Marketplaces – MTsireud

Local SEO & marketplaces: A product management view

$3,000 per day. This is the day rate for an SEO audit from an expert. In other words, $3,000 to tell how your web app sucks, not even how to suck less. . Why? Because SEO matters. There is good news too. You don’t have to pay a dime as long as you have your eyes on your users. This is how we did it at Timewith, my company, and ended up growing 1,000% within 6 months just from SEO.

Timewith is a therapy marketplace. Initially I was apprehensive about SEO. We were a pure-bred product/engineering team and SEO seemed like a dark art incompatible with our structured approach. How could we trust a process we neither understood nor controlled? But the prize was large so I decided to make a few shy attempts. Our first foray in search engine optimization was enough to teach me that SEO rewards products focused on pleasing the users. And this changed my mindset. Sure, there is depth to cover in the technical details one needs to master and sure, a Google “randomness” factor applies but equally, the biggest wins came from product and UX improvements. We also didn’t have any experts at hand, so perhaps you can do the same without spending $xx,000.

SEO and marketplaces in particular have had a natural love affair that has been going on since the humble beginnings of the internet. Amazon, Expedia, eBay, Airbnb, Craigslist, Thumbtack. The list is endless.

To make it visual and add some other nice examples, here’s a visual with just a few unicorns where SEO really made a difference.

SEO and local marketplaces. A recipe for success — SEO & local Marketplaces. A recipe for success. *Mtsireud.com, 2021.*

There is good reason for that. Marketplaces are not producers. Instead, their function is to organize the corresponding market’s information and make it easy for potential buyers to assess, compare and select the right product or service for them. A curated, thought-through aggregation results in a better web experience and makes customers happy.

And since marketplaces make customers happy, Google (and other search engines) promotes aggregated listings pages that offer true utility to the customers (not to be confused with scrapers/aggregators with no added context which are down ranked by Google).

On the other hand, SEO is critical for the growth stage and viability of a marketplace and there lies the basis of this fragile symbiosis.

SEO strategy in 2021: UX above it all

Due to the intrinsic value of SEO the level of competition steadily grew. Whilst a decade ago, simply aggregating and presenting information would serve a business well, nowadays pages need to abide by the principles of great UX, provide unique insight, and be supported by great engineering. Backlinks are another important factor but since they’re not location-specific, I will leave this out for now. However, there is one case (Thumbtack) worth looking into where they used local gov organizations backlinks to boost local listings. Amazing.

And since SEO is equally art and science it makes a natural fit for a Product Management viewpoint. “What does this mean?” you say. Think about the user, about the business case, think about engineering precision. And think about Google.

The ideal scenario is that is that when a customer visits a result link, they educate themselves on the offering, engage by progressing deeper down the funnel and finally make an action. Of course not all experiences happen in a linear, centralized or or sequential fashion but this is a starting point for the goal. The bad scenario is that the user can’t find what they want or are misled so they bounce. Google notes that as a negative signal.

Marketplace Funnel, a simple, linear perspective

So what is the best SEO strategy? It’s the one pleasing the users and a good way to go about this is by preempting their needs. Ask their questions beforehand.

How many of the customer queries can we answer?
1. Can we make the answers granular and specific?
2. Can we add unique insight to each answer?
Can we get the answers fast enough? Can we internally promote the right pages?
Can we allow users to find the right solutions for them?
1. Can we allow users to dissect, compare and filter the right solutions for them?
2. Can we create trust that we are offering the right solution?
3. Do we have the right products or suppliers to help them?

SEO as a series of UX optimizations. — SEO: A product management approach. *Mtsireud.com, 2021.*

Product functionality

Search & filter

Imagine you are looking for a therapist. What is it that you need to do to find, assess and compare in order to find the right one for you?

You might need help in assessing what you need help with. A short “test” might be available. This helps you create a relationship with the app and also the app benefits usually from capturing some data, prompting the user to come back.

Secondly, once you have a sense of what you’re looking after you might think about what is “the right therapist”. This might include credentials, experience, tone and attitude and therapeutic modality or it might be practical considerations such as location, price and spoken languages.

In general, to keep a customer happy you need to know the question they’ll answer and crete a series of search and filtering experiences that ideally serve the content before it’s asked through personalization and recommendations or at least allow the user to easily discover what they’re looking for.

Supply and quality of data

Supply matters. It matters to business model and it matters to SEO too. There’s a dangerous allure to onboard as many professionals or inventory to please Google. But this does not matter if the content is poor. In other words, a product, a shop or a professional are not a useful addition unless there is proper QA and the information around the products and services provided is detailed at least, unique ideally.

Social Proof and Trust building

Often considered a conversion tactic, Social proof and Trust building can take your SEO a long way. Generous cancellation policies, refunds, 24/7 customer support are great to create trusted relationship. Publications, accolades and reviews for social proof. Besides the obvious benefits, there are a few SEO specific reasons why this is important.

