This blog’s sixth anniversary

diverI published my first blog post 6 years ago today. I didn’t even notice the fifth anniversary of this blog, so I am going to indulge myself today instead. 277 posts, 3459 (non-spam) comments and over 1.6 million page impressions and I am still here, posting sporadically as time and inspiration allows. Maybe I will still be writing in another 6 years. Maybe not.

Here are some of my favourite posts from the last 6 years, in no particular order:

It would be nice to break 2 million impressions. I calculate that this will take approximately another year at the current rate of progress.

I have lots of ideas for new posts. But if there is any subject you, dear reader, would particularly like me to write about – add a comment below. I don’t promise that I will write about it, but I will certainly consider it.

Ephemeral

stonemasonMy grandfather worked most of his life as a stonemason. Much of that time was spent restoring the ruin of a Bishop’s palace in Sherborne. His work is still visible long after his death. The work of the stonemasons who built the palace is still visible after more than 8 centuries. How long after you stop programming is any of your work going to last? If it is a desktop app, I doubt anyone will have a working computer with an OS that can run it in 20 years. If it is a web app, it dies the day the hosting bill no longer gets paid. What are you going to show your grandchildren – some screenshots or faded printouts?

Everything is ephemeral over a long enough timescale. In the long run, we are all dead, as John Maynard Keynes famously pointed out. But it is slightly depressing how short the timescale is for software. It lives fast, but it also dies fast. Our work is more like that of an ice sculptor than that of a stonemason. I guess all we can hope for is that the software we spend so much of our life crafting brings us some fulfilment and improves the lives of our customers during its brief lifetime.

The brutal truth about marketing your software product

badwaterWe tend to hear a lot about software industry success stories. But most of us mere mortals have to fail a few times before we learn enough to succeed. In this guest post William Echlin talks about the hard lessons he has learned about creating and selling software products.

Probably, like you, I started developing my own software application a few years back. I had this dream of working for myself and becoming financially independent. The money side was a nice goal to have but ultimately I was looking for the fulfilment of working for myself. Sound familiar? Well, if it does, you may have learnt many of the lessons I’ve learnt. I don’t mind admitting now that I got carried away. I got carried away with building a test management application to the extent that I forget about many of the key things you need in place to build a successful business.

After a few years work I’d created the leading open source test management application (a product called QaTraq that’s still available on Source Forge but a little dormant). It had cost me time, money and effort. I’d achieved some success with building and marketing a free product. Next stop taking it commercial. This is where it gets brutal.

About a year into leaving a full time job I’m taking the last £1,000 out of the joint bank account. I’m making some sales but it’s damn tough. A few months later and I’m in the supermarket £15,000 in debt wondering if my credit card is about to be rejected for the families weekly shop. You read about this sort of thing in biographies on successful entrepreneurs. These guys take it to the limit and then succeed and make millions. Sounds so glamorous. When your wife, 3 year old son and 1 year old daughter depend on that credit card being accepted believe me it’s NOT glamorous.

Building a business has always been about balancing design, development, sales, marketing, support, testing, etc. When you’re a one man band that’s not easy. You try to do everything. You’re bloody brilliant at building the product. The trouble is, once you want to make a living out of it, that “building” is almost the least important bit. After I’d spent 5 years building my product I stumbled upon one very useful piece of advice. It was a little late for me but maybe it’ll help you….

“Learn how to market and sell before you build your product. Learn these crafts by picking a product that’s already been built and act as a reseller”.

That’s worth reading again (it’s counter intuitive). What’s being said here is that if you can’t market and sell a product (ANY product) then the odds of succeeding with your own product are slim. If you can’t “market and sell” what on earth is the point in wasting all that time, effort and money building your own product? If you’re never going to be able to market it, and sell it, why build it?

So find a product in a slightly different sector and sign up as a reseller. Save yourself the time and effort of building a product and practice marketing and sales with someone else’s product first. Create a web site, develop an ad words campaign and start promoting with social media. Sell the product! If you can’t get the hang of this why bother building your own? If you can get the hang of building your own marketing machine it won’t be wasted effort. If you’re clever and pick the right product / sector you just need to switch the product on your site a year or so down the road. Once you’ve built the marketing and sales engine switch it to sell the product you’re building.

I’m not saying that this is the only way to go about it. I’m just saying that if you don’t have the determination to learn, understand and be successful with marketing and sales early on, then it’s unlikely you’ll succeed with your own product. So why waste time building it. It’s a tough lesson to learn. One I learnt the hard way.

And the specific lessons I learnt the hard way? Well I’d do these things first if I was ever to do this again:

1. Create at least one lead generation channel as an affiliate for another product. That lead generation channel will probably be a web site and as part of that you’ll need to master things like:

  • Google Adwords
  • Social media
  • Email marketing
  • Blogging
  • Link building

All these things take a lot of time. Do you have the determination to learn and execute on all of this?

