Kenya: Investment Landscape

I have just completed a little under a week in Nairobi. It was a very enjoyable and enlightening experience — as well as being my first footsteps in East Africa.

After a quick tour of the picturesque Karen Blixen Museum, I met with a couple of local VCs to discuss opportunities in the region. I was curious to establish what sort of businesses were being started in East Africa, what the local risks were and to try to start building a picture of the investing landscape generally.  In no particular order:

  • Bitcoin & Mobile Money. Kenya’s M-Pesa is according to some reports involved in 70% of Kenyan GDP. Given the 1% merchant fee, this means its owner, Safaricom, might be generating gross revenues of 0.7% of GDP from this mobile payment system. Copy-cats such as EVC (by Hormud Telecom) in Somalia have pretty much replaced cash: that country may indeed (somewhat ironically) to all intents & purposes be the world’s first cashless society. These high penetration rates probably mean that bitcoin (via payment systems such as the excellent Bitpesa) are going to find it hard to break in, except for transactions such as overseas remittances.
  • Economic Openness. Of the big economies of the region (Kenya, Uganda, Ethiopia, Tanzania), Kenya probably offers the easiest path for overseas investors: it has been a relatively open economy for a long time, particularly compared with Ethiopia and Tanzania, so foreign investors can be fairly confident that what they put in, they should be able to get out again. Having said that, sadly there is still corruption so careful due diligence is key.
  • China. Although this is changing, being less commodity-oriented than, say Zambia, the East African economies have been less adversely affected by the significant weakening of commodity prices. However, as with Zambia, China has made infrastructure deals locally that governments are already in some cases coming to regret. As in other African countries, when the Chinese offer money for infrastructure projects it turns out to be a bit of an illusion: they bring their own companies, who import Chinese workers & materials. Thus there is little or no “multiplier” effect (although clearly the local economy still benefits from the new road, railway or whatever).
  • Logistics. Kenya’s not a small country: 45 million people covering an area a little smaller than Texas. Although it has some great roads, it also has some absolutely terrible ones. This means that getting raw materials & finished goods around the country becomes a significant challenge and represent a much larger cost than in Europe or the USA. Driving outside of Nairobi, I saw fully-laden trucks doing less than 10km/h – the road between Mombasa on the coast and Nairobi is, I hear, literally nose-to-tail with trucks crawling along. One study I was told about suggested that the most efficient form of transport here in some circumstances was still the donkey. Another study pointed out that small farmers in Kenya receive substantially lower prices for their goods than are available at the port or other main market-place — in some cases as much as 75% lower; but when these researchers investigated the costs for the farmer of actually getting their production to that market, they found little evidence that intermediaries were squeezing value from the farmers: transportation costs & losses due to theft, wastage, contamination etc were all far more important. This would suggest that there is massive scope for enhancing the incomes of small farmers by improving logistics — something that should not be beyond the wit of man.
  • Education. As I mentioned in another post, Kenyans spend ~45% of their income educating their kids. Yet inconsistent teaching in public schools & very large class sizes means results can be very poor. Bridge International Academies, a company founded in 2009 and backed by Zuckerberg & Gates as well as Pearson has been seeking to address this problem. Bridge-trained teachers read scripted lesson plans & classes are much smaller than in state schools. Teacher absenteeism is substantially lower than in the state sector, too. The approach is not without its critics, as the WSJ article highlights: for example, many of the teachers are themselves barely out of school and the training provided is nowhere near as extensive as a ‘full’ teaching qualification. But the statistics speak for themselves: a 0.32 StDev gain in reading & a 0.51 StDev gain in mathematics.
  • Agriculture. Kenya has 6 million small-scale farmers and agriculture represents more than 27% of GDP. Complex logistics are not the only problem: other issues include lack of price transparency, fraud & theft. It is difficult to quantify the impact these challenges have on productivity & farmers’ incomes; to date I have not been able to locate any studies that look at this.
  • Manufacturing. Many Kenyans are self-employed or work for very small-scale enterprises making all sorts of goods including jewellery & other ‘craft’ type products.  For some, this is subsistence-level whilst others may be substantially more successful. What they all have in common though is being limited in their access to markets.  Given the existence of the internet and high local penetration for mobile money, there’s no real reason a Nairobi artisan couldn’t set up a Shopify shop and tap into global demand for ethically-produced, almost-unique items; yet in practice that doesn’t seem to be happening. What these artisans need is a “bridge” to the world.

