As a general rule, marketers should major in the majors. We should pay attention to the things that make a significant difference. We don’t want to waste time trying to solve “edge cases” that nip at the periphery.
But sometimes, these “edge cases” are more important than we realize, and they can expose issues flying under the radar. It’s wise to take a second look and ensure you’re not missing something.
Edge cases occur when:
- Someone behaves in a way you didn’t expect.
- An action takes place in an environment you didn’t plan for.
- A customer has untypical circumstances.
- Two different systems collide.
- There’s a change in regulations or technology.
- A device is stressed, such as being low on memory.
A friend told me that he’d worked for months to perfect a computer system, and when his boss came to test it, he dropped both arms on the keyboard. The system immediately crashed. My friend said, “Why did you do that?” and his boss answered, “Because some customers will.”
Human beings rarely follow the rules.
Edge cases are usually considered to be more of an issue for software and hardware developers than they are for marketers. Software has to work for every user – or as close to “every” as can be managed. Marketers are dealing with audiences, and it’s understood that some people will fall through the cracks.
The thing is, sometimes those “cracks” are actually opportunities or market segments you’re missing. And other times, they’re just smoke.
When ‘edge case’ is an excuse
A standard use case for a customer data platform (CDP) is to convert unknown users to known ones. One way to do that is to append a customer ID as a query parameter to the links in your email. When the recipient clicks through, the CDP can capture that customer ID and convert a formerly unknown web profile to known.
But what if the recipient has forwarded the email, or has posted the link to social media? Whoever clicks on those links will now be identified by that customer ID, corrupting your data.
If your CDP tells you, “This is an edge case,” what that really means is that they don’t have a good way to solve the problem. It’s not an edge case at all. People forward emails all the time.
You need to press your CDP vendor for a better solution. Don’t accept “that’s an edge case” as an excuse.
Failing to consider edge cases led to some very dramatic and public failures. Here are two examples.
The iPhone 4 went on sale to much acclaim in June of 2010. Apple said it was the thinnest smartphone yet, and the faithful stood in long lines to get theirs.
Very soon afterward, customers reported the phone would lose its network connection when held in the left hand.
That sounds silly, and it’s easy to imagine the Apple engineers blaming it on “user error” and ignoring the problem. “All phones have sensitive areas.” Go away, kid, you bother me.
Steve Jobs got in on the act, and his condescending response became an internet meme.
I’m sure someone called it an “edge case.” It was a design flaw.
It became a big scandal for Apple – all because they were looking at the big picture and ignoring what seemed to be an inconsequential, rare problem.
In 2016, Microsoft created a chatbot called Tay. It was an experiment in “conversational understanding” and was designed to engage people in dialog while emulating the style and slang of a teenage girl.
It was supposed to learn over time. But, boy, did it learn. It didn’t take long before it started to tweet offensive things.
Nobody at Microsoft considered the possibility that someone might try to sabotage the experiment. Troublemakers on 4chan inundated the bot with offensive language, and the bot learned a little too much.
Failure to imagine what might go wrong – or what might be wrong with your process or data – is very dangerous and can affect marketing technology, too.
Not everyone has a smartphone
Assuming that everyone has a smartphone and that it will work the way you think it will work is a classic example of an “edge case” that isn’t really an edge case.
I was on a camping trip with my friends a little while ago. We called ahead to find out how to register for a site, and they told us we could register at the trailhead. That’s a normal thing at state parks. There are cards you can fill out, and a box where you can drop your registration fee.
But that was too old-fashioned. They had to bring things into the modern world, so they put up a sign with a QR code. Unfortunately, there was no cell service in that area.
We camped anyway. The next day the park rangers asked us what was up, and we explained the situation. They laughed and said when we got somewhere with service, we could log in and pay online.
The app didn’t allow you to pay for something in the past. The whole system was a catastrophe because someone assumed everyone always had a working smartphone.
I have a friend who is a quadriplegic and can’t operate a smartphone. I can’t tell you the number of times he’s run into impossible loops because the system assumes he can get a code off his phone – or some other procedure that he simply can’t follow.
It’s easy to think of “the guy who doesn’t have a smartphone” as an edge case. It’s not.
Emojis are big these days, and some say they can increase response rates in email campaigns. But they don’t always display properly.
Suppose you’re looking at the big picture. In that case, the overall benefit of using an emoji in a subject line might far outweigh the ugly results when a minority of email clients can’t render the emoji. Marketing is a percentage game, after all.
“That’s just an edge case,” your email marketing pro assures you. “Modern email clients handle that just fine.”
OK, but where else do emojis break, and are there any downstream effects? What if your emoji-powered emails are also sent to messaging apps or posted to social media and end up creating a mess?
When screen readers encounter an emoji, they describe them. This can cause trouble if you start your message with an emoji.
Digging into this “edge case” can help you create a better experience for your customers.
The basic concept of a funnel is that if you get enough people at the top and optimize each step in the funnel, you can get more people at the bottom – where they give you money. It’s a game of volume and percentages.
If some quirk (an “edge case”) in your system is causing you to lose 3 percent at the top of the funnel, that’s no big deal, right? You adjust your inputs to compensate. Get a different 3 percent.
But what 3% are you losing? When you’re dealing with numbers and averages, you’re likely to assume the 3% you’re losing is a representative sample of the whole. You might be losing your best customers.
For example, what if the quirk is cutting off people who use old iPhones, and that’s a characteristic of your target market? Say, “budget-conscious consumers who are late technology adopters.” Not investigating this “edge case” will make your whole funnel useless.
Dig deeper: If it’s not a sales funnel, what is it?
‘Edge case’ or outlier?
Outliers are small amounts of data that don’t fit the normal pattern. They’re those annoying points that sit way outside the nice straight line you’re trying to draw. They might be the result of a sampling error or misreporting. Or they might point to something important.
It was an outlier that led to the discovery of fluoride’s effectiveness in preventing tooth decay, and sometimes, these “outliers” might point to an important market segment, like a subgroup that operates under different rules.
Gleaning marketing insights from edge cases
It’s easy and tempting to dismiss a quirk in your data, process or customer complaints as an edge case. It’s easy to assume that everyone is normal, that they follow the rules and that the ordinary process that everybody else uses will be just fine for your customers.
It might. In fact, it probably will. But it’s worth applying a little of your creativity and investigative efforts to make sure. The downside of failing to do that can be significant.
Get MarTech! Daily. Free. In your inbox.