Ved Sinha, Former VP of Product at Upwork, on gig marketplaces

Another, which is more emerging is AI, or even like the no-code kind of tools. This is where the work to be done, instead of just hiring someone to do it, the service provider is injecting a lot of AI, and/or automation to do much of the work. Then there's just a thin layer of human labor that's ultimately delivering the work product to the end customer. You might put something like a Pilot.com, which is an accounting service or Smith.ai for customer service. Again, they're using AI, they're using automation, but then they put a thin layer of human service on top. You can see how, along these dimensions, more and more categories of labor  can inject AI to change the way the work gets done and delivered and the new kind of work-marketplaces you would want to create.

Permalink

Upwork was one of the first marketplaces to unbundle Craigslist. Now we have companies like Preply, Turing and Dribble, which are very specialized, and focus on one vertical. How do you see horizontal or vertical marketplaces competing with each other? What advantages do they have against each other?

For horizontal versus vertical labor marketplaces, it comes down to the size of the transaction - the frequency multiplied by the price. If you’re doing smaller size transactions -  that is, low frequency and low price – you're going to need a horizontal platform.  When the size of the transaction is not very big, unless I have different categories, I'm not going to be able to make up the cost of acquisition of the demand and/or supply-side. So for the frequency low and price low services, you need to be a horizontal like Thumbtack for home-services. Here you want to offer adjacent services to the demand-side so it’s not a question of having to re-acquire the demand every-time.

On the other hand, if you need a specialized or regulated supply with a specialized selection and buying process - then a vertical-only makes sense - like Vivian Health for healthcare clinicians or Vangst for regulated cannabis industry workers. The vertical platform has to manage the frictions of creating the supply or the nuances of the hiring process materially better than any horizontal. 

As for the famous Craigslist unbundling chart—those were completely different services that were being delivered on that platform - you had discussion boards to personals, to jobs, to someone selling an old bike. The point is there were very, very different classes of service that were getting basic liquidity on the Craigslist platform that got unbundled. 

When you look at a labor marketplace like Upwork, it's much less horizontal than you think, if you compare it to the whole scope of services that I talked about earlier, all the different types of work. Upwork, for example, is really just focused on digital work that's remote. By definition, that's going to be information work. But there are many categories within information work. The question is, will what happened to Craigslist happen to Upwork? 

If you look at something like one of the IT-focused labor marketplaces, like Turing, TopTal, etc. there's a big overlap with Upwork in the skills offered on them - obviously IT, software-development etc. 

In terms of the duration or the size of the contract – Upwork— right now, at least—is more in the mid-range, slightly more commoditized work that’s serving more of an SMB client. Whereas something like a Turing operates higher up in the sense of higher quality, more specialized work. There is a slight difference on that dimension.

Then you have the buying process, and that's really where the biggest differences are. With something like a Turing, these vertical-marketplaces are much more curated. Basically the value prop is around vetting of the candidate and the curation of the ultimate match. On the Upwork side, it's very scalable, it's all automated, it's all quality signals within the platform. On the Turing etc. side, there’s more vetting going on, more human interaction, and it's also highly curated, you don't get multiple proposals . That’s the big difference here. 

I see Upwork and Turing/TopTal as competitive at the higher end but not in the mid-range SMB space. So the middle + bottom of the pyramid is an Upwork, and the top is something like a Turing. Basically, there's a big, huge commoditized or mid-range bulk where, because of the size of the transaction, the frequency of the transaction, you have to be horizontal. But then, there’s space for a higher end, higher cost layer. They would co-exist - it's not like an unbundling, it's more like a high touch, high price layer on top of a bulk marketplace. 

We’ll have to see how Upwork, whether it can move up, how well it can move up, or not. Upwork is introducing features to go up there - the Talent Scout is essentially a higher-touch recruiting service that's layered on. The Enterprise capabilities, again, are a new thing that they've launched that allows them to move upscale. 

This kind of dynamic plays in the other categories too, like in Creative Services, where you'll have a more commoditized horizontal marketplace , because the transaction sizes are smaller, the frequencies are less. Once you acquire an SMB customer, you want the client to buy multiple things, otherwise it's not cost-effective, generally speaking. And yes, there’s competition at the higher end, with higher-touch and higher prices. 