I met with a number of businesses that are seeking to solve these problems, or helping other businesses solve them.

Many of my meetings were arranged by Growth Africa. They help early stage businesses hone their business plans, polish their presentation materials and raise working capital. I could not have been more impressed with the standard of the firms they’re working with and would recommend any young entrepreneur thinking of starting a technology business meet with them.

Below I feature the four early-stage companies I was most impressed by. Again, in no particular order. Although in some cases I highlight the positive social impact of these start-ups that’s not to say I don’t think they also all look like great standalone investments (subject to due diligence, of course).


Essentially, social selling. If you have a fan-base on Facebook, Pinterest or whatever, and you have products to sell (including digital music, event tickets or physical goods), why go to the trouble of setting up a separate shop on say Shopify (and risk losing the customer to a competitor) when you can sell direct from your FB page? Very smart, articulate founders; slick software; sensible revenue model. With 120 million Facebook users already in Africa, there should be plenty of scope for growth in this kind of application.


Soko offers jewellery from over 1,000 local artisans via a single portal; slick ERP software allows them to hold very low inventory of unfinished products. They can help artisans with training and also with working capital to invest in additional tooling. Finished jewellery is sold both wholesale & retail and the revenues are growing strongly. If you believe at all in “trade not aid”, this type of business is one NGOs should be taking a serious look at, as the positive impact on local communities can be huge.


By providing a platform that allows small-scale farmers to group together into collectives, tracking inventory levels, offering trusted transportation partners & connecting suppliers & consumers, iProcure is making the agricultural supply chain much more efficient. Again, the social impact to remote rural communities is potentially significant.

Mara Moja

Although Uber exists in Nairobi as well (presumably) as in other African cities, there’s a problem. Star ratings for drivers are all very well, but the Kenyans don’t really trust them. Instead, they’d rather use their ‘usual’ driver. And if he’s not available, they’ll ask a friend if they can borrow theirs. Mara Moja has cleverly turned this into an app by using people’s social media profiles & relationships to highlight to a rider those drivers that are used by people in their network. According to their research, a Nairobi rider would rather use a 2-star rated driver with connections to their friends, than an unknown 5-star driver. Which is pretty amazing.


Investing in Africa has never been regarded as easy by most people; it still has significant challenges, but for sub-Saharan Africa excluding South Africa I’d say Kenya was a very interesting first port of call for anyone looking to play the “Africa Rising” story. Nairobi has certainly earned itself a reputation as the strongest local technology hub and from what I managed to gather over just a few days it is now spawning some very interesting ideas and startups.

Jersey Bandwidth Sucks, @PhilipOzouf

Dear Philip

Thanks for following up on my Twitter comment.
As you’re aware, if the Island wishes to attract high quality businesses in almost any sector, but especially in technology then good connectivity is essential.
Over dinner last night with some local friends, I learned of one business that chose not to relocate to Jersey (with the loss of around 10 potential jobs created) because when they investigated Jersey’s services the feedback they received from local firms was, essentially, that most things were a rip-off and connectivity was terrible.
I have to say I agree.
I’m in the office in Queen Street using our JT Broadband. I spent 20 minutes in a hold queue to their Business Support team in order to find out the details of our contract. We pay £80/month for ‘Business Pro’ and for that we get ~20MB downstream with max 20:1 contention. In reality, as I write this we have 0.75MB upstream which is pretty pathetic. According to this report, there are cities in Romania that manage download speeds of over 100MB. In London, according to this report from Ookla, Virgin Media offers speeds of 136MB down / 44MB up: 50x faster uploads than Jersey! Even when JT manages to deploy fibre to our office, the best we can hope for is 50MB down / 5MB up.
At home in Corbiere, I have Newtel and the bandwidth is diabolical: typically 1MB down and less than 0.10MB up. That’s barely even describable as broadband. Yet today many people work from home and so it’s essential to provide higher-quality connectivity.
You probably saw my article on my experience with Sure and my mobile phone: over £6,000 of data roaming charges (charges that were already at least 10x what most UK providers would have levied) were eventually reversed when I demonstrated that the SIM was not correctly recognizing my physical location. Furthermore, when I’m in Jersey the 4G is poor and spotty.
This is not a package that looks attractive to technology businesses.
I look forward to your comments.

