Failure: The Key to Online Retail Success

Vintage Sports Equipment

In this post I look at how a friend of mine radically changed his approach to online selling, and introduce principles of an ecommerce philosophy that I call Lean Commerce.

Tom’s Store

Tom owns a sports equipment store.  It’s a small shop, but gets a lot of repeat custom. Most of his customers are buying kit for their school children, and that’s often left to the last minute. High-street stores still have something over internet shopping for rush purchases: fitting rooms and no wait for delivery.

Tom has also been selling online — mainly eBay — for around four years. He figures that if he already has the stock in-store he might as well put it online too, and it’s handy for selling off old product lines.

Tom and I were talking in the pub when the subject turned to business — his business. His online sales were good — a healthy addition to the store sales — but growth was flat over the last year. In fact, he’d cut shipping prices due to competition and eBay’s search changes. Still, he felt online selling was worth it.

The Question

Tom had an idea he wanted to run past me: he was considering adding a whole new product range to the online store: golf accessories. They don’t play golf in schools, so he wouldn’t offer the new range in his high street shop — this would be his first product range to only be available online.

Tom wanted to know if selling golf accessories was a good idea, or a reckless gamble that would leave him with a pile of stock to sell off at a loss. Later, I realized he was hoping for a pep talk — great, go for it! — but instead I asked a lot of questions. I was trying to understand his strategy, but it was turning into an argument. Before long Tom looked ready to blow up, so I excused myself to get more beers from the bar.

Filling beerFilling beer

At the bar I had time to think, and figured out why we were arguing: our ways of thinking were miles apart! Tom was a numbers guy. Before the shop he’d been an accountant, but he left the rat race and went into retail after a lot of analysis. Now he had been crunching the numbers in Excel again, and figured out that golf kit would be a big seller.

For a long time I’ve worked in software development: a profession notorious for delays and broken promises. The software world has improved a huge amount in recent years, by adopting (and making its own) ideas from manufacturing and science. The result is a huge shift in how we create software — a whole new philosophy. I’ve learned not to act blindly because it says what should happen in somebody’s spreadsheet or project plan. I need to see a clear connection between the plan and the end goal, a connection backed up by real-world evidence.

Lean Commerce Principles

The beers were on the bar now. I paid for them and made a decision: I would stop the interrogation, and explain the ideas I’ve been using in software development. Perhaps the same approach would apply to online retail? That discussion went a lot better than the previous conversation!

By the time we’d finished our drinks the mood had picked up, and Tom had filled a page with notes. Still, I had an uneasy feeling that he was holding something back. It was getting late and he was desperate to go, so we left — with a plan to catch up again in a couple of weeks.

So here are the main points of our conversation, boiled down into three key principles. I’ve taken inspiration from a few areas, like Lean Startup and Agile software development. What I hope to do here is put them into context for small-to-medium sized online retailers like Tom — and perhaps you!

Principle #1: Guess and Test

I didn’t know if Tom’s idea would work out. Tom didn’t know. In fact, no-one could possibly know if he would be successful selling golf accessories. Why not? Because all the education, experience, research and analysis in the world can’t tell you what will happen when you do something for real. Without actively doing something, it’s just guesswork — educated and informed perhaps, but guesswork all the same.

If I’m against guesswork why is this principle called “guess and test”? Well, to run a test you need to start with a hypothesis – that’s the guess. Guessing doesn’t mean you shouldn’t do any research — you should — but as a way of generating ideas to test, not to convince yourself that your idea is right. Tom’s guess (although he didn’t think of it that way) was that if he offered golf accessories through his online store they would sell just as well as his other products.

Once you have made a guess, don’t stress yourself out trying to figure out if it’s right, instead move onto testing as soon as possible. Why? Because pondering and planning can’t tell you what will happen when you do something you’ve never done before. You’ll waste time and draw the wrong conclusions, when you could be learning from low-cost, fast, real-world experiments.

Guess and test flow diagram

What Forward Did

Internet marketing agency Forward moved into online retail by looking for popular search phrases with few ads around them — indicating there might be high demand (lots of searches) but low competition (few ads). This resulted in a shortlist of “guesses” at products that might sell. They could have dived in and bought stock, or commissioned market research, but instead they ran a series of tests.

Why didn’t they make the most of these opportunities straight away? Because lots of searches don’t necessarily mean people want to buy. Perhaps the people searching want information, images or opinions and have no intention of making a purchase online? Be careful to identify the assumptions you are making — like taking it for granted that searching will turn into buying.