More clients convert if the brand seems legitimate. In fact, nowadays client’s won’t convert unless the brand is legitimate; the bar is higher than ever. From an SEO perspective, this is also great signal.
It can be a strong signal for search engines to use reviews.
It helps with retention and virality which creates a virtuous cycle for SEO too. When users trust a brand they sign up for newsletters, follow pages on social media and in the end come to the site “direct” i.e. typing the name of the business. All that is a strong signal for SEO and in the same time plays a valuable strategic goal of slowly setting up the stage for independence from SEO which is and will be owned by Google.

Information Architecture

Information architecture should be designed by thinking in terms of information priority or in other words, what are the questions users ask and in what order? This is where good keyword research comes handy.

Let’s return to our therapist example. In this case, looking at the keyword structure (UK) we conclude that the top Searches are in this order:

“Therapist in <location>” – between 500,000 to 1,000,000 searches per month. Example: Therapist in Manchester
<Counselling> therapist between 100,000 – 250,000 searches per month. Example: Depression therapist in London
<Approach> therapist – 50,000 – 100,000 searches per month. Example: CBT therapist

What does this tell us? It tells us what is the way people think. Firstly, users might think in terms of distance and locality. However this subset of users may or may not have a lot of knowledge in the sector. For instance, they might have not considered what the right modality for them is and thus, might also be of lower intent.

If you’re starting up and building your domain ranking, it makes sense to go for the niche, specific queries. When it comes to SEO, less search traffic is often correlated with lower competition and higher intent. Then you can build your way up to the fat heads (e.g. therapist in <location>).

However, from an overall architecture, here is what this could look like in terms of your information order and URL structure.

Information architecture could look like this:

Information order
1. Nearby areas
2. Actual results
3. Pagination
4. Related searches with area and approach
5. Stats about therapy inside the search area (custom content, continuously updating)
6. Featured therapist & therapy approaches (custom content, continuously updating)

The more technical view, including URL architecture could look like this:

Url structure – three SEO friendly URL structures
1. Location only:
  - Top location: /topLocation/?page=x
  - Sub-location: /topLocation/subLocation?page=x
2. Location & counselling area: /topLocation/subLocation/counselling/<area>?page=x
3. Location & therapy approach: /topLocation/subLocation/approach/<therapy_approach>?page=x

Rich (user-generated) content

Reviews. Insights. Comments. Recovery data. A dedicated section for “How our first session works”. Anything that is not automated, duplicated and scraped and offers insight in a question a user might have is gold.

Here the winnings are twofold: first, you keep users spending longer time on your pages and secondly they progress down the funnel due to trust and no cognitive barriers – you have answered their questions -.

This is a great example where product management and SEO overlap. Product managers think in those terms already to increase conversion, revenue and decrease CAC. But guess what? This is what Google looks for too.

Finally, deliver content that keeps your users coming back. Emails and Social media prompts are reasons for your users to come back.

Here’s what blending the concepts of tags, content and overall architectural considerations might look like for a results page:

Engineering

Finally, you got latency. When was the last time you waited for more than 3” to load a page? According to Google analytics, after 7 seconds, 32% of users will have bounced from a page. Why do you think Facebook, Twitter, and Instagram do all this work to load in less than 3 seconds?

So here are a few suggestions to get right early on to get any application to load in under 2 seconds.

Consider pre-rendered content. An example architectural solution would be a React application with Server-side rendering.
CDN caching – cache the pre-rendered pages for <x> hours/days.
Only load the information (content) required for Google to index the pages, e.g. don’t pre-load the availability of listings – this is pulled in after the whole page is loaded in the browser.
Use compressed versions of images together with lazy loading them as the user scrolls down the page i.e. don’t load the whole page just the first results like the Facebook feed.
Preload any CSS, JS assets so your page rendering isn’t blocked. There are trade-offs here between performance and UI (e.g. with preloading you may see an unstyled version of the page for a fraction of a sec before seeing the styled one) but hey; SEO benefits, right?
Finally, database design. If you’re using a relational DB, make sure the relationships in your database are such that serve the performance of loading the data.

When it comes to engineering specifically, I see this as free points. Every team control their own performance vs controlling users’ reactions in a new experience. That’s not to say it will be trivial or easy to do the above. It will time but this is also the most straightforward optimization conceptually. In other words, this is up to you.

Link equity and no-follow

Another important concept to consider is that of link equity. Think of your site as having a total amount of credit. Which pages should this credit be distributed to? Are your meta-tags reflecting this hierarchy of page importance?

Meta tags
1. First listing page <meta name=”robots” content=”index, follow”> – these are the pages Google indexes, so only the first pages are showing on google results. These are the pages where your link equity is diverted to.
2. Next pages <meta name=”robots” content=”noindex, follow”> — tells google to not index all other pages in the results but to follow all links on these pages.
3. Also, avoid to index pages with few results that don’t provide a meaningful search experience.
4. Finally, do not follow links in the footer section which you don’t require for SEO reasons (e.g. terms or privacy pages), because why pass link equity yo T&Cs?

And so just like that you got SEO game.

How much should you invest in SEO?

If SEO sounds like the land of promise, it’s not and there are pitfalls to be aware of. As time goes by, Google may change their algorithms or decide to capture a larger part of SEO for themselves by selling promoted listings.

A good example is the travel vertical of search. Type “flights to <location>” and you will see the google flight product, Google’s own flight search product.

