A former community discussing the business of software, from the smallest shareware operation to Microsoft. A part of Joel on Software.
We're closed, folks!
Doug Nebeker ("Doug")
Following a link http://www.business2community.com/online-marketing/turn-website-visitors-into-customers-via-conversion-optimization-0329212 from the article that was linked in a previous thread http://discuss.joelonsoftware.com/default.asp?biz.5.850888.12
Somewhere in that snakepit of social network buttons and sign up for newsletter boxes is content that claims to be good advice on designing a landing page. It took several minutes, but I managed to ignore all the tacked on interruptions and made it through the entire article. I even started to read the whitepaper produced by the same company that produced this article on online marketing concerning the wildly successful results this same company provided to one of its clients. I quit reading when I couldn't find any credible data to back up their claim of $100,000 savings for their client, although it might be in there somewhere, if you can be bothered to look hard enough.
If you really want to see what works in online marketing, go to the sites that sell the most online. Like Amazon and EBay. No big Call To Action button tastefully isolated on a pastel background for them. There is a single picture of something eye-catching, located below at least two rows of search boxes and links in small print. As you move away from that picture in any direction there are wall to wall hotspots, some are images and some are text, but the idea is put everything you might be looking for where you can easily find it. In the exceptional case when you don't see a link to exactly what you are looking for, you can keep scrolling down the page until you do find it.
No elevator pitch either. After all, visitors make a conscious decision to go a particular website, so they already know what they want. Your landing page has to show them where to go to get what they want, not convince them to buy.
I get asked to evaluate a lot of marketing services that get pitched to my employer, and you can probably guess what I think of most of them. When it comes to marketing, there is nothing wrong with imitating successful companies. And beware of experts who are more successful selling advice than real products that real people use every day.
I hear what you say. But a microISV (for example) is a very different business to Amazon or EBay. You would probably be better off looking at other successful microISV sites for inspiration.
Friday, February 08, 2013
AB testing is relatively cheap - we don't need to take anyone's word for anything in page design - we can test it.
Friday, February 08, 2013
There's so much wrong with the notion that you should model yourself after the current Amazon, Google or Craigslist that I can't even begin. But I'll try.
Every one of these sites has scale - pure massiveness - behind them. Amazon has hundreds of thousands of products, CL has millions of ads, and Google has the entire internet.
Every one of these sites is a franchise that has developed over a decade plus. People come to *them* for specific needs: Amazon to buy just about anything, Google to search, CL to barter (or maybe carjack :) .) Every single one is a household word.
An mISV or a small business has 1, 2, or a couple of dozen things that it can be selling at any one time. It is not a household word. It can't show rows of products and services and choices because *you just don't have them.*
Maybe you could put some real content front and center but then would the person get around to buying or are they just coming to your site because "oooh, shiny, keywords"?
The mega-site can't be used as a model for the small biz site. Well, maybe it can, in a way. Amazon was small once, in 1996. Google was once just search. Both of them were tight and focused in the beginning. Refer to the Wayback Machine for details...
Today some people come to Amazon for entertainment just to read the reviews, and then wind up buying something. With millions of hits, Amazon has a lot of traffic to play with, and people would now come there even if the design completely sucked (it's pretty complicated now with business like the "choose a category to sort" logic and the like.)
A small potatoes vendor is much better off with a big "buy now" button and simple text. The main problem of a lot of underperforming web sites is too much, too long content.
All these sites had dozens, even hundreds of competitors.
They killed all their competitors. That then ENABLED them to become big.
Ignoring that is dumb.
How did they kill their competitors?
Their sites all do the right thing that people want. They don't screw around with the customer.
Remember all those search engines that filled the home page with ads, news, and distractions? They made fun of Google. How can Google make money with no ads on their main page? Where is the stickiness to comp back? Where is the push content?
I picked those examples because their businesses are their web sites. Most books, antiques and cameras are not sold online. Nor do Amazon and EBay have online monopolies, but they dominate their markets like Heinz dominates the catsup market.
Too much of the accepted online marketing wisdom doesn't stand up to even a superficial analysis. For all the effort put into website design by mISV's, it looks like most of them are headed in the wrong direction. It's similar to the long copy debate. As long as everyone is ignoring facts, you are no worse off than your competitor, but don't forget that in most situations, your competition is a different method for performing a particular function, not other software using the same method with a different UI.
And don't get me started on A/B testing. Doesn't anyone study science? The only variable you control with a website is the content, and you have virtually no reliable information about the hundreds of other variables that affect who visits your website and what they do when they get there.
You can learn good things from sites like Amazon, such as using tabbed navigation.
However there are HUGE differences between selling physical products and being famous for doing so, and trying to convince cold strangers to try your software.
Think about the word 'marketing'?
It's about identifying, finding, getting the attention of and communicating your product's benefits and advantages to your specific *market*. That's an entirely different game compared to Amazon or similar, plus you have the additional challenges of NOT being so well-known or trusted etc.
I don't disagree with the basic premise, that many marketing companies don't practice what they preach. That doesn't alter the fact that large, clear call-to-action buttons and engaging text are proven conversion boosters for MiSVs and similar online businesses.
And yeah, split testing works.
Sunday, February 10, 2013
Howard; the point of A/B testing is that the other variables besides the content differences factor out over a large audience. That's why we demand statistical significance from results.
