The novelty factor: why data is dangerous | Heart Internet Blog – Focusing on all aspects of the web

Part 1: Other people’s data

You don’t have to search the web too hard to find a plethora of articles, videos and infographics telling you which colour buttons you should use for a greater click-through rate in your emails, when you should post links on social media for the best response rates, how long your web pages should be, and so on.

Unfortunately, there’s no one size fits all solution, as much as we wish there were. This is one part of data collection and analysis that becomes very dangerous: taking someone else’s research and blanket applying it to your own situation.

Let’s take posting on social media as the first example. Even if the data is nicely divided up further – for example, you can see from the random infographic you’ve found that the best time to target thirty year old women is 12pm on a weekday – it still doesn’t take into account enough factors for it to be reliable. Where do these thirty year old women live? What’s their typical job, if any? Are they parents? If so, how old are their children? What’s their typical lifestyle like? Are they single? Do they have much disposable income? What kind of education do they have? What do they buy?

The truth is, when you send that tweet at 12pm, you have no idea whether those thirty year old women are likely to be on their lunch break, checking their emails, giving their kids lunch, out shopping, in a meeting, travelling somewhere, working, or something else entirely. And neither does the person who put together the data. Their data on thirty year old women might be largely based around full-time mums, whereas your audience may be full-time professionals.

Some data goes a lot further, and instead of being one person’s experience, is an average of tens of thousands of tweets or companies. But it’s still not that accurate. For example, when you’re working with tweets at that kind of scale, you’re likely to be analysing data from a lot of average consumers. We work in what is typically not regarded as a consumer industry, so again there’s disparity there if we’re trying to apply that data to web hosting, design services, SEO or similar. A consumer might be more engaged during their lunch break or when they’re commuting or when they get home. An office-based business customer is more likely to pick up your message straight away…provided you’re not sending it on a weekend…

Culture is another important dividing factor to consider for both where the data has come from and your own target audience; a thirty year old Spaniard may be having a siesta at 12pm and miss your post entirely. An email talking about the joys of a snowy winter Christmas is going to alienate an Australian audience more likely to be hitting the beach for their summer Christmas.  A red button might be a positive for a Chinese reader likely to associate it with luck and good fortune, but may be interpreted as a danger sign

There are hundreds, if not thousands, more factors like the above. And when you start cross-referencing categories, it gets even more complicated. If your thirty year old women are also Australian, what does that mean? More to the point, what does it mean for what colour your buttons should be, how long your content should be, and when you should post social media updates?

Say you’re selling web hosting, and we release a lot of data about the times we get the most clicks on social media, the length of pages that get the best results, and so on. That’s going to be a bit more helpful to you, but not much. Even though the typical customer may be fairly similar, the context could be very different purely because of the different sizes of the businesses, locations, style of selling, preconceptions of the business and so on. It’s just as much about you as it is them.

The only way to know what works for you is to forget all that data and test for yourself. And keep testing. You can never know too much about your customers, whether it’s their favourite TV programme or the type of subject line that will grab their attention, because all of it helps to build a bigger picture about your audience, their lifestyles, and their expectations.


Part 2: The novelty factor

So when you’re doing your own tests, how do you know what’s successful?

Say you send an email out to all your customers and decide to test if the colour of your links makes a difference. You change all the links to a bright red, send the email, and then discover that your click-through rate has improved significantly.

So that means your customers really like red links, right?

Not necessarily.

The novelty factor makes a huge difference, whether it’s the fact that your links stand out from everything else in sight, or the sudden change to an entirely different colour. This is where a lot of research and ‘other people’s data’ falls down, because they rarely account for the novelty factor  – either in their own experience or when you try to apply their research to your own audience.

Whether you’re working with emails, adverts, web pages or anything else, always test, change, and retest. There are thousands of tiny different factors you can tweak to make a difference. Factors will also impact other factors, so something that’s successful on its own – e.g. a particular colour – may be affected by changing other aspects, e.g. text.

The main takeaway is that’s important not to jump to conclusions, or assume that just because something is (un)successful now, it’ll continue to be that way. The most accurate data available will always be that from your own campaign reports and tools, so as long as you’re proactively testing each email you send and adjusting them based on your previous results, your approach will always be optimised for your needs and customers.

Comments

Please remember that all comments are moderated and any links you paste in your comment will remain as plain text. If your comment looks like spam it will be deleted. We're looking forward to answering your questions and hearing your comments and opinions!

Got a question? Explore our Support Database. Start a live chat*.
Or log in to raise a ticket for support.
*Please note: you will need to accept cookies to see and use our live chat service