TalkTalk: Not Walking the Walk

A week ago, TalkTalk was hacked.

The company claims that hackers don’t have enough information from the hack alone to steal money from their customers. It’s unable to confirm how many of the 4 million customers were affected.

Having failed as a business to protect their customers’ confidential information (one for the Data Commissioner, surely) they have compounded this by treating their customers very badly. For example, they refuse to waive cancellation fees unless the customer has actually lost money. Their CEO is unrepentant.

Thinking about this the other day, it occurred to me that TalkTalk must have breached their own Terms & Conditions when they didn’t look after the data properly and that customers would be able to challenge them on this, and therefore terminate contracts without penalty. Turns out that is indeed the case, as This is Money explain in this helpful article.

Companies need to take data protection seriously. TalkTalk doesn’t Walk the Walk.

Customers should talk with their feet.


National Health Service: Not National At All!

Back in the mid-1980s when (as a freshly-minted science graduate) I joined Shell International Trading Company as a programmer, PCs were only just starting to appear on people’s desks: most of the ‘real’ work was done by big old mainframes located in Shell’s data centre at Wythenshawe, just outside Manchester. Oil traders, just like those involved in financial markets, are an impatient bunch and so whenever their systems didn’t perform quickly enough, they’d beat up on the Information Technology department to improve performance. The knee-jerk was always for the mainframe people to want to buy more hardware: faster disks, more CPUs, more memory, etc. My view then – and now – is that throwing more cash at the problem without really understanding the issues is basically throwing the money away. I think there’s a risk that’s what is happening in the NHS today. The NHS is seen by British people and politicians as so much a part of the fabric of the nation that its budget is not only sacrosanct – the political parties are actually competing to throw more money at it, even in these straitened times.

The NHS has a budget of £110 billion and employs 1.4 million people.

On a recent ski trip with my sons, we met three young British medics. They were a lot of fun, as well as being (of course) intelligent and I had some interesting and surprising conversations with them about medicine and Britain’s National Health Service. It turns out we don’t really have a National Health Service at all: at best, I think you could describe it as a loose alliance of healthcare service providers who all happen to be funded by central government.

One of our skiing medic friends was a pathologist. He told us how money was often wasted in his department. Here’s a typical example: a patient visits their local doctor (GP) with some ailment; perhaps fearing some form of cancer might be the cause, the doctor refers the matter to a consultant for further study. The consultant asks for a biopsy and in due course this is passed to the pathology laboratory for analysis. So far, so uncontroversial. Perhaps by this time it’s a few weeks since the patient first visited their GP. Eventually, a pathologist gets around to putting the sample under a microscope and inspecting the cells, trying to figure out if there’s some malignancy. Having done so, they telephone or write to the consultant with their conclusions, at which point they discover that in the meantime, the patient has died. This is not an isolated occurrence. Now, training a pathologist takes many years and in the private sector one imagines that their charge-out rate would be of the same order as decent lawyers: maybe $500 per hour? So if they spend a couple of hours on a slide for a dead person, the NHS has just wasted $1,000 because nobody told them not to bother. It’s easy to see how this can happen – after all, the GP and the consultant will both have dozens if not hundreds of patients on their books at any one time, and they’re expecting to have to react to new information, without time to proactively be monitoring on the situation for each person. It’s also easy to see how one, simple, joined-up computer system could help avoid the waste.

But the NHS doesn’t have one, single computer system. It has hundreds – perhaps even thousands. I was told about one hospital that had developed a pretty good computer system for handling patient records, including as much medical background as they were able to obtain, so that all departments in the hospital had the same information. They discussed with other local hospitals and doctors’ surgeries rolling out the same software so that, for example, a patient in a road traffic accident brought into one hospital would immediately have available all medical data regardless of which local doctor they used. Sounds entirely logical, but it didn’t happen because each technology department had their own, specific, mandate and there was no financial incentive for them to adopt another hospital’s system. Some hospitals even have different computer systems for different departments, and of course they don’t speak to one another…. Now, all that incurs massive costs for the system collectively: for every patient admitted to each hospital, someone has to ring around the other hospitals and GP surgeries to find that patient’s medical data before treatment can occur. The Data Protection Act doesn’t help, as often the gate-keepers to these (paper and electronic) records see only downside risk in sharing a patient’s medical files with another hospital – even in a life-or-death situation. Again, this is a pretty easy problem to cure: if the NHS is to have an internal market, there needs to be an internal market for data, too. So if I have to call up and get records manually, there’s a price. Add up the cost and pretty soon the bean-counters would figure that harmonizing their systems with neighbouring healthcare providers would make sense. And NHS has a data ‘backbone’ capable of handling all this securely: N3.