Is this an imminent threat in your industry? The answer is it depends how big your market is but it is an eventuality. When you rent a house, the landlord will eventually kick you out.

For the marketplace, the end game is to end up with the majority of the traffic flowing direct, create a destination site. Early on every business to solve for discoverability which can either be paid and expensive or it can be organic and SEO is a known and tested formula to do so fast but it’s dangerous to be the end-to-end play.

Does this mean not investing in it? Certainly not. Take on SEO early on, treat it like a core part of the product and you will have consistent, sustainable strategy that can take you far.

In the end, all networks converge: Facebook marketplace is eating up the markets.

When I think of marketplaces in the future I see Facebook against the world.

Facebook vs eBay vs Amazon vs Shopify vs Craigslist Vs everyone-that-comes-along-in-any-vertical.

A few years ago this comparison would seem nonsensical.

But in the ever-expanding internet universe, these businesses are not commerce, fashion or social. They’re all network businesses.

This makes sense once you look under the hood. Networks don’t derive value from individuals but rather from their interactions. And ultimately, monetize in similar ways:

1. Ads

2. Listing fees

3. Rake.

(4. SaaS – but let’s leave that for another post).

So is it any wonder that marketplaces are targeting ads?

Marketplaces go after ads and services

eBay, in its 2019 annual report, mentions that marketplace net transaction revenue grew to 8.86% up from 8.25% as a result of “promoted listing fees”, a nifty way to promote sellers’ items.

The marketplace recorded $85bn in GMV and of the total revenue 7.5% was attributed to…ads.

Amazon makes $12.5bn in revenue from ads, 23.5% up in 2020, which is not far from the rate at which AWS, the crown jewel of the company, is growing.

But it’s not only the behemoths.

Etsy made a controversial foray in advertising revenue. In contrast to the more commonly accepted model of “you may promote your listing” they forced advertisers to pay an extra % in case their item was sold as part of a cross-promotional effort namely Etsy advertising on their behalf across Pinterest, Facebook and Google. In other words, Etsy is no less than forcing their retailers to pay for the services of an outsourced marketing department which is genius and yet a step too far.

Instacart, Deliveroo, Uber Eats & basically any marketplace where sellers are not commoditized (will) offer a promotional boost and eventually seek the ad revenue as the next frontier.

Is it any surprise? Marketplaces exist to allow users to discover and for sellers to sell.

All marketplaces started by offering the same things to users: Search amongst a plethora of suppliers and better pricing as a result of information symmetry. Today, marketplaces dominate by improving the experience of transacting with others.

eBay was the pioneer of this new wave of marketplaces. First it established trust with its sellers reviews. Later by purchasing PayPal, it added further ease and value to parties. Payments not only increased the profit margins but allowed the marketplaces to expand in value-adding services. eBay, using PayPal’s capabilities, added the “money back guarantee” and in doing so, created a new playbook for marketplaces.

One where marketplaces would use their low cost structures and high profit margins to reinvest in streamlining the experience. In turn, users would be happier, buy more and return more frequently. Whilst eBay might be the pioneer it is far from an isolated case.

Amazon revolutionized logistics with Prime and changed eCommerce forever. AirBnb offers insurance to hosts. ThredUp offers to sift through the seller’s clothes, price them and sell them saving them the hassle. OpenDoor is willing to buy house’s upfront.

Meanwhile, internet penetration and digital marketing maturity are increasing driving paid advertising costs up. In turn, capturing a bigger part of the customer lifetime value turned from opportunity into imperative. Capturing a higher chunk of revenue is necessary and a mix of advertising revenue and services offered is the way.

Yet there is one outlier in the game. A network first, Facebook didn’t need to worry about customer acquisition. All of us are there.

From ads to transactional revenue: the Facebook way

Engagement, growth and more engagement. The Facebook playbook in a nutshell. If anything was obvious early on it was that user engagement needs to be protected at all costs.

Facebook holds its users’ real identity and and the responsibility to protect it is a business imperative. Over the years, Facebook has invested significantly in what is internally called “community integrity”. The aim was clear: remove all posts that are net detractors of engagement and stymie growth. This was a key move their predecessors had missed and a precocious one at that.

At the same time functionality built around identity propelled their targeting capabilities to levels previously unmatched – even by Google -.

Sign in with Facebook was a key move that subtly, yet critically, allowed Facebook to break out of its own limitations. Businesses all over the web would allow Facebook to track their users outside of the network. This was a fantastic move for two reasons: better targeting directly improves targeting and thus revenue but also allows Facebook to gain critical data it didn’t have before whilst rooting across the whole worldwide web.

Facebook had secured trust amongst its users. Now it also had intent. And with its advanced machine learning it could mine and classify all sorts of interactions. Recommendations requests. Preferences. And of course, commerce.

The social network had all the information it needed. The best part? Unlike any marketplace, users join and broadcast information for free allowing the business to continually reinvest in engineering tools and targeting capabilities.

What remained now, was to come up with an efficient way to expose that information.

Facebook launched its marketplace in 2016 reaching 800m users in 2018; its foray on search, however, had started much earlier. In 2011, the entity graph was introduced which, given its scale and time, was an engineering marvel.