2. Spend some time in a sales related role. Initially I was working in a full time job whilst building my own product in my spare time. The best thing I did was offer to help the sales team with product demos. I learnt lots from working closely with sales people (I didn’t like them very much, but that’s a different matter) and clients. If you can’t do product demos to clients, or you can’t talk to clients confidently then you don’t stand a chance of selling anything. People buy from people and a product demo is THE place to show case YOU (and the product)

3. Spend time learning about re-marketing. A lot of money goes into getting that initial lead. Don’t waste it! Understand Google’s re-marketing campaigns. These allow you to follow the people that came to your site and continue serving them banner ads on other sites. Understand email marketing once you’ve captured an email address. Yes I hate most of this when I’m on the receiving end. The reality is that it works though. That’s why companies do it (and why Google make so much money). I’ll tell you now that your business won’t survive if you don’t master some of these techniques. And if your business doesn’t survive then every ounce of effort you’ve put into building that application is wasted!

4. Spend time learning about cross selling. A significant amount of revenue can come from cross selling other products. When was the last time you went to a restaurant and they didn’t try to sell you a bread roll? When was the last time you flew somewhere and they didn’t try to sell you priority boarding? For you this might be in the guise of selling your leads to other companies that have complementary products. It might be providing different editions of your application. There are many other ways to add additional revenue streams to your prime product sale. These streams are absolutely critical to the success of your business.

5. Don’t try to become a sales person. You don’t have to be a sales man/woman to sell. Some of the best sales people I’ve worked with are those that just go out of their way to HELP the customer. They understand their niche inside out and have the gift, not to sell, but to HELP. People that are looking to buy something want help. They want an itch scratched or a problem solved. If you can help them with a solution then you’re most of the way towards making the sale. Forget all this rubbish about psychology and techniques to influence people. The best thing you can do is enter the mind set of helping! Go out of your way to help.

I don’t have all of this right by any stretch. I know one thing though. Products don’t sell themselves. And if you’re not prepared to start learning about sales and marketing you won’t sell your product.

It was all a bit ironic for me though. I spent years building my own test management product to help software testers. It even started out as the leading open source solution in it’s market for many years. I mastered SEO and created a great lead generation process (the oxygen of any business). I created a version which I put a price on and sold to companies. I even sold to a number of significant companies. But I just couldn’t do all of it. I couldn’t balance the design, development, testing, marketing, sales, support, etc. It’s brutally painful when this dawns on you.

In the end what I’d really mastered was lead generation. I ended up with a web site that attracted my target audience but failed to sell much. When you realise that, you realise that it’s the product. Nothing wrong with the marketing and sales. It’s the product. There were better products out there. Kind of tough to swallow but as soon as I did, I moved on. These leads, or rather people (because leads are actually real people), were looking for help. I just needed to provide them with the right product and services. So I started reselling other products and providing consultancy around those products on my test management website.

In the end I had one of the toughest bits right. If you get the lead generation right you’ve built a marketing foundation that you can build any type of business around. For me I just wished I figured the marketing piece out before I’d built my product. Now I just work on my marketing. Oh, and I help companies with their software testing and test management. For me at least, it’s much easier this way.

William Echlin has spent 20 years in testing, working on everything from air traffic control systems to anti-virus engines. He had a bad experience in his early childhood trying to effectively manage test cases with vi (he’s still a huge fan of vi but recognises that text files make a lousy repository for test cases). In an attempt to deal with these childhood demons he became a consultant on all things related to test management.

The 1% fallacy

Here is how to make a fortune writing software:

  1. Pick a large and established software market e.g. back-up, anti-virus or customer relationship management (CRM) software.
  2. Write a new product for that market.
  3. Get 1% of the market.
  4. Retire to your own island.

These markets are massive. The CRM market alone is estimated at around $18 billion per year. 1% of that is $180 million. How hard can it be to get one measly percent of a market? Ka-ching!

Except of course, it doesn’t work, unless you have massive amounts of funding or a brilliant idea that can completely disrupt the existing the market. Even then, you probably still need a fair amount of luck.

The competition in a large market, such as CRM software, is very tough. The top  companies have huge budgets and armies of developers and marketing people. Your chance of getting on the first few pages of Google results for a search term such as “CRM software” are as near to zero as makes no difference. And there are all sorts of network effects working in the favour of the established companies. For example, the biggest vendors will have an ecosystem of consultants, resellers, training courses, books, user forums and third party products that no new product can hope to match.

