All posts in “Startups”

Elton John invests in Qloo, a startup that analyzes your taste

Qloo, which bills itself as “the cultural artificial intelligence data platform,” has raised $6.5 million in additional funding from AXA Strategic Ventures and Elton John.

Yes, that’s the Elton John who co-wrote and performed “Tiny Dancer” and “Candle in the Wind.” He’s actually not the first big name to back the New York startup — actor Leonardo DiCaprio and Starwood Hotels founder Barry Sternlicht were both investors in the $4.5 million round that Qloo raised last year.

Qloo examines publicly available and open source data to try to understand patterns in consumer taste in categories like music, film, TV, books, podcasts, consumer products, fashion, dining and travel.

While the company has built a cultural recommendation app that anyone can download, it’s currently focused on selling this data to other businesses. For example, it says can help marketers understand which interests are shared by the fans of two competing sports teams, so that they can tailor their advertising accordingly during a big game. (The company says it has already mapped 750 million of these types of correlations.)

“We’re excited about the potential of Qloo’s cultural AI,” said Sir Elton (it feels wrong to just call him “John”) in the funding release. “It’s been a powerful tool for [his music company] Rocket Entertainment for brand partnerships across our entertainment company. We feel Qloo could be a driving force in the entertainment industry in years to come.”

Featured Image: Qloo Inc.

Marketplace liquidity

Liquidity is a crucial metric for all marketplaces. But how can we truly evaluate this liquidity? The three keys to answering this question are density, appropriately balanced demand and supply and category concentration.

Density, geographic reach and distance between counterparts

The first step to analyzing the potential liquidity of a marketplace is understanding its geographic extent. For example, Upwork will see transactions filled between freelancers and employers all around the globe, while Tinder will (typically) only see two people dating within the same city. Tinder would therefore require a city per city launch and a city per city liquidity analysis.

In local marketplaces, there is a strong correlation between density and liquidity. The higher the number of participants within a relevant radius, the higher the liquidity of the marketplace. We define liquidity as the number of transactions filled out of the total potential transactions in a marketplace. We refer to density as the number of participants within a certain geographic area.

Let’s use Tinder as an example. If in a given week Tinder acquires 1,000 female participants in Chicago and 1,000 male participants in New York, Tinder’s number of total users increases, but its liquidity remains the same. Now, if Tinder acquires 1,000 female participants and 1,000 male participants all located in the borough of Brooklyn, this higher density of users (from both sides of the marketplace) will lead to higher liquidity.

To determine the density in a marketplace, you need to first define the limits of the geographic area within which its transactions can be filled. This distance threshold (“r” in the visual below) is different for every marketplace and you will identify it by understanding how far your customers are willing to travel to complete their transaction. For example, a seller at Letgo might not be willing to travel 30 minutes to sell a used skateboard for $30, but a babysitter at will likely travel 30 minutes to make $150 for a day of babysitting.

Once you define this distance threshold “r” you need to maximize density. But how?

You start by optimizing your marketing strategy to achieve the highest customer acquisition within the established distance threshold. As a result, you will maximize the active participants within that distance of each other, and therefore maximize density, moving from the low-density to the high-density marketplace diagram above.

The evolution of a balanced marketplace

Maintaining the optimal balance of demand and supply in a marketplace is crucial to achieve liquidity. If a marketplace doesn’t have the right number of buyers per listing, or the right number of items listed per buyer, transactions will go unfilled.

How can we analyze what is the right balance of demand and supply? By looking at several ratios:

  • The average number of buyers relative to the number of sellers needed to fill a transaction
  • The number of bids per buyer required to fill a transaction
  • The average of items listed per seller required to fill a transaction

Let’s base our example on a used furniture marketplace in which the average ratio of bids per item listed required to fill a transaction is 100 to 1. So, every item listed needs 100 bids for it to sell. Now the conversion from this ratio to the ratio of buyers to sellers is determined by the average number of bids per buyer and the average number of items each seller is selling. This is the key to the analysis.

For simplicity, let’s assume the average items posted per seller is 1 and the average bids per buyer is 10. This means that the ratio of buyers to sellers is 10:1; it requires 10 buyers for every seller, to fill a transaction. However, it usually takes time for the average number of bids per buyer to be this high, and it will only happen once the company has reached a stable stage of liquidity.

At an early stage, the likely scenario is that the average number of bids per buyer is 1 or less, which means the ratio of buyers to sellers initially is more like 100:1. So, it requires 100 buyers for every seller to fill a transaction. The ratio of buyers to sell would evolve like this:

The first reason why average bids per buyer increase over time is that when a new buyer experiences a marketplace for the first time, they are hesitant to bid. The buyer needs to trust the marketplace before being ready to make a committing bid. Trust takes time. Furthermore, at a stable stage the diversity of inventory in the marketplace will increase the likelihood that a buyer will find something that interests them.