The original focus was on people’s relationships. This quickly extended to map entities around each user. Facebook was already building an advanced ranking mechanism for listings and items.

(https://thenextweb.com/facebook/2013/06/06/the-evolution-of-facebooks-entity-graph-the-structured-connections-behind-graph-search/)

And somewhat like that Facebook had the foundations for a strong marketplace of its own with built-in acquisition, trusted connections, never-ending inventory and incredible targeting.

Facebook marketplace model, Mark Tsirekas

The first steps were ultimately successful as the marketplace reinforced the existing flywheel.

It also provided yet another reason to be on Facebook which meant increase on the time users spent, data the network aggregated and ads served. So far so good.

But to compete in the future of marketplaces, Facebook will need more; it will need to create vertical experiences for buyers and sellers alike.

So the question becomes: Is there anything stopping Facebook from doing so?

The main thing in Facebook’s way is itself

Facebook’s ever-compounding flywheel is predicated on two things: zero cost of customer acquisition and targeting. This is the core asset powering everything very much like Google’s search is powering all its verticals.

But when it comes to commerce, what got the company so far can be a challenge for the future. Specifically, there are a few things that will be roadblocks:

Public market expectations
Finite real estate
The Facebook Brand
Opportunity Cost
Identity

Public market expectations

Most startups go through the “valley of death”.

This is the concise way of saying that even for the most successful companies, innovation is costly and to kickstart the flywheel a business – marketplaces in particular – requires heavy investments upfront which may never pay off.

Additionally as capital becomes cheaper and add-on services a strategy for marketplaces (think Opendoor), said experiments require increased costs per unit.

Facebook owns $55B in cash or equivalents. So not a showstopper so far, right?

Indeed, cash is not the core issue. Expectations are. Whilst a startup can stay away from the spotlight for years, Facebook can’t. Risky bets are likely not to be welcomed by analysts not just because of their short-term financial impact but also because they raise questions of strategic coherence. Each new service layer adds costs which are accompanied by more rigid organizational changes. Financing, delivery, insurance – all require customer support and bespoke operations.

Should Facebook start building teams around these activities who knows how that would change the company profile in the years to come? It’s a risk worth considering.

Finite Real estate

You wouldn’t buy coffee from the Coca-Cola company, would you? But you might buy Costa Coffee which is owned by Coca-Cola. And in essence, you’re buying from Coca-Cola without thinking of it like that. A brand is a powerful asset but has its limits.

Facebook diversified its brand on social networks by owning Instagram, WhatsApp and Oculus amongst others. But when it comes to marketplaces it still uses its own brand and accompanying real estate.

So far, Facebook has doubled down on two products: Marketplace and Dating.

Both of them have occupied prime real estate in the Facebook app facilitated by a navigation bar overhaul. The bottom navigation is dynamic serving whichever app the algorithm deems appropriate but currently seems to be pushing dating a lot.

The Top Facebook Updates You Need to Know: October 2020 — https://blog.hootsuite.com/facebook-updates-2/

The center of the bottom navigation is so prominent users will click on by accident. But what about all the other applications Facebook has created such as offers, movies and blood donations to name some? Surely, these can all be served by dynamic shuffling in the nav bar but to entrench a position in a user’s mind, more is needed. Facebook will need to spin off its successes for them to occupy their own space on the web and in the user’s mind. And of course that bears the question: would these be successful on their own, away from the mothership and how would that work?

The Facebook brand

Then we have the elephant in the room meta-problem: the better Facebook gets at targeting, the more creepy people deem it to be. This is not a problem that can be shrugged off; the resulting clash is more than a nuisance. It’s an orthogonality between its primary and secondary value creating activities.

To put this better in perspective, consider how a firm value chain works. The value chain represents the set of activities a firm must perform in order to compete. The key idea here is that the activities summed provide a value that is larger than the individual cost.

Each firm has primary activities that pertain to the production of its services and goods and a set of secondary activities that supercharge the primary activity.

As an example consider the cost of creating a pair of shoes.

Primary
Cost of material: £10.
Cost of labor: £10
Distribution: £10
Secondary

Design & branding: £10

Marketing: £10

Management: £10

Final retail price: £100.

The key here is that if a business is selling the material is only making £10. So is another firm assembling the shoe (labor). But it’s the business that owns the design, marketing and management of production that reaps the excess value of the final price.

And this is what great businesses do: operate an individual value chain synergetically to position itself as a dominant player in the industry value chain.

Under that lens, here’s what the Facebook Value chain looks like for marketplaces.

The primary activity for Facebook is to add more users to its network, keep them engaged, and then show them listings that are relevant to them across categories.

And all of it is underpinned by excellent targeting capabilities which in turn rely on an endless stream of information people give Facebook.

And here lies the conundrum. The better facebook gets at targeting and predicting the needs of the individual, the more disliked it becomes. It becomes “creepy”.

In all fairness, creepy is a compact way of saying that Facebook inadvertently surfaces the existential anxiety in all of us. Facebook’s algorithms expose our flaws, predict our desires and in doing so remind us how vulnerable and normal we all are.

Beyond philosophical implications, this is bad for business. Facebook is not “cool” and this is costing in terms of new community monetization and capturing value.