Then there are power laws which mean that you have to rank surprisingly high to get 1% of a market. The most famous power law is the Pareto 80/20 distribution. This is named after Italian economist Vilfredo Pareto, who observed that 80% of the land in Italy was owned by 20% of the population. Pareto distributions appear in all sorts of places. I have looked at various data for my own product and I have found the 80/20 distribution appears in my own data.  For example:

  • 77% of searches result from 20% of search phrases
  • 75% of sales come from 20% of email domains

If I could be bothered to crunch the numbers I expect I would find that  approximately 80% of support emails come from 20% of my customers and 80% of hits are on 20% of web pages. There is evidence that companies sizes are also distributed according to a Pareto type power law. Assuming a Pareto-type distribution, we can calculate what percentage of the market each company has according to their ranking using Zipf’s law :

Number of companies 1% rank
100 19
1,000 13
10,000 10
100,000 8

This table shows the rank you need in a market of given size to get 1% of the revenue of that market. For example, if there are a 1,000 companies in your market, you need to be ranked 13th to get 1% of the total sales.

How many companies are selling CRM solutions? I have no idea. Even in my little niche of seating plan software I have at least 10 direct competitors and well over 100 competitors with substantially overlapping functionality. I dread to think how many CRM products there are. At least a thousand I would have thought. What are your chances of coming from nothing to being the 13th biggest selling CRM solution? Also the conversion rates of customer visits to sales are typically around 1%. That means if you want to sell to 1% of a market and your main sales channel is your website, you need to get pretty much everyone in that market to at least visit your website. Good luck with that. Your best chance of getting a chunk of a big market is to create that market and grow with the market. But creating new markets is notoriously expensive and risky.

If you are a small software company, you have got a much better chance of getting a decent sized chunk of a small market, than 1% of a huge market. As a general rule of thumb, I would say pick a market for which you have got a decent chance of getting in the top ten Google results for important search terms (power laws again). You can even do this by going after a small segment of a big market. e.g. a CRM solution aimed at companies that trade on EBay. Or perhaps a CRM solution aimed at companies that trade on EBay in the Spanish-speaking market. You can always broaden your focus if you are successful in a small market.

Whatever you do, don’t stand in front of investors and pitch them the 1% fallacy. It makes you look an idiot. I should know, because I’ve done it.

How I increased sales 50% by adding extra price points

tinsHow much should you charge a customer for a product? From a pure economics point of view – as much as the customer is willing to pay. The airlines are masters of this. The people on a typical commercial flight pay a wide range of prices depending on factors such as: which class they are travelling, whether they are returning before the weekend and how far ahead they booked. The smug businessman in first class (who booked a week before and is returning the same day) might be paying more than 10 times as much for a seat as a someone in economy (who is going on holiday for 2 weeks and booked 6 months in advance). The businessman probably isn’t spending his own money, so he doesn’t care that much what the price is. Does the business traveller cost the airline 10 times as much? Of course not. The airline is simply maximizing its profits by charging more for the people who are prepared to pay more.

Supermarkets also use multiple price points by offering value, standard and gourmet versions of common products. The gourmet version has pictures of smiling farmers and tells you how it was lovingly hand-picked from a sun drenched hillside in an exotic country. The value version looks like UN emergency rations. The supermarket hopes the less price sensitive customers will buy the gourmet version, but they still want something they can sell to the more price sensitive customers. Is there much difference between the 3 products part from the packaging? Probably not.

When you start to look around you can see there are lots of different strategies businesses use to charge according to how much the customer is prepared to pay. Does a hardback book cost significantly more to produce than a softback book? No. But if you really want to read the book you will pay the extra for the hardback, rather than wait 6 months for the paperback. The gaming industry doesn’t even bother to change the product. Hardcore gaming fans will pay £40 for  a new blockbuster game. A year later you can get the same game (probably with bug fixes and add-ons) for £15. Two years after that it will be in the bargain bin for £5. Discount coupons are another common method you can use to charge price sensitive customers less.

I decided to try multiple price points for my table planner software. The graph below shows the 12 monthly cumulative sales[1] of my product for a year before and a year after moving from 1 to 3 price points. The red arrow points to the month I made the change. The revenue for the 12 months after the change were almost exactly 50% higher than the 12 months before.

multiple-price-points

Before September 2009 there was only 1 edition of PerfectTablePlan and it cost £19.95. Initially PerfectTablePlan was aimed at people planning their own wedding, bar mitzvah, Quinceañera etc. Typically they would only use the software once, so £19.95 was a sensible price. But as the product matured and improved it was increasingly being used by professional planners. It seemed crazy to be charging professional planners such a low price for software they might be using every week. So I decided to add additional price points at £49.95 and £199.95. The higher price points having additional features aimed at frequent and professional users.

I choose 3 price points because this seems a natural fit for the different types of people using my software (one-off users, frequent users on a budget and professional users spending someone else’s money). This turned out to be a big win for me. Not only did my average order value shoot up, suddenly I had more credibility with professional event planners, who might not have taken a £19.95 product seriously, no matter how good it was. Price is a signal of quality, after all.

