The Holistic Web

In a traditional company, you’ll have product or service to sell. There’ll be people making the product, marketing people promoting and creating demand, and sales people completing the sale.

In ecommerce, it shouldn’t be that different. But it is. This is what I typically see:

The person owning the business makes or sources the products. They get someone to make the website. Someone does the marketing - SEO, SEM, PR, etc. There’ll be a designer to make it all look pretty. There’s one thing missing though. Can you see it? Who is doing the selling? Well that’s easy you say, it’s the website itself. That’s the beauty of ecommerce, it replaces the need for sales people, it allows you to scale up. Ok, fine, I buy that. So who is in charge of making sure the website makes as many sales as possible?

That’s about the point where I hear silence and lots of ummms and ahhhs. It’s crazy - arguably the single most important thing, making sure the website actually sells, and it’s neglected. It’s a bit like getting an expensive car, marketing it and then asking an 8 year old to sell it. Sure, they’ll sell a few, but a seasoned, experienced sales person will sell many times more.
Don’t get sales blindness - sell your products!

I’ve been doing a lot of reading recently (as always), and one thing that has come up again and again and again is: you can’t trust what people say. Let me give some examples of stuff I’ve been reading:

• In a double blind clinical trial for a sleeping tablet, 90% of people reported it effective, versus 60% who reported the placebo effective. I read a lot of clinical trials (yes, strange, I know), and it’s quite normal to find that ratio of effectiveness vs placebo.
• In “Blink” (great book), a case where people were tasting Sprite and asked how much lime was in there. The answer would be directly related to the colour of the can - a greener can would invariably give a more limey taste, with no change to the formula at all.
• Adding a phone number to a web page increased sales dramatically - without generating a single phone call. A phone number generates a perception of trust. I didn’t do follow up research but I’m sure none of the new customers would attribute their signing up to the appearance of a phone number, but statistically, there’s a fair chance they did.

I’ve bumped into many, many more, but some are a bit too complex to be contained in a blog friendly nugget. The thing in common is that people say one thing - and truly believe it - but the real reason is not at all the same. They say it has more lemon, but it doesn’t.

Our minds are fooled very easily. As horrible as it might seem, the conclusion to reach is that many of our decisions are determined by our subconscious based on reasons we don’t know.We try and rationalise things, and so we willingly offer an explanation is prompted, but often that explanation is simple a rationalisation of what has already happened subconsciously, and isn’t the real reason.
Yet, so many people and companies still believe in things like surveys and focus groups (where as well as the above issues you also get group think). I’m not saying those things are useless, they have their place, but unless you understand these broader issues and make sure you are asking the right questions in the right way, you’ll get a result that seems right, but probably isn’t.

It increasingly seems that to be successful in the online realm, you need to be a psychologist as much as a marketer. Fortunately, the process we use is fundamentally based around people’s psychology - I’m growing increasingly aware of how right that approach is.

No matter how many times I do them, I am always amazed at how much of a difference split testing makes.

I had a difference of opinion with a client recently about how a landing page for a fairly large budget campaign should be constructed. We went with a compromise solution, but I still wasn’t happy. I offered to do a test on the compromise version versus my ideal version. To his credit, he willingly agreed. The test isn’t quite finished yet, but so far it’s showing a 113% performance improvement - more than double!
If you aren’t split testing, why not? Of course, you can let me know if you want help with it.

Edit: test is now complete, at 158% improvement! Wow.