So Forward created a test to answer a key question — did people really want to buy these products online? They created dummy websites with a few static product pages, then generated traffic using paid search advertising. Key measurements were the number of visits from potential customers, and the percentage who tried to add a product to their basket.

U Show Parrot Cage. Y U No Let Me Buy?!

Here’s the catch — there wasn’t actually a shopping basket in place! The point of the test was to get data, not make sales. It might sound crazy to turn down buyers, but Forward wasn’t running just one test — they ran many tests at once. They hadn’t found suppliers or created a working shopping cart — the extra effort of that would mean they could run fewer tests and might miss the best opportunities.

That test resulted in a shortlist of twelve products with high add-to-basket rates and affordable advertising. The surprising focus was pet enclosures, including bird cages, fish tanks and kennels. With the data in hand, Forward created an end-to-end business experiment: they set up working stores using Shopify, and fulfilled orders using drop-shipping suppliers. Twelve months after launch, sales had grown from zero to £200,000 per month. With buyer demand proven, the shortcuts they had taken to allow quick testing could be reworked to increase their profit margin.

Principle #2: Fail Fast

It’s not all that controversial to say you should test your ideas before diving in. But this principle is a bit more challenging! Yes, you should test, but you should also expect to fail, and even try to fail. And when you do fail, you should be quick about it.

This is a hard lesson to take; failure is a frightening thing. The popular story of the entrepreneur is one who battles against the odds, always believing that their idea is a winner, no matter what everybody else says. Finally, after many setbacks, they succeed and it’s proven they were right all along.

In that story, the hero entrepreneur succeeds because they stick doggedly to what they think is right. The moral is that determination, belief and hard work will lead to success. Many entrepreneurs internalize and live by that story, but it’s a myth. Most businesses fail, and they don’t fail because they don’t believe in themselves enough, or naysayers lead them off course. They fail because — in some way — their business is fundamentally flawed.

So, in a nutshell, you should expect and try to fail because your idea too is probably fundamentally flawed. To clarify: I don’t mean you should intentionally screw-up, but you should look for the assumptions in your plan, test them, then adjust course based on that feedback.

Fail You Will. Fail Fast, So You Should.

Amazon: Experts at Failing

At Amazon, when a manager has an idea the first thing they do is write an internal press release announcing the finished product. Let’s be clear here: the internal memo is the first thing they do — with no actual product or even a prototype.

Instead they describe a problem they have seen, explain how their idea solves it, and set out the benefits for buyers or users. Then their colleagues pick holes in it. More often than not, they rip it up and try again — either with a different angle on the same idea, or something completely different. To submit your ideas to scrutiny that early takes courage. You have to embrace failure, appreciate that it’s probably coming and not take criticism personally. If you try to protect yourself by extensive researching and refining, you will probably just fail slow instead of fast — and waste a huge amount of time.

Internal press releases are only the first stage in refining ideas at Amazon. They are big believers in using metrics — specific measurements of users’ behavior — to drive their decisions. If an idea progresses to implementation, it is not refined exhaustively before it is tested once again. Instead, a change or new feature can be put onto the live site for a small proportion of users and the effect tracked.

Depending on the change in metrics, the idea can be released more widely, reworked or abandoned. Changing or scrapping an idea is a good thing — it means we have learned something important and instead of lamenting lost time can move on quickly to something else.

Principle #3: Learning Before Profit

Business is all about profit, right? Right.

But there are times when it is healthy for a business to earn less than they possibly could – when it allows them learn a lesson quickly. So, instead of planning to maximise profit from the start, plan to test assumptions and maximise learning.

How Zappos Started

Shoe retailer Zappos was sold to Amazon for $1.2 billion in 2009. It had humble beginnings. Founder Nick Swinmurn got the idea for an online shoe store in 1999, when he discovered that 5% of Americans bought their shoes from mail order catalogs. That’s a good indication for people buying shoes online, but investors were not convinced.


So to test his idea Swinmurn created a basic website using photos taken in local shoe stores, and when he made sales bought from the same stores at the full retail price. Clearly, that’s not a way to make money, but it provided the missing evidence that people really will buy shoes online. And without that loss-making experiment, Zappos may have never got off the ground.

As another example, you read how Forward used paid advertising as a quick way to learn about products with high demand but little competition. Yes, they had to pay for the ads, but the cost was much lower than setting up a real, efficient ecommerce business which would not have succeeded in many of its product categories.