Having 3 editions of the product with different feature sets also allows me to offer an increasingly sophisticated product to ‘power’ users without overwhelming more ‘casual’ users. This is a big bonus for all my customers and it reduces my support burden considerably.

There are various ways I could have set the price points. For example I could have set the price points based on the maximum number of guests at an event or on the duration of a licence. Charging according to the number of features seemed to fit best with my market and existing licensing.

I thought carefully about how to introduce the extra price points part way through the life of the product so as not to confuse or alienate existing users. I decided it would be too complicated to add the new price points at the same time as doing a major (paid) upgrade from v4 to v5. Instead I released the new editions at the same time as the v4.1 upgrade. I announced ahead of time that v4 would become v4 Home edition and that 2 new products were being released: v4 Advanced edition and v4 Professional edition. I was careful to ensure that I added plenty of new features and didn’t remove any existing features between v4.0 and v4.1 Home edition, so users who didn’t want to upgrade didn’t feel cheated. They were few complaints. I encouraged existing customers to pay the difference to upgrade edition and many did.

All 3 editions of the product are contained in a single executable and customers can switch between the editions dynamically at runtime. This was more work initially than using #defines to create 3 separate executables, but I think it was worth it as it allows the customer to easily trial or upgrade to a different edition without reinstalling or re-starting PerfectTablePlan.

A lot of software products have 2 or 3 editions, with the most expensive edition costing 1.5 or 2 times the cheapest edition. This seems far too narrow a range to me. I’m confident that a professional event planner can get at least 10 times the value from the product compared to someone planning their own (hopefully) once-in-a-lifetime wedding event. So I decided to go for a 10:1 difference between the cheapest and most expensive edition. If the airlines can do it, why can’t I? In retrospect I think this was a good call.

Having multiple price points is not without its downsides. It makes the sale more complex and it is an extra decisions for the customers to make. People are demotivated by having too many choices and I think having multiple price points has reduced my visit:sale conversion rate slightly. So don’t add too many price points. 3 is probably plenty in most cases (the supermarkets should know). But the slight drop in conversion rate has been made up many times over in a significant increase in average value per order. Also I should point out that the increase in sales wasn’t ‘free money’. I had to do a lot of work to add the extra features to sufficiently differentiate the 3 editions of the software, overhaul the licensing, tweak the website etc. But it was definitely worth the effort for the increase in sales. I think it also been beneficial to my customers as they now have a choice of which edition of the product best fits their budget and requirements.

[1] Each point is the total sales for that month and the previous 11 months.

DarkMarket

darkmarketThis book is an interesting and disturbing glimpse into the world of cyber-crime, particularly online credit card fraud (‘carding’). It also touches on related areas, such as cyber-warfare. It is written by a journalist, so don’t expect much in the way of technical details. But, if you can get past the tacky cover artwork and dubious sub-title (‘how hackers became the new Mafia’), it is a fascinating read.

The story mainly centres around the eponymous ‘DarkMarket’, a forum in which cyber-criminals conducted their murky business. For example buying and selling stolen credit card numbers. The story of DarkMarket is known in some detail, as it was infiltrated by various government agencies and some of its key players brought to trial.

There are lots of different characters mentioned in the book, many of whom have non-English names and online aliases. This makes the story quite hard to follow. Perhaps that is inevitable given that it is a story about deception and duplicity involving many people. Nevertheless, it still provides lots of interesting insights into this dark underbelly of the Net.

Online fraud is a cooperative effort. For example, some people specialize in stealing credit card numbers, others in selling credit card skimming devices and still others in employing armies of ‘mules’ to make withdrawals from ATMs (the riskiest part of the operation). But criminals are hardly likely to trust other criminals they have never met. Especially given that some criminals (‘rippers’) specialize in ripping-off other criminals. This is where forums such as DarkMarket come in. They act as a trusted third party, providing escrow and other services to cyber-criminals. The backgrounds and motivations of the cyber-criminals seems to vary considerably. Some start off as curious hackers withot any criminal intent, but turn to the ‘dark side’, often in small increments. Often such people seem to be motivated by status and reputation more than money. Others are simply in it for the easy money.

There are many ways in which your credit card details can be stolen. For example, you hand your card to a petrol station employee. The employee quickly swipes your card through a hidden credit card skimmer before swiping it through the legitimate device (they might pretend they have dropped something behind the counter to disguise this). A small camera hidden in the ceiling records you typing your pin. The criminal has a copy of your credit card and your pin number. These can now be sold on, perhaps through a forum such as DarkMarket, to other criminals who specialize in extracting the money. They will then clone your card and instruct their ‘mules’ to extract the money from an ATM and pay it into another account, keeping a percentage for their trouble. Some of the ‘work from home’ and ‘I made £2000 in a week’ ads you see in spam emails and attached to lampposts may be from cyber-criminals trying to recruit ‘mules’ for this purpose. Sometimes the criminals will withdraw small amounts over a long period as this is less likely to be noticed than one big withdrawal.