Now you mentioned Preply, and that’s more of an education or tutoring market with a consumer focus. Upwork hasn’t been going after the consumer, per se - I don't see them as competitive  because it's a different service. How you deliver the tutoring, it’s completely different compared with B2B information work.

Permalink

Upwork has two sides—the customers and the contractors. Which is the more difficult side to build - demand side or supply side?

Going back a while now,  in the early days, it's a marketplace and a standard chicken and the egg problem. For Upwork, getting the supply was typically not as challenging as getting the demand side. Getting the demand required more of a behavior change of the buyer - especially in the early days. Even now, the buyer has to be comfortable with remote work and distributed work. Today, of course, we have video-conferencing and collaboration tools making it many times easier for remote work than it was 10 years ago. 

For the supply, or the people who are already consultants or freelancers or have those skills, it's a no-brainer. For them, this becomes an easy marketing channel, acquisition channel. And they get the money, and the flexibility, so the value prop is very clear. 

It's a little harder to find the segment of demand that’s open to hiring online, and understands that they can do this. That they can hire someone remote in, say, Nebraska, and that person can get the job done, and do it safely. The level of trust that you need to build up, and the level of confidence needed was always higher on the demand side.

What was the GTM for the supply side? How did that evolve and inform the product strategy as the product evolved?

In the early days, on the supply side, what we were looking for—given the type of service we are providing—was as much unpaid organic acquisition as we could handle. That meant a tremendous amount of investment in SEO, and making the unpaid organic acquisition work for us. There’s a virtuous loop here, because the more jobs we got, the more content there was for SEO. The more profiles we got, the more content there was for SEO. The more data that we could generate about the market, like hourly rates and salaries, and trends and skills, the more content there was for SEO. 

We followed it very, very consistently, like how do we generate data that is interesting for our marketplace participants, but also interesting, therefore, for the search engine? For example, making the profile and the job get as much new content as possible, whether it's samples or updates or response times, and basically keeping it alive. There's a really nice virtuous cycle here, where you increase the number of jobs and you get more people showing up, increase the number of profiles, you get more people showing up. More people showing up in turn increases the number of jobs and profiles. 

We saw the share of organic really expand a lot, I'm talking a few years back, when we were starting out. That was a huge advantage in terms of cost of acquisition, especially when you compare with the cost of a traditional agency. 

Viral acquisition was hard to make work here with the users being competitive with each other. At the same time, of course, you want to invest in paid, and take advantage of new online channels, and the traditional performance-marketing stuff.  And then there's earned content, earned media. A marketplace obviously has a lot of insight on the market, the salaries, the trends and the skills, what's up and what's down, and that generates a lot of interest. 

All the growth that happened and all the acquisition was without any sales people, like zero, so it was 100 percent digital. I'm talking about the early days. Even now, the supply-side acquisition is predominantly digital, and so in that sense, product-led.

Permalink

What was the GTM for the demand side? Did you start with SMB and move up the stack to enterprises, or start with enterprises and then expand into SMB?

The enterprise motion is more recent for Upwork, in the last few years. It's really been more of an SMB and startup play. Even now, I would say, if you look at the latest earnings reports, 80 to 90 percent of revenues is still SMB, with enterprise growing fast. The enterprise is more recent, and you have to build new processes and new features to support the compliance, the onboarding, the virtual talent bench, to support that. And, of course, the sales layer and the account management layer. 

But that’s really more recent, it's the SMB that was a bulk of the demand-side we went after. And it was, again, using a lot of organic and paid, meaning digital. Now we’re seeing an expansion from SMB to enterprise.

Permalink

What are the differences in the business model of a traditional staffing agency vs. Upwork?

Ultimately the service they're both providing is very similar or almost the same – it's helping an employer get the job done using talent, using a worker. 

A company like Upwork is leading with technology, and reimagining it with technology. The traditional companies are doing it well, but using traditional operational processes. The biggest thing that I would say is just the level of productivity per employee is so dramatically different. Upwork has 10 to 15 times more revenue or GSV per Upwork-employee, compared to a more traditional agency. Upwork would be something like a $4-$6 million GSV per employee. A traditional staffing agency—and this is all public info - like a Robert Half, for example—would be around $300,000-$400,000 per employee. It's a huge difference, and that's because of the technology that’s doing a lot of the work. 