Many people have used this technique to gain real, measurable improvements in their profits - by comparison the ideas your promoting sound like they're based on faith & opinion. I would never advise a business to change their marketing strategy without imperially proving the new strategy is superior if it was possible to get some numbers; split testing is just one way to achieve that.
Sunday, February 10, 2013
>And don't get me started on A/B testing. Doesn't anyone study science? The only variable you control with a website is the content, and you have virtually no reliable information about the hundreds of other variables that affect who visits your website and what they do when they get there.
A/B tests are widely used in subjects such as psychology, pharmaceuticals and behaviour precisely because there are lots of variables you can't control.
Sunday, February 10, 2013
You don't test pharmaceuticals without a full medical history, reporting of diet during the test and so on. I'll use another example, when Proctor & Gamble tests a new toothpaste they don't pull the old Crest off the shelves, because they know they will lose existing customers. For a new product, they will do focus group testing, then launch in test locations where they know the income levels, age, sex, etc. of shoppers, and if they think they have a winner, then promote it to their regular distribution channels. What software vendors call A/B testing is just a variation of the old Pepsi Challenge except you are competing with yourself.
If your market is larger than a single person, you have diversity. If you want to test different advertising for the same product, run both versions of your ads at the same time. Or run them at different times so your audience doesn't quit paying attention, but don't expect to be able to say B is better than A because you sold more while B was running. If you randomly present visitors with different versions, you dilute the effectiveness of your advertising because if you present B to someone who is more favourable to A, you will never get them back.
There isn't a constant stream of homogeneous prospects that you can dip your bucket in whenever and however you feel like, and expect to get the same results time after time. Which is another problem with A/B testing, you can't repeat your results, because of the all the factors you can't control, and you can't model the results because you can't measure those factors either. It's witch doctor marketing.
Andy, Dr. Goldacre may have some valid viewpoints on the suitability of randomized control trials for evaluating pharmaceuticals and medical procedures, although the little I know about this topic suggests that his ideas are far from universally accepted (to put it mildly). Being a well-known commentator does not make him right.
If most people reacted to specific drug therapies in the same way, regardless of how many times those therapies are applied to the same people, or even if their reactions changed over time in a consistent manner, you can make the argument in favour of using RCT's for testing drugs. Especially in the absence of better tests, and the reasonable expectation of doing more good than harm. If it turns out that a specific therapy is no longer effective, or is not really as effective as the trials indicated, we quit using that therapy.
I am extremely skeptical about the use of RCT's for any arena of public policy, even public health policy. It is much easier to argue in favour of RCTs when they are uncommon, because almost all policy failures did not rely on a RCT. Those failures do not prove the efficacy of randomized control trials in setting public policy however.
The population of policies (successful or unsuccessful) implemented solely on the basis of RCT's is way too small to draw valid conclusions from. In the history of mankind there has never been a 100% objectively formulated policy, never mind exclusively based on RCT's. And even allowing for the possibility that some policy decisions were significantly impacted by RCT's, you would have to disregard the argument that statistical validity is not correlated strongly enough to reality for enough points in time to make life and death decisions for anyone other than yourself. I can decide to take a particular drug, but when government tells me I have to take it, something is seriously wrong.
But what about web design? It's not a matter of life and death, but that doesn't mean A/B testing accomplishes anything, other than possibly making the tester feel better. You need a large number of visitors who closely resemble your total market, and you need to repeat your test several times. Over time your market changes, and the population of visitors changes as well. Unlike a drug test, where once you have a winner, you can give it to the control group; with marketing, once your visitor has a negative view of your product, changing the website isn't going to turn that visitor into a customer. By all means use analytics and any other methods to learn about your market. But don't waste your time with A/B testing.
I have to agree with Howard here. Competing in the B2B space and winning the attention of IT, business analysts, programmers and propeller heads is not easy. Awesome content and awesome product are KING. I've done different things, ran different campaigns - tried to "describe the benefits" or "illustrate the value". But it seems to me that by the time people visit my site, they've *already* done their Googling due-diligence to have found me in the first place. These aren't casual browsers for the most part, and when they are, there's enough data about my products to understand what it does via online user guide
It all boils down to enabling potential customers to do the following EASILY and PAINLESSLY:
1) Google searching and finding me via premium content
**2) Landing on my product page which IS my evaluation page
3) Downloading the fully-functioning TRIAL with OPTIONAL email address registration (yes, there's a whole debate about "well, your buyers aren't serious aka they're unqualified if they can't even leave a corporate email address" - part of me agrees, but guess what? I'm more inclined at this stage and with the experience I have thus far to deem this as yet-another-hurdle-from-keeping-potential-customers-productive-because-now-I've-annoyed-them-with-pesky-questions). People want measurable results NOW; not after a few dozen clicks, sales rep callbacks, form submissions etc. It's 2013, like it or not.
4) Product runs within minimal fuss; brainless install process, performs all capabilities and quietly degrades into crippleware after 30 days. License key re-enables it for enterprise-mode.
**The desire to emulate Amazon, Google, CL here is clear: they get to the point, provide you with ONLY the information that's necessary in order for you to arrive at a decision-making frame of mind quickly. I'm almost tempted to incorporate my entire products' online presences into single, independent one-page URLs. No landing pages, about us, etc. Maybe one more link/page for a User Guide (which would provide the literature/information of the product/domain implicitly and double as a source for indexable content that Google feeds off of).
You can lead a horse to water...
Monday, February 11, 2013
This topic is archived. No further replies will be accepted.Other recent topics
Powered by FogBugz