Cyber-crime is difficult to prosecute. It is hard to establish the real identity of the criminals and the they are often based in a different legal jurisdiction to the victim. The security services have infiltrated many cyber-criminal forums. The DarkMarket server was eventually being run by an under-cover FBI agent. However even security services from the same country (e.g. the FBI and Secret Service in the US) don’t seem to be able to play nicely together and end up investigating each others agents and informants and generally tripping over each other. The author believes that the Russian security services has infiltrated many of the Russian-speaking cyber-crime forums, but have no interest in shutting them down as long as they are careful never to steal from other Russians. The banks also aren’t keen to cooperate in investigations. You and I are ultimately paying for the fraud through our credit card fees. As long as the banks are making lots of money they don’t want to upset the apple cart by revealing the scale of the fraud. It might affect their bonuses.

So don’t expect cyber-crime to go away any time soon. But do stay away from dodgy websites, keep your credit card in sight at all times, cover the keypad with one hand while you type in your PIN with the other and check your statements!

DarkMarket on amazon.com (affiliate link)

DarkMarket on amazon.co.uk (affiliate link)

First impressions of the Retina 13″ Macbook Pro

macbook pro retinaMy table planner software runs on Windows and Mac. Previously I took a Windows laptop with me when I was out of the office or on holiday, so that I could provide technical support to my customers. But it isn’t (legally) possible to run Mac OS X on Windows, so this made it hard to give Mac customers the best possible technical support. For example I couldn’t send them Mac screenshots or replicate Mac-only bugs from a Windows laptop. However it is legally possible to run both Mac OS X and Windows on a Mac. So I decided to buy a 13″ Retina Macbook Pro for my new laptop, so that I could have access to both OSs when out of the office. I was also attracted by the screen and design of the new Retina Macbook Pro. I choose the 13″ model, rather than than the 15″ model, simply because it is more portable (e.g. more likely to fit in a hotel safe). I also paid to upgrade the SSD from 128GB to 256GB to have room for both Mac OS X and Windows setups. I have now had my Macbook Pro for a few weeks, so I thought I would share my initial impressions.

The good points:

  • The 2560 x 1600 Retina screen is gorgeous, both in terms of sharpness and colour. It makes the display on my old Toshiba Windows laptop look very tired.
  • With a 2.5 GHz Core i5 processor, 8 GB of RAM and an SSD it is very responsive.
  • The aluminium chassis is beautifully designed. Despite being only 19mm thick and 1.6Kgs, it feels very sturdy. The power brick is quite small and light as well.
  • The keyboard is nice. Automatically backlighting the keys in low light is a nice touch.
  • The power cable attaches magnetically. This means it is easy to attach. But, more importantly, accidentally kicking the cable just pops the cable out, rather than the whole machine crashing onto the floor.
  • Gestures work very nicely on the touch pad. For example you can drag two fingers up and down to scroll or tap with 2 fingers to simulate right click.
  • I have Windows and Mac OS X  stored on separate partitions. I can access Window either by booting into Windows or from inside Mac OS X using Parallels (cost approx $80). Being able to access Windows from Mac OS X without a reboot is very useful.
  • The speakers are surprisingly good.

The not so good points:

  • The Macbook Pro is expensive compared to Windows ultrabooks with similar specs.
  • There are a limited number of ports:  2 USB, 2 Thunderbolt, 1 headphone, 1 HDMI and 1 unidentified (SD memory card?).
  • There is no Ethernet port. So if there is no Wifi, you are out of luck.
  • There is no Kensington slot, so you can’t physically lock it down. This is especially annoying given the price.
  • As someone who mainly uses Windows I’m still struggling to get used to the Mac keyboard. For example there is a Backspace key, but no Delete key. Also there is no Home or End key. Worst of all the cursor sometimes  jumps to a different line while I am typing. I have no idea why.
  • Parallels and Windows don’t do a brilliant job of handling the very high resolution of the Retina display. Fonts are sometimes shown very small or very large in Windows. Some applications, such as Keepass v1, looks very fuzzy. Hopefully this will improve with new versions of Parallels and Windows as Retina-like displays become  more common.
  • Not all applications support gestures. For example Firefox on Mac OS X doesn’t support pinch zoom. But I expect this will improve as new versions of applications are released.
  • Not having a built-in DVD drive is a pain. I was able to install Windows 7 by ripping a DVD to a USB stick using Infrarecorder (free). But this didn’t work for Photoshop Elements 11, so I had to download it from the app store, despite having it on DVD.
  • I couldn’t get Windows 8 to work. I installed it into a Bootcamp partition, but when I booted into Windows I just got a mouse pointer and  blue bar down the side of the screen. Apparently Windows 8 is not yet supported by Bootcamp.
  • I am told the graphics performance isn’t great. But I’m not intending to use it for games, so that doesn’t bother me much.

