Navigate back to the homepage

Platforming the future: Safaricom

September 2nd, 2020 · 6 min read

“Gentlemen, there’s only two ways I know of to make money: bundling and unbundling.” - Jim Barksdale

The Great Unbundling: Banking

Safaricom unbundled the big banks in Kenya during the late 00s with mobile money transfer service M-PESA. It’s market shares stands at 99% with 40M+ customers.
In this article, I will explore options for what’s next for Safaricom from a perspective of where the internet is headed.

Success & Market Share
Back in the days, there was a notion among bankers that middle to low-income consumers were twopenny-halfpenny & you couldn’t make substantial money banking them. M-PESA confuted this. In 2007 the average distance to the nearest bank was 9.2 KM (urban areas). In 2015, the average distance to the nearest M-PESA agent was a mere 1.4 KM with 167,000 agents registered. The company IPOed in 2008 with shares at KSH 5 ($0.05). Today they are worth KSH 30 ($0.3), a 6x return with a market cap of $10B (10 times than the largest bank). With smartphone penetration at 97%, mobile transactions totaled to $43.5B annually while Subsaharan Africa recorded $456.3B in 2019. It’s worth pointing out however, the M-PESA interface feels pre-internet 13 years later. The software is boring but works. Businesses struggle with integrating the API into their modern systems.

Internet 1.0
The first phase of the internet involved two things:

  1. Connectivity.
  2. Information Arbitrage.

    We are almost done connecting everyone online. But is every industry online? Not really. The use-case is just beginning. Retail is still largely informal and offline.

Internet 2.0
The internet is evolving to quality & tastemaking. Let’s explore two case studies:

1. Google Trends. A comparison between two search terms: ‘Best’ (in blue) & ‘Cheap’ (in red). More users are interested in finding out what is best for them rather than the cheapest option in the market.

'Cheap' 'Best' searches comparison

2. TikTok. Specifically user retention & the speed they achieved product-market-fit. Average user session is ~294s with each session having >100 content units. Traditional social media trails this. Twitter session is ~114s with 50-75 content units. Tiktok has managed to attain this stickiness from their near-perfect recommendation algorithm. As you use it, gives you the impression it knows you better by the interaction. Same with Spotify. Spotify & TikTok like many digital products serve content using recommendations in-turn boosting the user experience, rendering what you just want to see.

Failed Experiments?
3 years ago (2017), Safaricom launched The model was sketchy. Safaricom acted as a retailer but it was also opened to 3rd parties. However, they cut 3rd party vendors in November 2019, with a new revamp site that is still awful.
What went wrong?
E-commerce is hard in Africa. Some facts & numbers to back this:

  • Purchase power is low: $42 is the average amount consumers spend online annually in Kenya. Globally that amount is $2,594 for the same period.
  • Last-mile challenges - Cost of transporting goods in sub-Saharan countries is 5x (per unit distance) in some sub-Saharan African countries compared to the US.

Last year, the retail revenue was $20B, but e-commerce has managed to capture $640M. That is 3%. Most consumers bought Electronics & Media ($430M) more than any other product category. Copying & relying on traditional modes to E-Commerce won’t be viable for the current market, or is it? Amazon ranks as the 13th popular site in Kenya, 2 spots ahead of Jumia.

But this doesn’t mean it’s impossible.

Platforms & Aggregators
One distinction we need to make is between Aggregators and Platforms. An aggregator sits between the user and service providers. They own the user & scale the relationship. They help users & suppliers find each other. An example is Google. Google aggregates users who come on the site to look for information. They sell these users to relevant advertisers. The main revenue is from advertising. A Platform from Bill Gates quote is ‘when the economic value of everybody that uses it, exceeds the value of the company that creates it’. An example is Apple. They sell phones but they create much greater value from the apps via the app store. It’s a whole economy.
Here is an illustration by Ben Thompson on the difference.

Aggregator,platform comparison

Platforms let their users and customers shine. My favorite analogy is Platforms are enablers while Aggregators are regulators.

At this point, we need to answer two questions:

  1. Was/Is Masoko an aggregator or a platform?
  2. What are the alternative paths for the product?

My answer to the first question: An aggregator.

Did suppliers gain any more benefits than listings? Did the users get any more value than being connected to suppliers? Users churned to classified sites like Jiji & PigiaMe, which are far better competitors with superior product & catalog.

The Marketplace

What are the alternative paths? Let’s explore the cards — no pun intended — in Safaricom hands.

