-time by reaching only $1.4m in the first ten months of 2019 for a target of $5.5m. That means that less than $5m was generated from ISP over 10 months which is very low. In comparison “small” app developers easily reach tens of millions of revenue from their mobile apps. There could be a few reasons for this such as too few developers offering ISP for premium content or consumers not willing to purchase that content. Personally, see two main issues there: quality and discovery

  • Skills quality is a typical problem of stores in the early days

There are some high-quality skills available in the Skill Store, but there are two main problems: there are too few of them and they are not standing out from low-quality skills. What I mean by too few is that the barrier to entry to create a skill is very low…but the investment needed for a high-quality skill offering premium content is very high. Too few developers have introduced really superb skills that would justify users to pay for their premium offerings. There is a large investment needed on one hand and limited revenue expectations on the other hand, so why should the developers take (most of) the risk?

Also, brands need a safe environment for them to move in: I was talking with a marketing VP at a mid-size consumer brand the other day and she explained to me that one of the several reasons they were still reluctant to build a skill for their brand was the “junk you find on Alexa (ie the Skill Store)” and that she didn’t want “(their) brand and logo appearing next to a farting-noise skill”. She had a point: at the time of writing this section of the article, a quick look at the French Skill Store showed Virgin Radio right next to one of the two farting skills appearing on the “homepage” of the Skill Store. So much for curation and signaling quality…

Now, if you remember the early days of the App Store, it was very much the same with apps like whip or beer drinking noise apps. Things have changed quite a bit since and I’m sure that the smart folks at Amazon will fix this. Hopefully rather sooner than later.

  • Discovery is a real problem

Those having developed a high-quality skill and introduced ISP to monetize it now face a new and massive challenge: discovery. There are only two ways for a user to activate a Skill:

  1. Invoke it by name on the device
  2. Activate it in the Skill Store (web or mobile)

In the first case – which would be the equivalent of a user typing directly your URL into a browser – the challenge is for the user to both remember your name and know you have an Alexa skill available. If you enjoy high brand awareness, then things get easier, but you still need to communicate about your skill and what benefits it offers. If your brand awareness is fairly low, then it’s an uphill battle.

In the second case, the user will be either on the web or on the Alexa app and discover your skill. Here two scenarios. In the first one, your skill got promoted either directly (ie someone in the Amazon team decides to put your skill forward) or indirectly meaning that with your skill usage, retention, reviews….etc. you get surfaced in the ranking (there is a science behind it in understanding the algorithm and pushing the right buttons called Skill Store Optimization, but that should be the topic of a separate post). In the second one, you actually drive traffic via paid or earned marketing campaigns to your skill page within the Skill Store.

The similarity between all those cases? In all of them, Amazon controls the user flow. And the fact that there is no way to track and attribute skill activation is a real problem for skill developers willing to put money behind their skills. Imagine running online marketing campaigns to drive traffic to your website without having Google Analytics.

Now to the good news…we (an awesome team of developers and myself) have been working on this problem for some time and have found a solution that allows developers to track and attribute skill activation. We have been testing it for some months and are now putting the last touch to the user interface. If you read this post until here and are interested to hear more about it or to test it out, feel free to reach out at alexis@voxalyze.com

Alexis Hue

Alexis Hue

Alexis is Voxalyze cofounder. He is passionate about voice as a user interface and loves using data to unlock and enable growth.