Overall it is an impressive machine and I’m pretty happy with it so far. In particular it is great to able to support Windows and Mac OS X on a single machine. But it does have some annoyances and I wouldn’t have purchased it to just to run Windows. No doubt PC ultrabook manufacturers will have copied its more innovative features fairly soon.

A couple of tips:

  • If you are  in the UK, consider purchasing from John Lewis. They will price match other ‘bricks and mortar’ vendors that have stock. But (at the time I purchased) they were offering 2 years additional warranty. I got them to price match PC World’s £100 off deal. I got £100 refunded and 2 years extra warranty. The service at John Lewis is usually also very good.
  • You probably want to get some form of case or sleeve to protect your lovely and expensive new Macbook. Beware that many of the existing cases are designed for the old Macbook Pros. Consequently they might not be a good fit for the thinner Retina models. I bought this Cool Bananas case. It is a good fit and quite well made. I just wish it had a bit more padding. No doubt more cases and sleeves will become available to fit the new models.

VAT basics for software vendors

The dreaded VAT. Ugh. Value Added Tax (VAT) is the European equivalent of sales tax and it is a Royal Pain In The Arse. However, if you are running a business that makes sales in Europe you need to understand VAT. In particular it has important implications for your choice of payment processor, even if you are based outside the EU or below VAT registration thresholds. I have put together a few pointers here in the hope that it will help someone grappling with the complexities of VAT. But please note:

  • I am not an accountant. If you need proper advice, talk to a proper accountant.
  • The VAT rules are complex and may be interpreted differently by different people.
  • The rules may be different in different countries.
  • The rules change over time.

Only VAT registered businesses have to charge VAT. You have to register for VAT once your sales reach a certain threshold. At the time of writing,  UK-based businesses have to register for VAT if their EU sales exceed £77k in a 12 month period (technically it is UK sales, but the ‘place of supply’ for EU consumers is classified as the country of the seller). You can also choose to register for VAT before you reach the threshold. But it usually isn’t worth it, unless perhaps you think having a VAT number is essential for your credibility. Personally I waited until I couldn’t avoid it any longer.

Even if your business is not based in the EU, the EU still expect you to pay VAT on any sales inside the EU once you reach a threshold. This is controversial and it isn’t clear to me exactly what the EU can do to enforce this if you are based outside the EU. Talk to your accountant.

The VAT rate varies between countries. At the time of writing it is 20% in the UK and 19% in the Netherlands. It also varies over time, it used to be 17.5% in the UK.

The UK also has a simplified flat rate VAT scheme with a lower VAT rate. But you can’t claim back VAT on purchases. Worse still, it appears that you will effectively be paying VAT on sales outside the EU. So that doesn’t seem at all attractive.

The VAT rules are complex and depend on:

  • where you are based
  • where your customer is based
  • whether your customer is a business or a consumer
  • whether you are selling goods or services

Technically you do not have to charge VAT to an EU business, even if they aren’t registered for VAT. Apparently they are then responsible for “self-charging” the VAT. However the burden of proof is on you to show that the customer is a business. So most vendors require a VAT number as proof of business status.

There also seem to be disagreements over whether software is goods or services. What if you ship a CD?

Here is a simplified summary in pseudo-code of whether a seller needs to charge VAT on software as I understand it:

paysVAT()
{
    if ( seller registered for VAT)
        if ( customer in EU )
            if ( customer is a business )
                if ( customer in same country as you )
                    return TRUE;
                else
                    return FALSE;
            else
                return TRUE;
        else
            return FALSE;
    else
        return FALSE;
}

Except that people in Norway and Switzerland (which aren’t in the EU) pay VAT in some circumstances. Don’t ask me why. Also you don’t pay VAT on some items, e.g. postage. And outside the scope of VAT (O), not rated for VAT (N) and zero rated for VAT (Z) are all different VAT codes meaning no VAT is payable. As I said, it’s complicated. Not complicated and interesting like quantum mechanics or the love lives of celebrities. Just complicated.

The only upside of being registered for VAT is that you can claim back the VAT you pay on any purchases you have made (make sure you get a VAT receipt). Or, if you are buying from another EU country, you can tell them your VAT ID and they shouldn’t charge VAT (see above). So any equipment you buy in the EU is now 20% cheaper. This is small recompense for the giving 20% of your sales in the EU to the VAT man. Try not to think about that. Instead give yourself a pat on the back for having reached the VAT threshold. A lot of businesses never do.

Note that when you register for VAT you may be able to claim back the VAT of products purchased before you registered. When I registered I could claim back VAT paid on goods purchased within the last 3 years and services purchases within the last 6 months. So keep your VAT receipts.