While disliking the brand won’t be a showstopper for the daily mindless feed scroll, it certainly is an issue for services that Facebook would be primed to capitalize but require a different level of trust and brand association.

Brand is important for a simple reason. Every purchase we make is a vote in the community and culture the brand stands for.

eBay introduced a world of discovery and bargain. Staying at an Airbnb still maintains the edge of the experimental. Etsy supports the creators.

But Facebook is a threat.

Would anyone want Facebook to become their general practitioner, insurer and bank with which they pay both of the above? Unlikely.

Opportunity cost

Corporate strategy argues the firm should leverage its existing assets to optimize the chances of continuous innovation. In practice this means that you don’t expect a grocer to produce shoes or a fashion retailer to do research on chemicals. Each to their own.

Similarly, Facebook’s core asset is the astounding engagement of its user base. From the perspective of building marketplaces, Facebook has no cost of customer acquisition (CAC). Its users keep coming back and that is an opportunity to promote more ads.

Facebook advertising cohort net revenue, Mark Tsirekas

This creates a virtuous flywheel. User A brings User B. Together they stay longer in the network and consume more ads which in turn help businesses and Facebook grow.

There is a stark difference between the growth curve of the advertising business and Facebook’s. Whilst the business grows its cost and revenue somewhat proportionately, Facebook’s net revenue increases while its cost structure remains the same. In other words, Facebook’s profit margin from advertising increases without further effort.

This is part of the Facebook magic. And so here lies the other question:

When does it make sense to risk enraging your advertisers to pursue a new business line?

Identity

When a user posts something in the marketplace, it is visible to all. Again, staying true to the flywheel logic this is perfectly sensible. Every post is another attempt to keep people on their screens.

However with dating, Facebook had to break that link. The app is designed with privacy in mind and its marketing rhetoric revolves around that.

Despite the marketing attempts, users will still feel wary of being exposed to a certain extent. After all, it’s their own identity. And since dating requires time and exposure, there should be doubts over long term user retention. Same goes for all activities which are deeply personal.

So, looking ahead what is Facebook’s marketplace endeavor going to look like?

The future of Facebook marketplaces

Here’s what seems likely to happen then in the next 5 years and how nascent marketplaces should view the giant.

Facebook will invest in engineering driven services with high profitability to add value to its marketplaces (e.g. payments).
It won’t venture in categories where it would require an overhaul of mindset and cost structure (e.g. iBuying).
It is likely to launch further marketplaces in verticals with low set up costs and a strong human touch (e.g. recruitment)
Finally, for the attempts which will work we might see them to be separated from the core site/app and try to capture their own space (e.g Dating, Marketplace).

Facebook will invest in payments & engineering driven solutions

Facebook will aim to capture the transaction in-network. Whether that is within the Facebook marketplace, through WhatsApp or as a tool via Facebook Pay the intent is clear: own payments when a transaction occurs.

This makes a lot of sense. Payments bode well with the general strategy of the business. Rely on technical implementation, have low marginal costs and keep the users in the network.

Similar services: ID & verification, insurance underwriting, psychometric matching.

Won’t venture in new categories

Facebook is unlikely to go after innovations that require an overhaul in its structure and mindset.

For Uber to scale, Uber partnered with maps, car manufacturers and even set up physical stores to serve their drivers’ queries. And Uber is still in deficit.

Whilst Facebook could rely on individuals to create its own mobility network, immersing itself in the operational side of it seems unlikely.

Service based marketplaces

But it is likely that Facebook will not stop fighting for services. Should dating work, this will herald a new playbook for Facebook. One where it can monetize 1-to-1 relationships.

And in services there’s all that. Recruitment being a close relative to dating and a huge market. And given Facebook is already eyeing up work, similar experiments in job seeking seem sensible.

Facebook will spin off apps

As Facebook tries to deal with the finite space it has to entrench itself as many more things than a social network in the mind of its users, it makes sense to spin off its successful apps to extend the real estate it owns within our smartphones.

Consider this as an emancipation moment and a filtering process:

Is an app established enough to stand and grow on its own? Spin it off

Is an app in need of the Facebook mothership? Position it in the middle of the users’ screen till it becomes a habit.

What does this mean for vertical marketplaces?

In the categories Facebook will venture into, marketplaces will face increased competition. This might not be felt immediately as the eCommerce market is growing but it will intensify the need for differentiation.

I expect that more and more marketplaces will compete on the Experience layer of the transaction (delivery, fintech, services) which in turn implies a higher cost structure to start a marketplace.

For the moment, this is ok since Venture Capital is at an all time high. From a cultural standpoint it becomes difficult to be an founder running an indie marketplace (like yours truly).

However it seems unlikely that Facebook, despite its unfair advantage, will become an unstoppable marketplace factory anytime soon and likely that consumers will reap the benefit of increased competition and corresponding innovation.

Interesting times ahead.

Delivering food, capturing attention: what restaurants can learn from the publishing industry

Before the internet, publishers were in a very privileged position. Controlling the printing press was a true competitive advantage. Similar to other mass-produced goods, the machinery necessary to replicate the news of the day at national scale came at a high cost and the right to participate in the news creation did so too. As a result, the publishers controlled their geographies, their audience, and the advertisers who wanted to reach their audience.