Tom’s New Store

Tom and I had agreed to catch up again in two weeks, but real life got in the way, and two months passed before I saw him again.

OMG Did I Really Buy That?

We sat down with our drinks, and Tom started the conversation by telling me he had a confession to make. He asked if I remembered his sudden rush to leave, and now he explained the urgency: he had already placed a large order for golf stock from China. So he really didn’t want my opinion on the idea, he wanted reassurance — it’s no wonder my questions wound him up!

Realizing he was putting a lot of money behind a fairly wild guess, he had rushed back home and cancelled the order. But Tom had also prepared about thirty eBay listings, ready to go as soon as the stock was landed. That time was a write-off too.

Or was it? Tom didn’t think so, because he ran the listings anyway. If he made a sale he would buy from another seller — at a small loss — and get it shipped direct to his buyer. A kind of ad-hoc dropshipping. The buyer might be a little confused when the parcel arrives with a different seller’s packing slip, but in truth most eBay buyers pay little attention to who they’re buying from.

Now, most retailers would rather eat their own heads than give money to a competitor while suffering a loss themselves, but Tom had risen above that. And importantly, he learned something: golf kit was a weak product line for him, because he hardly made any sales.

What he did next was really smart. He realized that guessing was a good thing, but he needed to make his guesses more educated. He put some time into researching the eBay market, looking for opportunities in other sports equipment. He found a few maybes, but didn’t fall into the trap of continuing the research in search of the “perfect” product – research alone can’t get you that far.

His candidate products were in the school sports market he was familiar with, but in niches like lacrosse and fencing that relatively few schools played. He ran some listings — again without buying stock — and got a reasonable number of orders. But when he caught the start of the playing season for some of those sports, orders got a big boost. This was now familiar territory, and he felt comfortable enough with what the data was telling him to make some sensibly-timed (and sized) investments in stock.

Traditional vs Lean Commerce

So Tom stayed in the school sports market after all, and was expanding his range into niche sports. Not only that, he had a taste for the power of data and testing, and was full of testable ideas for express shipping upgrades and timed promotions that could appeal to last-minute buyers. It turned out that it’s not just football boots that get left to the last minute.

That’s Tom’s story. I’d love to hear yours in the comments below.

* * * * *

Note: Tom didn’t want me to reveal his business name, so I’ve left that out and changed some other details so he can’t be easily identified. He thinks he might be breaking eBay’s product availability policy for listing items he doesn’t have in stock!

6 comments on “Failure: The Key to Online Retail Success

  1. Nice article Andy and I like the comparison to agile development (formerly called RAD ;-)). While I agree testing is good, even if you lose money initially, it is not always the best strategy. The strategy proposed here totally ignores long tail business. Long tail strategies can reap big rewards for eCommerce because a physical store cannot carry near as large an inventory as an eCommerce specialist can. I have certain categories that we carry a large variety of items. While a few consistently sell out, I have others that we may sell one or two a month. These one or two a month items make the difference between profitability and just covering overhead. You do have to watch your cost of holding inventory as well as inventory financing – there is no free lunch, but with all the consumer choices today you cannot ignore the long tail business.

    1. Thanks Jack, that’s a good point! My philosophy is that testing in general is good, but what that means for a particular business will vary. I agree that for sellers with a very large product catalog (e.g. many media sellers) testing each product line doesn’t make sense. But it should be possible to test other things such as different pricing and loyalty strategies etc.

      By the way, I worked with both RAD and Agile and the big difference I saw in practice was that RAD often lead to poor quality applications (just getting it to work was good enough). With Agile there’s a big focus on quality, craftmanship, maintainability etc. There are certainly good and bad examples of both though!

  2. The danger of this approach is a long delay in getting the first orders out for new lines of product from new suppliers, I would always advise that the store owner have tested the supply chain at least 1 time so as not to disappoint customers.

    1. Actually, I’m not suggesting that sellers make an order from a supplier when they get their first orders but directly from another retailer so there shouldn’t be any delay for customers. That way they get to test several product lines easily, quickly and risk free.

      Lines they want to continue can be sourced later, and sales either put on hold or continued at a loss in the mean time.

  3. Maybe you can try drop shipping for the item you do not sure whether it is going to sell or not to sell. For the popular items, I always recommend stock up your inventory.

Leave a Reply

Your email address will not be published. Required fields are marked *