Congratulations on making it this far. Here is the important bit. How you process payments has important implications for VAT. When someone pays you via a payment processor, such as PayPal, legally they are buying from you and the payment processor is just handling the payment on your behalf (like a bank cashing a cheque). So you are responsible for collecting what VAT is due and paying it to the appropriate government. This can be a major headache if you are selling hundreds or thousands of licences per month.

When you use a reseller, such as Avangate or Fastspring, legally you are selling your licence to the reseller and the reseller is then reselling it to the customer. The reseller is then responsible for deciding what VAT is due, collecting the VAT and doing the paperwork. They then pay you net of the VAT and their commission. Leaving you to sort out the VAT for their one payment to you per month.

Using a reseller is a big win if you are registered for VAT. I am registered for VAT and use Avangate as my payment processor. They do the heavy lifting in terms of calculating, collecting and paying the VAT on my sales. But if you aren’t registered for VAT be wary of using a VAT registered reseller – approximately 20% of your sales will be disappearing in VAT (which the VAT registered reseller has to charge) which you could be keeping if the customer bought from you direct. So if you aren’t registered for VAT, a reseller such as Avangate or Fastspring may not be the best solution for you. Check out e-junkie.

VAT admin is fairly straightforward. To keep the VAT man happy I have to file:

  • an EC sales list every month
  • a VAT return every quarter

My Quickbooks accounting package generates the numbers for these. It only takes a few minutes to file reports online once all the transactions and VAT codes are entered correctly into QuickBooks. The VAT man then debits (or credits) the appropriate amount from my business account each quarter. It is not too bad, as long as I don’t think about the wheelbarrow loads of cash Avangate keeps to pay the VAT man. Maybe they roll around naked in it on the last day of every quarter. I probably would.

When I first registered for VAT I tried adding the VAT onto my existing prices. But I found that sales dropped more than 20%. So I ended up keeping the gross price (including VAT) the same, whether the customer pays VAT or not (Avangate gives you this option). Whatever you do, make sure it is clear whether any prices you quote include VAT. EU consumer expect to be quoted prices inclusive of VAT and won’t appreciate it if you try to sneak on an extra 20% at the end of the purchase process. You may be legally required to quote the price including VAT in some countries.

A final note of warning. The VAT man has a lot of powers. I understand the UK VAT man can kick your door in and seize your equipment without needing even a warrant. He might not be impressed to find out that the computer you reclaimed the VAT for is an XBox. Do not mess with the VAT man.

If I have made any mistakes, missed anything out or if the rules are substantially different in your country, please add a comment.

** Please note that this article was written in 2012. It doesn’t cover changes since then, notably ‘VAT MOSS’. **

Thanks to Marcus Tettmar of Macro Scheduler automation software for checking this through and advising me on some of the finer points.

Buying a lean, mean, compiling machine

Nearly two years ago I wrote an article about speccing my ultimate development PC. Somehow there was always something more pressing to do. But I finally took delivery of my shiny new PC this week, partly spurred on by the fact that I wanted tried and trusted Windows 7 for the OS. Also my current development PC is getting increasingly crufty after 5 years of continual use.

I emailed my requirements to the top 3 custom PC companies in the UK as rated by PC Pro magazine : Chillblast, Cyberpower and CCL:

Hi,

I’m looking for a PC for developing software. Prime requirements in order of decreasing importance:

1. reliability
2. cpu + disk speed
3. quiet
4. value for money

Here is my wishlist of components:

-i5-3570K CPU
-an SSD (at least 128 GB) + 2 fast and reliable HDDs (7200 rpm, at least 1 TB each)
-ASUS, Gigabyte or EVGA motherboard supporting USB 3.0 and SATA/600.
-16 GB of fast RAM
-Windows 7 64 bit professional
-quiet is good, open to suggestions on sound insulation, fans and/or passive cooling
-AMD Radeon HD 6850 graphics card
-at least 2 USB ports on the front and 2 USB ports on the back (ideally more, ideally including USB 3.0)
-DVD drive
-Gigabit ethernet
-full size case
-kensington security slot, so I can lock it to the ground
-I don’t need a monitor, keyboard, mouse etc
-I don’t need WiFi
-It has to be *super reliable* – I want reliable SSD + HDDs, good quality motherboard, good quality branded power supply etc.
-target price, not more than 1,500 inc VAT, less is better obviously

Can you build something to meet this spec or get close? Please send me the spec and your price (including UK delivery).

Both Chillblast and Cyberpower sent me quotes for a system fairly close to what I wanted within 1 working day. I then spoke to their sales people and went online to tweak their suggested systems using their web based system ‘configurators’. In the end I chose Chillblast over Cyberpower due to:

  • higher rating from PC Pro readers
  • cheaper for a comparable system
  • better warranty
  • better online configurator (I found the number of choices on the Cyberpower online configurator a bit overwhelming)

However there really wasn’t a lot in it. CCL took nearly 2 whole working days to respond to my initial email,  so I discounted them as insufficiently responsive.