Then along came the internet. Fixed costs for content production dropped down to (almost) zero. What used to be restricted information became abundant and personalization rules in abundance. Google and Facebook are the clear winners of the information economy and advertising online was the prize.

Mistaking a platform for a channel

Both won by imposing their own rules in the discovery and distribution of content, and eventually owning the audience. Initially, both were heralded as great marketing innovations.

Publishers were mesmerized at the potential of a new global, growing reach. Before programmatic advertising and the maturity of hyper-targeting this was great news. Google and Facebook were seen as channels serving them, not as platforms serving themselves.

The shift from paper to digital delivery of news left publishers with a fraction of the advertising revenue they used to gain and a steep decline of the print circulation revenue. A double whammy.

This has been a bleak and deteriorating reality ever since. The scariest part of which is that it’s been 26 years since major publications launched online and it’s only now that the first signs of economic sustainability appear.

Perhaps not obvious at first, but the restaurant industry is following the same footsteps, commoditizing itself one delivery at a time.

Food delivery as a utility.

In the UK, Deliveroo, like many a great company, started with providing value to a niche. Casual dining was the target. Connecting those establishments which didn’t offer delivery to the upper-end consumer who valued quality and convenience above cost, the goal.

“We are a restaurant, not a takeaway place” was a common thought for a lot of restaurants initially.

Deliveroo’s offer (and their sales) was hard to ignore. Off the back of the normal operation, why not make extra money? Deliveroo would make money from the extra fees clients paid. The restaurant didn’t pay but a small fee (if anything).

In the beginning, Deliveroo was the perfect utility. Since the cost was zero or very low, restaurants were incentivized to inform their local base that now they could reach them even more conveniently; they could deliver the same quality of food at home.

This was fantastic and it didn’t affect the restaurant’s economics. Deliveroo was a (near) free marketing channel.

Abundance, commodity and owning discovery

While hindsight is 20/20, in the moment it is not easy to spot the inflection point where a short-term opportunity turns to a perilous force. And this is because there was no qualitative change in the nature of the service to inform of the upcoming disruption. It was the users’ attention that shifted as a result of repetitively using the service.

Initially, customers would use Deliveroo or the other delivery companies (mainly Uber Eats) to get their favorite meals delivered, from their favorite places delivered. As the value of the service was higher than the delivery fees associated, usage increased and eventually the nature of the question changed from “how can we get X delivered” to “What are we eating tonight?”.

What started as an “add-to-basket-and-go” experience, was now turning into a discovery session.

As customers got habituated to using one place to get food delivered and, more importantly, to get inspiration, the power dynamics gradually shifted. Customers had less of a reason to stay loyal to the local pizza shop nearby. Why would they? Another authentic Italian Pizza place with great reviews was now able to reach them, sometimes enticing them at an even lower price, courtesy of the delivery companies.

To the discerning eye, the power to sway customer preferences is a pivotal moment. The channel was on its way to become the platform.

At the same time, this was but an imperceptible change for restaurants themselves; each place kept seeing their delivery revenue steadily increase and so did the confidence they placed on this quickly growing stream of revenue, devoid of quantitative concerns stemming from a drop in customer lifetime revenue. After all, such are the concerns befitting a tech company, not a restaurant. To their detriment, this is a view a lot of restaurant owners share as they prefer focus on the creative part of the experience.

The delicate balance of indulging their customers in trying the various options available whilst continuously growing revenues for restaurants was sponsored by the humongous growth of the delivery economy with the Venture Capital industry picking up the bill. In other words, the volume of orders for each restaurant was a derivative of market growth, not the performance of an individual operation.

The result might seem obvious now but not so much then, especially not from the perspective of the restaurateur who experienced a booming business. For every new order they accepted however, the relationship with existing clients was diminishing and new clients were not represented by a name; a fleeting order ID sufficed.

More choice for clients equaled a lower volume of repeat orders to individual restaurants. In return, as restaurants had a wider addressable reach of clients, an equilibrium was found.

Finally, customers were now loyal to food delivery companies which dictated the discovery, delivery, and payment for food.

And for this endless growth to be sustained the rake doubled to about 30%, charges were placed on the restaurants rather than the end consumer (or both) and delivery fees became “variable” (increased). Delivery companies now control the value chain and it’s time to squeeze as much value of the proceeds as possible, leading restaurants to re-think their razor-thin margins and business relationship with delivery companies.

Sound familiar?

COVID and the acceleration of disruption in the restaurant industry.

This was a slow process but then COVID-19 happened. And any lingering doubts were removed. Restaurants had two options: sign up to the delivery platforms or shut down. Or both. But one thing was clear: this was not the promised land they signed up for.

At the current pace and given the dynamics of the industry here’s what to expect.

Kitchen and restaurant unbundling. Low cost space “dark kitchens” become mainstream. Restaurants are struggling to compete in an industry where their sole contribution to the value chain is a commoditized product. All other value-adding services have been stripped away from them. Marketing, payments, delivery operations, and real estate (dark kitchens) belong to delivery platforms.