It shows up at every stage of the process. Whether it's a lower cost of acquisition, a lower cost of qualification and conversion, a lower cost of matching, a lower cost of contracting, I mean, through the whole chain. That's why the productivity is so different between the Upwork and the traditional. 

The other big difference is the take-rates are so much higher in traditional agencies. Take-rates in traditional agencies can be 60 to 75 percent, whereas an Upwork take-rate is about 15 percent. So there's a huge margin gap there and I have to believe that Upwork having a lower cost profile is going to be better in the long term. 

Another key difference is the usage of W-2 employees versus 1099 workers - where Upwork is more 1099 versus the traditional agencies using W-2. But now it's becoming more and more hybrid. The best traditional agencies are investing in technology, and Upwork is investing in a human layer, a recruiting layer, like the Talent Scout as well as increasing their W-2 payrolling capability.

Permalink

With Upwork having a strong enterprise sales motion does it mean more bespoke, customized service with less tech? Do you see Upwork actually becoming more like a staffing agency and less of a tech company at some point of time?

The starting points are so different. Upwork is completely reimagining the value delivery using technology, versus a traditional agency starting from a recruiter doing all the work as a way of operating. I don't see it as converging, like meeting each other, I just see a movement towards more hand-holding on the Upwork side, using their technology foundation. There's a long way to go before these two sides are going to have a similar cost structure, if ever.

Permalink

We're seeing marketplaces built on top of software. For instance, Workrise has a full-fledged SaaS powering the workforce management contractor and payroll. What are your thoughts on the importance of embedded SaaS within these marketplaces? What role would they play, and who is deriving value from that? Is it the contractor or is it the companies?

That's obviously a big trend that we’re seeing, where you're getting workforce management platforms targeted at specific industries. Workrise, for the oil and gas industry, Snagajob and Instawork, targeting hospitality and retail. In the regulated cannabis industry there’s Vangst. In healthcare and nursing, there’s Vivian Health, for example. So, absolutely, because of the differentiation and the skill that's required, and the way the work gets done, there’s a huge opportunity to create these verticals.

It’s simply a question of what is the customer experience that is needed to get the job done? What are the dynamics of the supply and how hard is it to get the supply? What are the specific frictions in the supply network creation? What is the buying process that has to happen? And post-match, what’s the workforce management on the backend, for example, shift scheduling, that needs to happen? Or what level of training? For contractors, what on-boarding, what time-tracking and co-ordination is needed? 

The benefit is, there's efficiency gains for everyone. There's efficiency gains for the supplier in terms of the training, in terms of having job after job after job, in terms of the automatic invoicing, the fast payments. There are efficiency gains for the buyer, getting trained and qualified talent, faster time-to-hire, and seamless payment approvals. So, in the best cases, there's efficiency gains for both sides, it's not one side or the other.

Permalink

What’s the advantage a company gets when paying contractors through a platform? If they ever decide to take those contractors outside the platform, what do they stand to lose? What stops them from doing that?

On the company side, there’s the convenience of setting up the payments, where the invoicing and the payment processing is all automated. You can do it not just for one particular contract or one particular employee, but across all of your flexible workforce, along with analytics and reporting on it. So there is  a convenience factor. 

The other thing that companies look at is the trust and dispute management. If something goes wrong, they have a third party that they can go to, that third party being the platform. That is a good amount of leverage over the talent, because it can affect the ongoing reputation of the talent. If you go off the platform, you lose that dispute management and that talent-reputation leverage. If you stay on the platform, you retain that leverage. 

Now, in addition to that, the SaaS tools in many labor marketplaces provide value beyond payments after the match. On Upwork, for example, you get what they call a work diary, which keeps track of the hourly work in an automated way that builds trust in the payment amounts that need to be made. 

For some companies, particularly in SMB, to the extent that they build a reputation on the platform of their own responsiveness, payment records, disputes etc., that makes it easier for them to hire in the future. It's not as big an issue compared to the contractor side, but it’s an issue; the more people they hire, the more valuable they'll be seen, so they can hire more people more easily. 

Now, if the platform is performing the role of employer of record and serving as the W-2 employer, that's 100% a reason companies want to stay on - where the company gets compliance with employment regulations. That's what staffing agencies do, and for general labor marketplaces like Upwork do to a much lesser extent. Labor marketplaces more typically work with 1099 employees, and help with filing taxes for that in an automated way. 