Scheduling is another area that seems to be handled sub-optimally. Here are two examples.

I’ve heard of patients being sent by taxi from Cornwall to London (at least four hours drive) in order to visit a cardiologist because there isn’t one in the southwest. That can’t be a good use of NHS funds. Surely it would be possible to collect together all the patients in need of a cardiologist in that region and fly one down for a day? And why isn’t there one there, anyway?

I’ve also heard of ward beds being ‘blocked’ by a patient that should really have been discharged, but no ambulance or relative was available to take them home. I’m guessing a ward bed costs £1,000 per day, so put the patient in a taxi, for Heaven’s sake! In the end that’s what the registrar did, but of course there was much consternation over how to account for that, financially.

Bottom line: there’s inefficiency within hospitals and there’s inefficiency between hospitals, and the system’s not currently set up to deal with either particularly well. And throwing more money at the problem isn’t going to make it go away – in fact it’ll probably just make it worse.

Now, large government-backed computer projects are famously disaster-prone the world over. In 2011 NHS scrapped a £12bn computer system, for example. But that’s not what is needed here. The technology departments in hospitals and surgeries can learn from the way modern software projects tend to work: a cycle of prototype and refinement, and gradual adoption rather than a “big bang”. So, one hospital develops some software with nifty features, neighbouring hospitals adopt it and pretty soon you have a large enough collective budget to tweak version 1.0 to add all the features those other hospitals need. So then more hospitals join and so forth.

Next time someone tells you, “The NHS is in crisis,” think about why that might be.


This article first appeared on LinkedIn 30 December 2014 

Putting the Gini Back in the Bottle

An article by The Economist on Twitter caught my eye a couple of days ago.

It shows that the wealthiest 0.1% of Americans control around 22% of the nation’s wealth — almost exactly the same proportion as the bottom 90% of citizens. This hasn’t happened since the late 1930s and the article seems to be lamenting the passing of a period where wealth was more evenly distributed.

One problem here is that there probably isn’t any really good data on income or wealth inequality going back further than the early 1900s; because I’m pretty sure that if there were, we’d see that the post-WW2 period was the outlier.


I’m not arguing that wealth inequality is necessarily good, just that it’s probably natural.

I’m not too concerned about the Paris Hiltons of the world…. “rags to riches and back again in three generations” comes to mind; I’m sure they’ll fritter it all away if they’re not smart enough to look after it — or their parents will follow Warren Buffet and give it all away rather than passing it to his offspring.

At the other end of the spectrum though, think about the technology sector, for example. Innovations like Apple’s App Store have made application development tremendously accessible: it’s literally now possible for a team of maybe half a dozen teenagers with a good idea and some coding skills to put together an app in a relatively short space of time that could gross tens or even hundreds of millions of dollars in revenues. Not every app is going to succeed, but how many pursuits were there in, say 1965 (when I was born) where someone potentially without any formal qualifications and with very little capital could make that kind of money?

On a slightly larger scale, look at Finnish game developer SuperCell. Following $12mn of VC investment in 2011, they developed two games, one of which was Clash of Clans — an app that grossed nearly $900mn in 2013. I don’t know how many employees they had in 2011-12 when the game was developed, but I’ll bet it was a pretty small core development team.

To me, this is a kind of democratization – literally with nothing but a laptop and an idea, it’s possible for a talented coder anywhere in the world to come up with something pretty awesome and sell it globally. Of course, if they succeed they will make a ton of money and in doing so “mess up” the Gini coefficient; but honestly, who’d want to live in a world without that possibility?


This article first appeared on LinkedIn 11 November 2014

Mobile Telephony

I had an interesting and salutary experience recently when travelling in Germany. I was using my Blackberry to read my emails, careful (I thought) to be connecting to WiFi in the airport, hotel and so forth.

When I got back to Jersey, I received a text message from my provider, Sure, alerting me that I had high mobile data usage. I was surprised, since I thought I’d been using WiFi almost all the time.