However, the conclusion here is not just that the bids per buyer increase over time and that this determines the evolution of a balanced marketplace. It is that you need to closely monitor the second-degree KPIs (like bids per buyer and listings per seller) to understand how to adjust your strategies to maintain a balanced marketplace over time. In the same way that it is common for average bids per buyer to increase over time, it also is usual that the average items listed per seller change over time.

Category concentration and diminishing marginal returns

Once you have defined the distance threshold under which your marketplace will fill transactions, you need to breakdown the liquidity per category within the marketplace. For example, if our used furniture marketplace has 1 million active buyers at a certain point in time, how many of them are in the market for a table? How many are looking for chairs? How many want a couch? Only a small segment of the 1 million active buyers will be interested in each category. This means you cannot just evaluate the overall liquidity of the marketplace, you need to do it on a per-category basis.

It is often the case that most participants are interested in just a few of the categories served by the marketplace. This means that as you spend marketing dollars to acquire users, the higher-concentration categories will yield liquidity faster than the rest. And in most cases, the low-concentration categories will require a much larger marketing investment to achieve liquidity.

Let’s go back to our furniture marketplace example to illustrate this point with the following assumptions:

  • 50 percent of the buyers are interested in buying a couch while the other 50 percent are equally scattered across 50 other furniture categories
  • We are spending $100 on marketing
  • Our CAC is $1
  • The liquidity threshold per category is 50 buyers (the liquidity threshold is the number of buyers needed for a transaction to fill)

We would have 50 new buyers in the couch category, and one new buyer in each of the other 50 categories. The 50 new buyers in the couch category will substantially increase the likelihood of a filled transaction within its category, while the additional one participant in any of the remaining categories barely increases its chances to fill a transaction in their respective areas.

If we take this example to scale and we assume:

  • Marketing budget of $500,000
  • A liquidity threshold per category of 250,000 participants

Only the couch category would reach liquidity. And even if we were to invest $1 million in marketing, only the couch category would reach liquidity because all the other areas would only reach 10,000 participants each. And the same would happen until the invested amount goes above $25 million. This means that such a marketing strategy would lead to diminishing marginal returns from a liquidity standpoint:

Again, the best way to solve this issue is to align the marketing strategy to the category concentration. In other words, allocate the majority of the marketing investment to the most predominant category (which in our example would be the couch category). The liquidity chart in that case would look like this:

This is, of course, an extreme example.

A more realistic case would have several concentrated categories and a long tail of dozens of others. To illustrate this, let’s assume 30 percent of the buyers want to buy a couch, 20 percent a chair, 10 percent a table and the remaining 40 percent are scattered across the 40 other outstanding categories. This time, the liquidity chart would look like this (which is more realistic):

Density, a balanced demand and supply and category concentration are crucial drivers of marketplace liquidity. To maximize your return on marketing investment, you need to align your marketing and liquidity strategies. To monitor the evolution of your liquidity, you need to track your second-degree KPIs.

VCs should also dig into these three principles when evaluating a marketplace investment opportunity.

Featured Image: LEONELLO CALVETTI/Getty Images

The Lemonade insurance social experiment results in $53K donation in year one

Lemonade has raised $60 million to date with the thesis that insurance companies can cut costs and automate more of their processes if they can somehow incentivize users to act less selfishly and commit less fraud. Traditionally insurance companies take the spread between premiums and claims as profit, but Lemonade aims to donate unclaimed premiums to charity once a year in a “Giveback” to discourage fraud. Today Lemonade made its first annual donation of $53,174 or 10.2 percent of first year revenues.

Lemonade founder Daniel Schreiber has a term for the evils many of us succumb to when filing claims with our home insurers — letting the devil loose. Its common practice for homeowners and renters to file claims at inflated values and to keep granted claims even when stolen or missing items turn up. All of this bad behavior accounts for billions in lost revenue across the industry and creates an adversarial relationship between people and their insurers.

Schreiber is the first to say that the idea for a Giveback wasn’t about being generous — it’s business. During an interview, he recalled two Lemonade customers who have called to voluntarily return money given to them for stolen goods that were ultimately recovered. This is a heartwarming story for sure, but it’s still too early to saw whether the model and the behavior scales.

Daniel Schreiber (Lemonade) at TechCrunch Disrupt NY 2017

There are really two ways to evaluate Lemonade. One is to accept that people are heavily influenced by their environment and that the startup can actually incentivize people to act in better faith. The alternative, considerably more cynical, reality is that people are people and Lemonade’s early progress is merely a product of early adopters embracing a like-minded company — not everyone will go out of their way to buy Toms so a shoe can be donated, but the positive corporate ideology attracts its own cohort of loyal customers.