And just like that, restaurants are in the shoes of the publishers. Losing their customers attention means gradually giving up on their sovereignty.

Dominant marketplaces inevitably give birth to individual creators

But we have seen this cycle before. When a network reaches maturity and moves to a value capture phase there’s inevitably an economic incentive for creators, solopreneurs and SMEs to compete against the network and capture value from customers on the fringes of the network, kicking in another cycle of creative change.

As Amazon grows to unprecedented level, Shopify is growing even faster.

Substack grows in tandem with Medium and traditional media as the appetite for content is at an all time high.

This should inform us as to what might happen in the restaurant industry too. Restaurants have traditionally been low-tech, focused on fly-by traffic and didn’t have incentive enough to act any different.

However, there have been significant changes in the past 5 years. It is now possible to outsource delivery, have a direct relationship with customers and finally, capture the whole transactional value. And of course, now it is a matter of survival. How can they compete? Customer loyalty.

Lessons from the publishing industry moving forward

I’d be surprised if the dominance of the platforms was to be challenged anytime soon. In a way, it’s too soon. It’s the boiling frog effect once again and things won’t change before they get any worse. Delivery platforms play a significant part in keeping the lights on and, as such, it’s difficult to envision radical change.

But were things to take on a drastic change of course there’s a few lessons from the publishing industry and there is a few reasons for drawing that parallel.

In the attempt to own their audiences, publications have turned to multiple media and in doing so generate revenue from multiple sources. Subscriptions are starting to work, podcasts attract a different set of advertisers, events capture the value of an audience and social media drives premium newsletter sign ups that allows publishers to connect with their audience. Additionally, these exact same tools are available to micro, small or large enterprises.

A similar path seems possible for the restaurant industry.

From loyal customer to member: a viable path ahead

Despite the recent slump in retention, restaurants are indeed a high loyalty sector which becomes obvious with a quick glance at Deloitte’s report on restaurant spend. 49% of visitors spend at least 25% of their total restaurant budget in one place.

Deloitte, Second Helpings:Building consumer loyalty
in the fast service and casual
dining restaurant sector

A loyal customer would be someone eating there once per month which looking at the graph above should be represented by the segment of people who spend 25% of more of their total restaurant budget, assuming eating out once per week that would be once per month. If that was the case there’s 49% of people that each given restaurant could target to convert.

Of the roughly 100 customers an average restaurant (25 seater) serves per day, or 3,000 per month we can assume that at least 10% will become repeat customers. This means that a restaurant hosts roughly 300 loyal customers monthly.

The second important metric that Deloitte points out was that (in casual dining) 9% claimed they’d be amenable to a subscription offered by the restaurants they visit and they’re likely to dine again.

It’s reasonable for a restaurant to convert 10% of its 300 loyal customers each month and expect these customers to do so with an increased spend. This would equal 1% of the total footfall.

In other words, this restaurant, assuming no churn for simplicity, should aim at 360 members in a year.
Taking an Italian restaurant as an example and £15 as the cost of a pizza, it’s safe to assume the average member spend could be £25, the price of two pizzas discounted by ~16%, as an example of a simple membership perk.

That would amount to gross sales of £108,000 in annual recurring revenue or roughly 60% of the total revenue the restaurant would need to be viable.

The restaurant owns its audience.

Starting as low tech as a QR code on the food box offering a perk and later through a combination of email, SMS or even a Facebook Group the new wave of restaurant operators would have the ability and the mindset to harness a community.

In an era where people are spending more time than ever in their local area and their social fabric has been eroded, the community feel (even virtual to begin with) is more important than ever.

Community, multiple touchpoints and revenue diversification.

On the basis of creating a community, a restaurant would seek to use an array of channels to deliver and capture a higher part of the value of their members. Educate on all culinary topics, offer exclusive perks (e.g. private dining after hours is a great one), exclusive meals (specials of the day) and/or cooking classes “how to cook X on your own” are some very basic examples. Especially since once owning a community, monetization opportunities present themselves.

Restaurateurs as creators

Operating a community, subscription management, and payments is a solved problem and very much in the scope of being a creator. Finally, delivery can be outsourced to a 3rd party, dedicated delivery service companies like Stuart.

And this is already a big part of the value chain.

Expect steps in this space and along these lines to be few and far in between as well as clunky for a while until dedicated SaaS platforms (an opportunity there) to arise to streamline these operations for a fraction of the cost.

I’d be keen to hear from operators. If you’re in the space thinking along these lines, feel free to reach out, I’d like to hear from you.

Notes

Restaurants* there is an implicit assumption that for sheer takeaway places things might stay the same since their margins were already low; in this article I mainly refer to restaurants offering seating which also didn’t offer delivery traditionally aka the casual dining part of the market.

When Managed Marketplaces Don’t Work And Why SaaS Enabled Ones Do

Sometimes a marketplace is flawed and headed for a tough time before it even begins. But it’s not unlikely that this does not become evident until £20m has been invested. So, it begs the question: how can you know if that is the case and what can you do about it?