It seems that even when the little WiFi logo is showing on my phone, it may decide (according to Sure) that the connection is insufficiently stable and use GPRS instead.

When I challenged the bill, the only itemisation they were willing to provide showed dates & times of downloads during my stay. One of these — upon arrival in Munich and whilst I was still in the airport on WiFi — was for 500MB of data in a single transaction. I asked for more details and was told nothing further was available. Since I don’t watch movies on my Blackberry, it’s hard to imagine what that 500MB could possibly have been, notwithstanding that in any case I had been under the impression I was using WiFi not Data Roaming.

That really surprised me — in an era when all our electronic correspondence is in all likelihood being monitored by the security services, and when even our internet search history is being recorded, is my telecoms provider seriously claiming that they don’t know what IPs I connected to and/or what was being downloaded? I find that incredibly difficult to believe.

What makes it even more interesting is that effectively this means that Sure can just make up my bills — there is no challenge I can apparently make to these charges: as far as they are concerned, their word is law. Yet for fixed line telephony there have been numerous examples of customers being billed for calls that were never made, so what sanction do I have? I’m not even sure there’s a regulator in Jersey I can refer the matter to.

This little problem also revealed just how awful the Sure customer service ‘experience’ is: they were totally implacable throughout the process. Needless to say, we are taking our customer elsewhere.


This article first appeared on LinkedIn 1 November 2014

Why Isn’t Software Regulated?

Great swathes of the economy are already subject to government oversight: finance is the most obvious, but of course there’s also healthcare, air travel, car safety, workplace-related legislation, food safety, building regulations, telecommunications and so forth. The list is pretty much endless.

At a recent lecture I heard that something like five times as many Americans are subject to some sort of permitting in their job as was the case in 1950s (I’ve tried to track down the statistic but can’t find it — if you have it please provide a link).

So in some sense it feels like something of an aberration that something as pervasive as software would have no regulation attached to it.

There are not many activities where software does not in some way impinge on our lives these days, after all. And while it may not matter to anyone if some random little game App downloaded from iTunes doesn’t work properly when you reach level 50, there are other situations where the performance of software is of very great importance.

In the UK, several banks have experienced partial or total crashes in their ATM networks (see here, for example, and here). Whilst I’m sure that the banks’ regulators would have required them to perform extensive testing of the software, and to have all sorts of complicated business continuity planning in place, I doubt the software vendors themselves have left the door open to any form of claim against them for shoddy workmanship (almost all the licence agreements one has to click in order to install software contain some sort of “no warranty” or “as is” clause which would not cut any ice if you tried the same thing when selling a car).

According to Wikipedia’s article on software bugs there have been some other pretty major problems caused by poor software, including large-scale power outages, incorrect social security payments, aircraft crashes and security breaches (most recently, Heartbleed for example).

Now, perhaps with open source software one could take the view that, “you get what you pay for”. I work in financial services, and even if I were to give my advice away free to anyone who wanted it, there are countries where the local regulator would wish me to submit to their jurisdiction (sometimes there are carve-outs, for example in USA if you have very few clients and don’t manage much money you’re not subject to as much regulation by NFA as a big institution). Not that it would be easy to regulate open source software, anyway, but even something as pervasive as Linux has attracted for-profit companies such as Red Hat which seek to package up the open source operating system and provide useful support and add-ons to make it a more ‘corporate’ product. Under that same model it would be possible to imagine Red Hat’s version of Linux, for example, coming with many more warranties about its performance in specific business environments than the plain-old ‘free’ version would have, in much the same way that one can buy any number of (unregulated) magazines which will tell you what stocks to purchase whereas regulated advice will generally cost you more but come with some additional protections (such as the prospect of making a claim against the advisor in the event the advice is inappropriate).

Don’t get me wrong — I’m really not in favour of wholesale regulation of the global economy; regulation is a dead-weight cost and there seems currently to be a tendency amongst both politicians and the general public to want to regulate as many aspects of life as possible. What we need is smarter regulation, not necessarily more.

Nevertheless, every time I use a bit of software — normally paid-for stuff like Microsoft Windows, Office, Google Apps for Business — and it doesn’t work properly, I wonder if perhaps the world has a little bit of a blindspot about software compared with what consumers (both retail and business) demand in other products.

This article first appeared on LinkedIn 28 July 2014