Lemonade and its investors deeply hope that they have it right and the first assumption rings true. Schreiber turns to the other benefits of Lemonade (beyond charity) and his incredibly low customer acquisition cost as additional support to back up his thesis. Compared to other home insurance offerings, Lemonade is both inexpensive and simple to use. This has led to a 10X reduction in CAC.

The behavioral economics that Lemonade is rooted in tells us that there is a science to criminal behavior — that the right circumstances can lead people to do almost anything (see Stanford prison experiment and all things Milgram).

[embedded content]

“Insurance ticks every box of what not to do,” Schreiber tells me when discussing the nuts and bolts of designing a system to incentivize good behavior.

The good news for Lemonade is that the company doesn’t need every single user to be altruistic — a few bad actors doesn’t ruin the experience for everyone. The Lemonade team is closely monitoring claims and keeping its fingers crossed that future Givebacks remain proportional to about 10 percent of revenue. It’s clearly still very early, but Lemonade can count year one as a success by those metrics.

You can buy Tovala’s $399 smart oven starting today

Tovala, a $399 smart oven out of Y Combinator promising to perfectly cook ready-made meals in under 30 minutes is finally available for purchase.

It can bake, boil and steam meals at up to 550 degrees Fahrenheit in 10 to 30 minutes and works with either prepackaged Tovala meals or your own food. However, the real money maker for the startup will be in the meal delivery service.

It’s a crowded space with Postmates and other restaurant delivery as well as meal kits more similar to Sun Basket and Blue Apron. However, Tovala offers what it believes is a quicker way to make meals and with less clean-up.

“This is one of the biggest complaints with the meal kit companies – after all the cooking, you’re left with a ton of pots, pans and plates to clean. There’s literally no clean up required with our meals.” co-founder David Rabie said.

Right now, Tovala has two subscription plans: 3 single-serving meals for $36 or 3 double-serving meals for $72. Compare that to Blue Apron’s offering at 3 double-serving meals for $60 a week or the family plan serving two recipes per week at four servings each for $72.

Tovala tells TechCrunch it will eventually add a family plan as well as breakfast and dessert meals. For now you can choose from entrees like miso-glazed salmon or Moroccan chicken with chickpeas, goat cheese & corn pasta, for example.

Each Tovala packaged meal can feed up to four people, are made from scratch in the Tovala kitchen, and consist of 400 to 800 calories per serving. The startup also has a recipe library for creating meals with special diet restrictions such as paleo, vegan and gluten-free, if those are your preferred options.

We interviewed Tovala’s co-founders Rabie and Bryan Wilcox back in March when the company first launched. Check out the video below to see our original review a few months ago and get an idea of how this smart oven works.

Those interested in purchasing one for themselves or finding out more can go to

Sendence closes $1.5m seed round to simplify deployment of real-time applications

Sendence, a new startup based in New York City, announced a $1.5 million seed round today. The company is working on a platform for simplifying deployment of real-time applications.

Investors include Boldstart Ventures, Greycroft, Contour Venture Partners, Notation Capital and Resolute Ventures.

As company CEO and co-founder Vid Jain sees it, more companies are hiring data scientists to develop applications to make decision based on large amounts of data in real time, but the problem becomes dealing with the complexity of running these applications, whether in the cloud or a data center.

What Sendence purports to do is remove all of that complexity by handling the underlying infrastructure plumbing and making it scale automatically whenever new resources are required.

Developers write their applications on the Sendence Wallaroo platform in their language of choice, which Jain says covers most popular choices such as C++ or Python — most anything except Java. The idea is to simply create your application as you normally would, let the system know where your infrastructure resides, and it should deal with launching and managing the application with the correct level of resources — and modifying that over time as requirements change.

These types of applications tend to run across dozens or even hundreds of servers, and employees end up spending a lot of time on the infrastructure side making sure the data isn’t lost, load balancing, all of the infrastructure maintenance and management tasks. He maintains that this a huge waste of time and resources.

Once the application has been launched, you can monitor the performance in an analytics dashboard.

Photo: Sendence

For smaller companies where valuable data scientist time is spent configuring the infrastructure side of things, this could be a boon, but even for larger companies with data engineering and infrastructure teams, it could help automate a lot of activities that require human management today.

“Even if there is a group geared toward data engineering or infrastructure, when [the data scientists] are developing the application, they still need to think about the underlying infrastructure and how to scale it. When you write an application as data scientist on our platform, you don’t have to think about any of that,” Jain said.

The company plans to open source the core technology shortly and are hoping that a community builds around the open source project to help give the product some traction as they head out to market this year.

Sendence currently has 10 employees and several customers in Beta. The company’s short-term goal is to get a couple of paying customers by the end of the year, then begin to grow the company from there.

Featured Image: gilaxia/Getty Images