Sanity-Checking Twitter’s Valuation

Kate Craig-Wood

Kate Craig-Wood, 29/01/2010, posted in "Analysis"

Kate Craig-Wood is co-founder and MD of Memset, Britain's leading dedicated server, virtual server and green hosting technology company. In addition to her interest in using her positions to ...more info

Kate Craig-Wood is co-founder and MD of Memset, Britain's leading dedicated server, virtual server and green hosting technology company. In addition to her interest in using her positions to represent the needs of local businesses and SME's, Kate has three principal special interest areas: Green Technology, Girls into technology, and improved care for the transgendered. ...less info

• 
• 

• Tweet
• Email

• Tweet
• Email

At the end of last year Twitter signed a contract to take an investment of $50 million valuing the company at roughly $1 billion.  On a per-user basis, this valuation makes Twitter worth 1.5 times more than Facebook.

An impressive amount for a technology startup with no clear means of generating revenue, so I thought I would do some calculations of my own:

Twitter has about 18-20 million users in the US, according to Mashable

Earnings, earnings, earnings

There are 300m people in the US and 750m in Europe. Let’s assume that the ratios are the same for Twitter users. Therefore, the total number of Western Twitter users (ie. people who would potentially pay for anything) in 2010 is:

(20m / 300 ) * (750 + 300) = 70m users

To justify a $1bn (£625m) valuation they need to be able to realistically generate at least 5% of that as profit (20x earnings multiple, which is basically unheard of outside blue chip corporates) ie. at least $50m/year.

Therefore, in round numbers, they need to be able to realistically expect (in future) to generate earnings (profits) of $0.70/year from each user.

So far, perhaps not unreasonable; once micro-payment systems start working properly one could imagine users paying a notional $0.10/month to use Twitter, or advertisers paying that per user, at least so long as Twitter is the only one of its kind.

Twitter ain’t so special

That is where we realise the valuation’s big flaw: Twitter is not doing anything special. To build another system that replicates their functionality would, in my opinion, take 2 good coders, 2 good system administrators and one good web designer 6 months, tops. Add in some management and marketing capability for operational running and if you are a generous employer your wage bill might be £500k/year ($800k).

What about hosting costs? Well what it boils down to is a large and very active database:

Assuming 100m users
100 reads & 10 writes per user per day
= 10 billion reads & 1 billion writes per day

Squeeze into 16 active hours (just looking at the West)
= 120,000 read & 12,000 write transactions per second

To host a system which is capable of those transaction levels you would need at most 100 1U quad-core machines loaded with RAM (eg 24GB), costing £200/month each, or £240k/year ($385k).

So, I reckon that Twitter’s operational costs should be under $1m/year. But what do I know, right? I’ve only been designing, building and hosting massively automated online business systems for a decade, not to mention being one of the UK’s leading tech entrepreneurs.

A more rational valuation

So why are Twitter’s investors valuing such an easily-and-cheaply replicable business so highly? I suppose that they are banking on a land-grab effect and user stickiness, but we have seen what is happening to MySpace’s diluted popularity faced by new competitors.

When faced with such valuations I fall back on common sense and base the valuation on a cost-plus revenue model, ie. “what would it cost to provide the service, plus a modest profit margin”? All businesses and markets eventually commoditise down to that price point after all, and such commoditisation can happen very rapidly in the online world.

Viewed from that perspective I estimate that Twitter’s services are probably worth about 1.4 cents per user per year at present (estimated $1m/y running costs divided by 70m users). Lets be really generous and assume that they are able to generate 50% profits on that (first-mover advantage etc), so we get $500k/year profit, which at our mentalistic 20x valuation ration would give a valuation of a whopping $10m (£6.25). Hmm.