The whole thing has taken only 2 weeks from the initial thought of "I should give this a shot," to getting some usable data. Here's what I've learnt.
- I'm better at this web malarky than I thought.
- It's nice to be reminded I can make things. Often at work it feels like I'm making little progress, and I've got a notebook full of never-implemented or 25% done personal projects. Real artists ship, and it feels good to kick something new out the door.
- Amazon's catalogue is shit when you get away from media/electronics.
- I wanted to link to clothes, shoes, jewellry, some accessories. Do they have them? Do they fuck. I will link out to other stores that don't pay any referral fees and hope that people also use the Amazon links.
- eBay's business model is fundamentally retarded.
- eBay's got an affiliate scheme but it's a confusing mess, not least because eBay is stuck on the auction format. A giant part of eBay's business is from small business sellers offering products for a "buy it now" price, with essentially infinite stock. But you can't generate a link to a specific product because the auction fetish listings expire and have stock limits on each lot. I am ignoring eBay because of this.
- Omniture SiteCatalyst makes you take good analytics for granted.
- At work we use Omniture SiteCatalyst to track user behaviour, and I'd taken that quality of insight for granted. There's two parts to this:
- Amazon's affiliate reporting is crap. When you generate an affiliate link back to Amazon you include a tracking code. When you click through, Amazon can tell exactly what was clicked. Do they show you this data? Do they fuck. This data would be super-useful because you know what products are tempting and what aren't, letting you rewrite the copy or replace the product with something else.
- Google Analytics is alright, but not superb. It doesn't do any click-tracking so I can't tell how users are leaving my site. Nor does it track AJAX requests, so it's not much good for anything other than a visitor counter. This is a drawback of taking the single-HTML-page approach.
- Amazon's affiliates scheme is per-locale, which is ridiculous.
- What? If I want to sell something to a user in the States I have to have another account with its own code? And I have to rewrite all my links? And do this all again for amazon.de, amazon.jp, and so on? Amazon could easily have handled this all for me. I should be able to include my referer ID in a link and let you deal with all that. I'll even let you give me a crap exchange rate, or just throw me a buck every time a US customer buys something.
- Amazon's sign-your-REST-requests requirement is a steaming pile of turds.
- To make a request to Amazon's web services (which you have to do if you want to display prices) you must sign your request. You do this by taking the parameters, running them through a hash function, and appending the hash to the request. It stops other people pretending to be you and it gives Amazon some more control over their web services. But they could at least have the decency to document the fucking thing. There's a bunch of getting-started examples in the manual that don't work because they don't include the signature, and their "Here's how to sign your requests" is outright wrong (it excludes the mandatory timestamp, for example). Nor do they provide example implementations for you to use and only the barest of worked examples. I tried writing my own implementation and it didn't work, so I googled for one and used that instead. The API is a bit shit too – it'll tell you the lowest price for an item but not the lowest price including postage, or the official Amazon price.
- Selling stuff is better than advertising.
- Whenever the question of "But how does it make money?" comes up online, everyone bellows "Advertising!". What bollocks. I've never trusted advertising to bring in anything – you're essentially selling the opportunity for someone else to sell something – and much prefer physical products. Both deal in small change but I'm much happier offering pysical products instead of trying to write linkbait articles and hoping someone's dumb enough to click a banner ad.
- Facebook ads are useless, but really easy.
- I've not found anywhere else that does ads this well. Most advertising networks don't let you buy £10 of ads and see what happens, but Facebook do. They've got a neat ad builder and control panel that shows statistics and reports about who's clicking your ads. They've got great targeting too. The trouble is that nobody clicks ads on Facebook and your fellow advertisers are scum, but it's a cheap way to get some traffic.
- Writing is hard, especially when you don't believe something.
- I find writing hard enough anyway, but trying to write something engaging about a mediocre product is hard. This whole project is soulless but I don't want to lie either, so I've put effort into squeezing out something truthful and exciting. Writing took way more time than building the site.
- Pick your target audience carefully.
- If the people you're targetting don't have any money, you're fucked. Mine are broke teenagers so I shot myself in the foot there. Similarly, if there's no online ecosystem around your niche you're out of luck. I can't believe there isn't for mine, but that's how it looks. There are no good blogs, no online celebrities and associate hangers-on, no real forums. Starting one would be effort I'm not willing to expend.
How am I going to rescue this trainwreck? I'm probably not, but that's OK. It was an experiment, I've exercised my web developer & writerly chops, learnt some things, and it's not cost me anything except £8 and some evenings I'd have pissed away anyway. I've picked up a few tips from this guy (who is putting way more effort into this whole thing than I am). I'm going to make the following changes and then leave it alone:
- AJAX in prices from Amazon. Amazon won't let you hard-code prices into your page – you have to retrieve them at least once every 24 hours. I've written the code to retrieve prices and to display them in the page, and just have to cache the results. Buy links with prices displayed get something like 7 times more clickthroughs than those that don't, so this should make a real difference.
- Link out to non-Amazon stores for certain categories. Even if I can't make money from people clicking through it's all content grist for the Google mill.
- Create some new categories. I started with a small selection of categories. More content looks better and gives people more chances to buy, but releasing early and iterating is better than crafting the perfect emporium before publishing.
- Guest-write some blog posts for people in exchange for links back. There are a few blogs out there – writing stuff for them in exchange for links helps 'spread the word' and helps people find you. So does participating in forums etc, but I haven't found the energy for that yet.