Demystifying Usability

Ad conversion rate influenced by time (not click rate)

Time is an important design variable to understand. Your user experience is effected by it no matter what user experience you are serving up and the rules are different for every context. For example, the "three click rule" (users must get to their destination within three clicks) applies to e-commerce primarily but not to mortgage education, financial services usability or reading the New York Times online.

Time Usability 101: Design for asynchronous interaction. Asynchronous means users don't do it all in one setting. Environmental distraction is inherent to any user experience: the pasta is burning, the kids are screaming, the documents are being tracked down, printer cartridge is being changed, someone interrupts...

Here is some fascinating research about conversion rates and the influence of time:

Up to 85 percent of conversions occur after the day the most recent advertising impression was served, according to new research from Advertising.com. The results seem to show online ads can have an impact on buyers' purchasing decisions long after they view an ad.

For the five-day monitoring period, approximately one-third of all conversions happened on the same day that the impression was served, but only nine to 11 percent occurred within three hours of the impression being served. The longer periods of monitoring revealed that up to 85 percent of conversions occurred days after a user was served an impression.

"These findings show what a significant portion of conversion activity takes place well beyond the impression," said Scott Ferber, chief executive officer of Advertising.com. He said that the data shows that metrics such as per-click or per-action may not give marketers an accurate picture of campaign results. "It also reveals how lasting the effect of an impression can be," he said.

This data supports the theory that users' consumption of the Internet is often patterned and repeated on a daily basis.

If you know when your audience is going to be online or converting, there is some value to targeting by time of day."

The implications for web analytics analysis, advertising and content management are interesting:

1) Ad impressions go through a 48 hour or so window of decision making activity. Decision making being influenced by time and potentially access to external information, persons or environments.

2) Interpreting user behavior through a web analytics perspective ought to include a sensitivity for the impact of time (i.e. conversion rates are non-local meaning they aren't isolated to immediate on-page actions as is the current view of traffic log analysis).

3) Understand how you are supporting time usability. Do you offer good time-out messages, good error protection from browser instability (if serving a web based application)? Does your content support good decision making across time?

4) Consider your website usability strategy as a variety of ad impression. Already, independent studies show that website usability (ease of use) increases the likelihood of purchase. More on this later.

5) Get ahead of the competition and synchronize ad impressions with conversion behavior. In other words, throw out the purchase cycle across time and serve up product and service content that parallels return visitor activity. In some cases with high ticket software solutions, getting clear about this may lead to strategic leaps in revenue generation that otherwise was pegged as just low conversion.

If you'd like to check out the science of time usability, have a skim of this paper (PDF 196K) Daniel Loewus-Deitch and I (Frank Spillers) published at the workshop on the Temporal Aspects of Tasks in September at HCI 2003 in Bath, England.

Best Wishes,
FS

Patenting usability improvements- going over the top?

Summary: Questioning Jakob Nielsen’s recent advice about competitor data and his advice to patent usability best practices.

Slashdotters react to Jakob Nielsen’s recent column. And they raise some good points:

• "The reason this logic doesn't follow for many software engineers and software patents, is that the stuff that gets patented is sometimes simple, and a horrible waste of cycles and time happens globally as people work around the patents".

• "Usability is required on *all* sites, not just the ones that have the resources to patent everything".

• "So if I come up with a great usability enhancement, I should patent it? How does that increase the usability of the web overall? We live in a sea of unusable web sites and horribly designed programs.. now he's saying "hey, the goal is actually not to make web sites more usable. The goal is to come up with usability enhancements that one or two web sites will use. The other sites can go stuff themselves."

• "...is just going to be a directory of patents? Just swap out any point where there is actual usability advice with the relevant patent number and you're done".

The key issue here, I think, is that Jakob Nielsen has an unhealthy monopoly on Usability Consciousness. He promotes best practices, he preaches obedience to his guidelines and when he postulates opinions they are interpreted as instructions.

More later on how this has eroded Jakob Nielsen’s popularity within the usability community.

Best Wishes,
FS

Email Marketing- New Legislation

In case you missed the big news in the last few weeks, there is a new anti-spam law that was enacted on January 1st 2004. The rules for sending email have changed. Marketing Sherpa has a good summary:

Update Memo: CAN-SPAM Good News & Bad,
Bad, Bad, Bad News

SUMMARY: We just spent two and a half hours quizzing the FTC's attorney on exactly how the new law affects emailers. Here's a quick, handy round-up for you:

Good news: It's fairly easy to comply with message content.