I would add that on these labor platforms, they extract value in a tiered way – meaning the larger the amount of money you're spending, the lower the cost to you over time - to help reduce disintermediation.

Permalink

Contractors multi-tenant across different platforms. What does Upwork offer to keep them on the platform?

On the supplier side, it really depends on the labor market when it comes to multi-homing. We saw that we were able to retain our best performing supplier contractors. One big driver is the reputation they're building on the platform. The more hours they work on the platform, the more ratings they get, the more reviews they get, the more they invest in their profile by adding samples. The more we track—because we can, these platforms can track the work that's getting done—the more we can see verifiable work that's been completed and other quality signals, and all these things add up to a reputation and a brand for the supplier. That brand, for the best people on the platform, not necessarily for a new entrant, is a big reason to stay on, because it makes it so much easier to get your next job. 

In fact, the people with the best brands will get work coming to them just automatically. Because they would show up higher in search results, they have different badges that show that they are best performing. Platforms, like an Upwork, can give them marketplace tokens that allow them to apply to more jobs. Basically, the ability to get new work increases a lot when your reputation goes up. That's not very portable, that reputation is not portable. If you went to a new platform, you would have to build all that up again and go through that.

In the case of services marketplaces, there are services that are differentiated, services where there's a relationship between the buyer and the provider, and with those services, reputation matters. In the case of a fungible service, a completely transferable service, like a taxi service, really it doesn't matter who the supplier is, as long as they meet a certain threshold and you can get from A to B. But if you're hiring a programmer, or you're hiring a product designer or a marketer, then yes, their reputation matters, and how you differentiate compared to other professionals matters. That's where the reputation comes in, and that's what makes it less amenable to multi-homing, once you get established. 

Then beyond that, of course, you have productivity benefits, like the automation of the workflow post-match, the automated invoicing, the payment protection, the speed with which you get paid. All those benefits, post-match, are a secondary value proposition that provide differentiation to platforms. 

I would say the biggest is the brand - it's just like in the real world, once you build up a brand in a certain area, in a certain city, if you go to a different place, you have to build up your brand again. For differentiated services, that's pretty sticky.

Permalink

Is there also some component of a lower take rate? If I do more jobs on a platform, and spend more hours, I get better ratings? Does the company’s take rate go down, does it stay constant, or go up, how does that work?

Absolutely. As I mentioned, there’s a step function decline in the take-rates to help reduce dis-intermediation. The take-rate model is aligned to the delivery of value . It's highest when you have the initial match, because the supplier just found work, there's huge value being delivered. Then, as you continue working together, because the supplier and the provider are building trust in each other, the platform take rate goes down. As the work-volume increases, you keep decreasing the take rate. 

At the same time, the platforms put in these other tools, like the automated invoicing, automated payments, the 1099 filing, a single platform for all your tax work etc. that kicks in. But bottom line, yes, as the amount of payments for ongoing work increases the take rates would typically go down.

Permalink

On the competition side, a company like Deel came from nowhere and captured the contractor payment/EOR market. Upwork, Fiverr and Freelancer already had payments and contracting. Do you think this is something that was consciously not taken up by the marketplaces or is it a missed opportunity for these marketplaces?

Traditionally, staffing companies provide employer-of-record or EOR services with W-2 employees. The new work platforms typically wanted to be 1099, or non-employee, in order to be very asset-light. 

Now, what's happened is that because of COVID, it's created a space for hiring remote workers - permanent and contractor - in different countries, not just in the US . There were companies doing international employment compliance before, but it wasn't as big an opportunity compared to after COVID. 

So yes, because of their historical focus on 1099 contractors versus EOR/W-2, platforms like Upwork are missing a big opportunity. They can today handle 1099-class workers anywhere in the world. But for W-2 payroll, Upwork has a product that's only on the US side. It seems to me that there's a big opportunity to offer that same W-2 type payrolling service, not just in the US, but overseas as well. 

Because a company like a Deel would be developing a global supply base, supply network, and it's totally logical that they would eventually build a marketplace layer on top of the payroll.

Permalink

What can marketplaces like Upwork do to improve the margin profiles? Is it about improving take rate, moving to enterprise, or more services, like layering in financial products?