Network effects, liquidity, TAM, are some of the many things that a marketplace needs to get right. All very aptly described by Bill Gurley here. The list in the post is informative, concise and certainly correct. But if we were to put these in a timeline, gap related to the execution in the middle emerges. For instance, you can know if you have a big market from the get-go and by the time you experience network effects it all gets easier however, everything in the middle is murky, complicated and multivariate.

In a 2018 post, Eli Chait published a very insightful piece where he correlates the fragmentation of buyers to marketplace success.

Effectively, his research reduces the complexity of the first list to one metric that’s a bit meta; he’s positing that for a marketplace to be successful there is an inherent characteristic the market needs to possess and that is a very large amount of buyers (1 million buyers for one business) and for those buyers to also be fragmented.

This is a useful point because it identifies the principal component of success amongst many other variables and thus simplifies the problem. Can you get a million buyers? If not, then the cards are stacked against you.

Once more, if you have 400,000 buyers already this is a meaningful insight. You can approximate the value per added buyer in the network and roughly calculate if you can escape velocity through network effects.

But what if you’re just starting? What if you have 100, 1000 or even 10,000 buyers? Are the above enough to ensure that a marketplace is the right model for a given industry? How can you tell if your unit economics can mature to reflect a £1bn marketplace?

Perhaps there is a way to know what will NOT work. And there if there’s one thing to steer clear from is services with inherent repeatability of transactions between the same supplier and buyer. If that is the case, then scaling a marketplace to the 1 million buyers, 1 billion GMV will be tough.

And there is both an intuitive and analytical approach to explain why. But first, why is the lack of repeatability a benefit? Uber is a good example.

From the perspective of the supplier (driver), Ubers works like this. A driver can reach clients previously unreachable and get access to a set of value-adding services to make the job easier. In exchange Uber takes a fee per ride. Every time a new rider connects with a driver, Uber has taken care of: finding the client, navigation, safety measures and, seamless payment in the end. But since the whole CLV is (most likely) equal to the value of the ride, the next time the driver picks up a new rider, (s)he will reap the same exact benefits from Uber and Uber will receive the same fee.

In other words, supply and demand see each other as a commodity. Therefore a platform facilitates the transaction. Uber is there to ensure that a code of conduct is enforced (safety, trust, politeness etc) and enables transactions that would not be possible before (eBay, AirBnB are also good examples). This is an example of Bill Gurley’s point on “technology expanding the addressable market”.

From the supplier’s perspective, Uber’s value per ride is the same.

However, there are industries where the above hold true and still have fundamental issues in their economics, because of the loyalty repeat transactions create. Take therapy for example. The more therapists, the better the matching. A marketplace can expand the market via remote therapy, matching and also by creating the necessary privacy and trust for clients to make the first step. It’d be closer to how eBay grows the market (remove friction) than how AirBnB, Fiver or Uber do it (without them it’d be impossible).

But still, in the UK alone there are at least 2,500,000 people that will receive privately funded therapy per year. Trust is necessary initially. Structural and attitudinal barriers can both be addressed by marketplaces and therapists have a constant need for new clients as some come and some go. Yet, a therapy marketplace won’t work. Why? Because of the repeatability of transactions. A therapist is not able to see more than 25 clients (max) per week (50-75 uniques per year) and a client will only see one therapist.

That would not be a problem if the marketplace could capture a chunk of the value across the lifetime of therapy. However, the value-add of a marketplace declines after the first session because the client forges a trust relationship with their therapist. By definition, therapy is about the trusted relationship.

From the perspective of a therapist, a therapy marketplace exists for referrals.

In fact, from the perspective of the therapist, the marketplace doesn’t only stop adding value but can detract value. After the first session, a therapist will start adding notes for a client, have constant communication with them via their preferred medium (e.g. WhatsApp), receive the payment in a centralized way (e.g. PayPal) and agree on a fixed schedule which on a per-client basis at least, is simple.

Consequently, after the referral is done, the marketplace introduces a cognitive cost to use its add-on services (e.g. automated payment collection). If the therapist wants to use the marketplace they have to do extra admin on another space to keep their operations centralized.

Therefore, taking the client away from the platform is natural. Disintermediation is inevitable and thus, CLV real < CLV expected, which leads to unsustainable economics and does not allow the marketplace to scale.

Enter SaaS enabled marketplaces, or SEM

So, repeat transactions present a structural barrier for the marketplace to achieve strong unit economics and scale. Still, suppliers do need services to grow and manage a practice.

This is where a SaaS-enabled marketplace can provide a solution for a few reasons.

The positioning of the product is not necessarily around lead generation
Consequently, pricing is not pegged on “getting more clients”
Finally, the supplier might bring the clients onto the platform voluntarily which alleviates the platform from the burden of CAC but also positions it to capture value across the lifetime of the client.

From a cash flow perspective, this allows for a direct reinvestment of revenue per supplier to acquire new suppliers. Eventually, when there is a critical mass, then the platform can aggregate them and introduce a new service on lead generation but without the pressure of that being a promise. Shopify’s Shop app is a good example of that.

So, are repeat transactions intrinsic between supply and demand in a market? If so, it will erode the marketplace unit economics and pose challenges in scaling to network effects. Looking at the “come for the tool, stay for the network” approach is the more appropriate strategy for these kinds of markets and can still take to the same end point but in a more sustainable way.