Bad News #1: It's hard to comply with opt-out rules. Welcome to suppression file hell...

Bad News #2: The law applies to more types of organizations and email than you think.

Bad News #3: There's no "grace period."

Bad News #4: And, almost anyone can sue you.

Read on... and then forward to your legal department:
http://www.marketingsherpa.com/sample.cfm?contentID=2576
(Open access = permanent)

I don't know about you, but my spam hasn't reduced since the start of the year :-(

Best Wishes,
FS

Site Search can be flattened by Usability

If you're using a search solution on your website (Ask Jeeves, EasyAsk, Google, Inktomi, Mercado, Verity etc) or simply offer a search interface that users can find products, information or documents with-- you may be missing the main feature that will make it work (and give you the return on investment you want): Search Interface and Search Results Usability.

The search interface runs into these typical problems that most people don't consider when shopping for a search solution:

1) Search Interface: how the database works typically defines how the search interface will look (see the "Boolean logic" and "Natural Language" issues below)

2) Search Results: how the user configures the search is going to determine how great the results are. Also how the search database works will determine how *robust* search results can be displayed. Displaying search results in a readable, understandable and helpful manner is instrumental to designing the user experience effectively.

Search appears to be perceived as a highly valuable resource by users and marketing managers as an essential "help" feature on a website. As many of my clients can attest to, finding a good search vendor is a bit of a headache. One of the big issues is the feature many search vendors sell called "Natural Language Search".

Question: What is Natural Language Search?

Answer: Natural Language Search (NLS) is a more robust version than standard "Boolean" searching. Boolean logic says "And this, Or that, Not that" such as with this search interface http://search.internet.com/ (Probably the worst search interface on the web?) In that example the interface exposes the *needs of the database* e.g. you must enter this and that but not that.... the problem with *exposing Boolean logic* parameters to the user is that if they get it wrong, they may end up with nothing or something off-topic.

NLS really is a back-end feature. NLS database matches "fuzzy" records in the database so users can type "What's for lunch?" and get intelligent results back (supposedly). Ask Jeeves is the classic example (though it is not that great to put it mildly). Google of course has perfected it and Amazon too...but those are very expensive databases used by those two companies.

The main usability question to ask when reviewing a search vendor or evaluating the user experience of your own search solution is, "Can you customize the results?" and "Can you simplify the search interface?". Those two things matter for ease of use and value to the user. Search Results are as important as the interface.

I noticed Amazon is "culling" search results more...so that if you search "Interaction Design" on Amazon they cut it to 2 top results and then "View 230 others". The search rule system is imposing something onto results to prioritize them by popularity (or other variables).

Question: Why is it that when I want to find a Starbucks in an unfamiliar place, Starbucks store locator search is useless?

Answer: Starbucks coffee shops occupy multiple addresses in the same area of town across the country. When a colleague you meet wants to have coffee and says: "Meet me at 1134 NW Glisan (Portland, OR) Starbucks" (more specific then meet me at the one Downtown on 3rd for example), when you type in the address it gives you: two in the area, and then, a list with details for 163 stores in the entire city. Even when reducing the default 5 mile radius to two miles, it acts the same way. The Glisan store is on one of the 20 pages of search results! (try it for yourself)

What is happening I think is that the database is not matching inputted data to the store details page but instead is serving up all results based on geographic data. Result: annoying and unhelpful. Possible solutions include filtering results to match address to actual store pages; adding semantic data to the result (e.g. "across from the Library"); allowing users to search with visual maps of the city and filtering stores based on simple user selection (as in the example below). Starbucks unhelpful search has caused me: to go to the wrong store; pick up the phone and call to confirm; trade 3-4 additional emails to confirm (with the business person I am usually just meeting for the first time). read: hasstle

Here's an example of a successful search implementation:
My client used a Boolean search interface before with a 94% search failure rate during usability testing, we changed it to this one, leading to a 44% increase in registrations, two months after the change in search interface and search results usability.

Best Wishes,
FS

Quote this as: (2003) Spillers, Frank. Web Usability Best Practice Handbook. Experience Dynamics.