Following some suggestions made by the sales people I spoke to, this is the spec I ended up with:

  • Chillblast Fusion Longbow
  • Windows 7 Professional 64 bit
  • Onboard High Definition Audio
  • Corsair CX 750W 80 PLUS Bronze Certified PSU
  • Sony 24x DVD-RW Drive
  • Seagate Barrcuda 2TB 7200RPM Hard Disk
  • AMD Radeon HD 6850 1024MB Graphics Card
  • Intel 120GB 520 Series Solid State Drive
  • 16GB Corsair PC3-12800 1600MHz DDR3 Memory
  • Asus P8Z77-V LX Motherboard
  • Akasa Venom Voodoo Ultra Quiet CPU Cooler
  • Intel Core i5 3570K Processor 3.40 GHz (No Overclocking)
  • Fractal Design Define R3 Low Noise Case – Black Pearl – USB 3.0 Edition
  • Total price: £1089.80 + VAT (inc MSOffice Home Edition)

I take security fairly seriously. I have a motorbike style ground anchor in my office and I want my shiny new box physically locked to it. But I was told that almost no PC tower cases have a Kensington lock slot. This seems crazy to me. My current Dell tower has one and the cost of one tiny little extra slot in the chassis must be pennies. So I had to buy a lock adaptor kit. It’s not the most elegant solution, but it works fine.

I ordered the system on 02-Oct and it arrived on 22-Oct. Here are a couple of photos of the new system with the side panels off.

The PC took a few days longer than the originally advertised time to arrive. This wasn’t a big issue in my case. But I only found out it was going to be late when I emailed them after the expected completion date. It would have been a lot better if they had been more pro-active and emailed me first. Other than that I am fairly satisified with the service from Chillblast so far.

There are a few issues with the case, which aren’t really Chillblast’s fault. It looks rather lovely in its big, black, minimalist sort of way, a bit like an obelisk from ‘2001 a space Odyssey’. But the case scratches rather easily if you lay it down to change a component on my laminate floor. The side panels are also a bit fiddly to get on and off (my old Dell PC is better in this regard). Worst of all, it has rubber grommets (is that the right word?) that fall out into the case (and potentially into the fan or heat sink) if you even look at them funny. This means lying the case down to retrieve them, struggling with the side panels and more scratches. I have had to do this at least 4 times so far. It seems that they have made them of rubber that is far too soft for the job. Grrr.

How fast is the new PC? It certainly feels very snappy. I benchmarked it against my old Dell development PC (Dual Core 2.13 Ghz, 4GB RAM) building my event table planner software from scratch. This is 83k executable lines of C++ according to SourceMonitor:

Build time Old PC New PC
Debug build 6 minutes 56 seconds 1 minute 32 seconds
Release build 6 minutes 23 seconds 1 minute 28 seconds

So it is more than 4 times faster than the old PC at its key task – building software. Admittedly it isn’t a ‘fair’ comparison of the hardware. The older machine has a different version of Visual Studio, a different OS and probably some unnecessary services running in the background. But it is the best I can do in the circumstances and I doubt a ‘fair’ test would be much different. Despite the fact that I only went for a mid-range graphics card, the new PC can also handle playing Half-Life 2 on full 1920 x 1200 resolution without any noticeable issues. Hopefully the faster build times will give a significant boost to my productivity (as long as I don’t play too much Half-Life 2).

The new PC is also eerily quiet. I would guess more than 4 times quieter than my old PC. Even when it is doing a build, all you can hear is the faint whir of a fan.

Only time will tell how reliable it is.

The world’s fastest Rubik cube solver is made from Lego!

CubeStormer II is the fastest Rubik cube solving robot in the world. It set a Guinness World Record of 5.270s for the fastest robot solving of a Rubik’s Cube in November 2011. I saw it in action on Saturday at the Swindon Lego show. Click the image below to watch a video I took.

(there is also a .mov version on screencast.com)

The project was commissioned by ARM Holdings and designed and built by Mike Dobson (who built the Lego robot) and David Gilday (who wrote the software). A custom Android app on a Samsung mobile phone images the cube and instructs the Lego robot what turns to make. The robot is made from 4 Lego Mindstorm NXT kits.

I got to speak briefly with David Gilday and he told me that the software is optimized for the robot’s capabilities, so it computes the quickest sequence for the robot, rather than the minimum number of moves. The software uses pre-computed look up tables of moves for speed. Apparently the limiting factor on the speed is the power of the motors. CubeStormer II can manage about 5 moves per second, whereas the best humans can manage 9 moves per second.

It didn’t work every time. But it is an impressive achievement. Especially considering the software was written by a hardware engineer! ;0)

(there is also a .mov version on screencast.com)

More details over at Wired.