The question around expanding margins and achieving profitability in labor marketplaces can be a long haul. You’ve got to grind your way to getting to the scale. For labor marketplaces, like an Upwork, the opportunity of margin expansion simply by expanding take rates is absolutely there. Because the status quo competition is a staffing agency. The staffing agency take-rates are, as I said, around 60 to 75 percent compared to a take rate of 15 percent on the labor-platform sites.So there’s absolutely an opportunity to simply increase the take rate, whether it's adding onto the base rate or adding different fees to it etc. 

My sense is that all of the platforms want to build that scale first. The penetration is so low right now. If you look at staffing alone, that could be over $500 billion in spend in the US alone. And the penetration of these platforms is so low, sub one percent, and that doesn't even account for the expansion of the TAM, because of COVID surge in remote work. So they're focused on just expanding their market share before extracting more of the take-rate. Number one is simply increasing the take rate, and a few percentage points itself is massive, because the competition ultimately is charging a huge amount more. 

The second thing is payment processing costs. When you're taking only 15 percent of payments and the payment processing costs are two to three percent, that's a huge chunk, again, that can be reduced. Just making that direct deposit or other ways of getting that money, that's a substantial margin expansion, using payment processing. 

Then, you have the traditional benefits of scale, where the platforms can scale and expand using the same technology. The cost structure is ultimately going to help over time. 

You can also add more upmarket services, higher margin services on top. But even these first three opportunities are so huge. The payment-related products are certainly an opportunity, because it’s billions of dollars that are going through in payments to contractors, freelancers, and providing a financial services layer totally makes sense. My sense is that the companies will want to focus and invest first in the matching and hiring value proposition, and just getting that at a sufficient penetration before extracting value from the payment processing. But the option is there.

Permalink

We’re seeing the evolution of gig workers to solopreneurs, people who run their life and work as a business. Companies like Patreon, Memberful, and Substack,help creators monetize their talent. In some ways, that’s exactly what Upwork does, helping workers monetize their talent. Do you see these companies acting as competition? Or do you see them as catering to a completely different class of gig workers?

For companies that you mentioned, like Patreon, we use terms like “the passion economy” or “the creator economy.” They are very different compared to the traditional labor marketplaces. The job to be done for the end consumer is very different. On the gig-marketplace side, the employer is saying, ‘I need this work to get done,’ and the contractor is delivering that to the employer. On the creator economy side, it's a lot more of the contractor acting on their passion, and creating content, and deciding what to create in order to be consumed by an audience. 

In one case, the employer is directing it, and in the other case, the talent says ‘hey, I want to create this content, and I want an easy way to monetize it and find my audience and find patrons.’ They're not competitive.

Permalink

As the marketplaces evolve, what do you see as a persistent economic moat? Is it the scale of talent, is it SaaS, or is something different?

First, of course, is the marketplace liquidity and selection: the fact that a marketplace can get transactions done, because they are attracting supply, that attracts demand, and that attracts more supply and so on. There are network effects here, because my selection choices as a business go up as more workers come onboard. On the worker side, my opportunity to find new work goes up with every new business that comes onboard. That’s hard to start, and once you reach a certain scale, hard to dislodge, witness Craigslist traffic still today. That's a big moat.

Another huge moat is the proprietary data that all these platforms are collecting about these different classes of work, whether it's about the performance of the client, or the contractor, how successful they've been, how they perform, there are all kinds of quality and differentiation signals. User behaviors allow platforms to create a quality score, and there's an immense amount of data that's being collected about the contractors. Similarly, to a lesser extent, you're collecting data on the client side about how responsive they are, how fast they do things, how many times they have disputes. The data that's being collected ultimately helps, it gets fed back into the system, it makes it easier to do the matching and hiring, and it increases the chances of job success - ultimately it helps with the outcome the client is seeking. 

Related is that proprietary data also enables AI/ML across every stage of the hiring process. When you're starting out, you have no data and so you have no AI, and you have no optimized matching, whereas at scale, all these things kick in. These two things are the way I look at it.

Permalink

This transcript is for information purposes only and does not constitute advice of any type or trade recommendation and should not form the basis of any investment decision. Sacra accepts no liability for the transcript or for any errors, omissions or inaccuracies in respect of it. The views of the experts expressed in the transcript are those of the experts and they are not endorsed by, nor do they represent the opinion of Sacra. Sacra reserves all copyright, intellectual property rights in the transcript. Any modification, copying, displaying, distributing, transmitting, publishing, licensing, creating derivative works from, or selling any transcript is strictly prohibited.