Consumer Behavior

Low-income consumers are underserved by ‘traditional’ e-commerce models which aim for upper-middle to elite consumers. For consumers in sub-Saharan Africa, shopping is a daily/weekly endeavor. Instead of large malls, the retail sector is dominated by small local shops (‘Dukas’), line shop kiosks, table-top kiosks, market & roadside stalls mom-and-pop shops. The quantity is low but the frequency is high allowing them to maintain a close working relationship with their customers. Most of these shops don’t have any modern form of inventory management, no POS, low cash flow, low margins, and credit needy shopkeepers.
One important feature in these shops is the presence of an MPESA agent(s). Safaricom has managed to work with the small local shops seamlessly when it comes to money transfer.


Example of a Duka with an M-Pesa agent.

The Opportunity: Bundling of Retail

In the same way that the banking experience of the middle to low-income users was improved, their shopping experience too can. The need for an Online to Offline platform is massive.

Let’s envision how this platform would look like:

  1. 3rd Party centric.
  2. Data-Driven.
  3. An Eco-System.
1. 3rd Party Centric (3P)

This means that Safaricom acts as an enabler, rather than a retailer. Suppliers & retailers can setup up a shop in the platform and list their catalog. (In 2014, Amazon’s margin from 3P was >85% while the 1P was ~25%). The platform aims to drive maximum sales value while providing all the back-office tools needed to the retailers. The platform’s revenue will be a function of Gross Merchandise Volume (GMV). The network-effects creates a virtuous cycle, like a rolling wheel, that never ends & is hard to escape (explained below in moats).

Network effects wheel

Illustration from a16z.

2. Data-Driven

Good data creates better recommendation systems. Better recommendations create the best user experience. TikTok overstates this. Pre-Internet, customers didn’t matter much. Their aggregation at scale pushed consumers to first priority.
The gold mine: Got a date tomorrow? Safaricom knows it. Need an overdraft to sort the bill? M-PESA got you.
Safaricom through its services has managed to control the financial, communication & the internet markets. They are the gatekeepers. Such data is utilitarian in retail. With users looking for the best services/products, the platform can attain a buying decision coordination that’s unmatched, retaining users & suppliers. Data is the business, everything else is a means.

3. An Eco-System

It’s tough to build a platform that fulfills all consumers/suppliers needs. An open & friendly platform where developers & partners can build upon & integrate with the core platform is essential. Economics should be aligned to incentivize this

The network effects are two sided. Developers build (micro-apps) on the platform attracting users, which in-turn attracted developers. Again, once this wheel starts rolling, it’s extremely hard to compete with. An ecosystem is more powerful than platform lock-ins. Apple & WeChat have perfected this.

What are the advantages and moats for such a platform?


  1. Network-Effects. (discussed below)
  2. Capital Efficient: No need to set up warehouses & inventory, near-zero cost of acquiring users & suppliers since they’re already captured in M-PESA.
  3. Unlimited selection of products: Everything you need can be curated. Credit can also be provided to either the user to shop or the supplier to stock.
  4. Community: With supercomputers in our pockets, most of the work will be offset to the community using the platform. Since it provides a win-win situation, everyone will work to improve the ecosystem.

Moats (Competitive Advantages)

  1. Network-Effects: They create monopoly naturally. The network gets better on every use. Users attract users by improving the experience, users attract suppliers, suppliers attract users, both attract developers to build on the platform & the cycle repeats. The wheel never stops. It’s hard to compete with this.
  2. Supply Chain: With MPESA agents just near the corner (every 1.4KM), they can act as a pickup point. So can retailers.
  3. Brand: The essence of a brand is knowing what to expect. Safaricom ranked 1st in Kenya & 10th in Africa under the top brands.
  4. Data & Identity: With the ocean of data Safaricom collects, the possibilities are endless.

Final Thoughts

“Of course, you always are bundling and unbundling. You can’t stand still…” - Jim Barksdale.

There is a striking paradox in the technology business: The rich-get-richer and you get wiped out for not innovating. It’s the deadliest sin.

Safaricom’s failure to scale M-PESA to 1.7 billion unbanked people is something between social activism & communism. It hasn’t captured the maximum value. These experiments are hard to re-run. Sometimes we are overwhelmed by our inability to do complex coordination and describe a future that is different from the present, but we should give it a shot.

Thank you for reading.

Join our email list and get notified about new content

Be the first to receive our latest content with the ability to opt-out at anytime. We promise to not spam your inbox or share your email with any third parties.

More articles from Roboflank

Media Consumption (Aug 2020)

Tracked Media Consumption for Aug 2020.

September 1st, 2020 · 1 min read


My productivity suite.

August 15th, 2020 · 1 min read
© 2020–2021 Roboflank
Link to $ to $ to $ to $