Whether we’re talking about everyday life, or the world of business platforms – we believe that decisions, big or small, should be made based on data. This forms the crux of our technology – so how does one make effective data-driven decisions regarding customers? How do you innovate the experience, and optimize the content they view and consume? Ultimately, how do you make customers more successful?
There’s been a lot of content structured around these very questions – do a quick Google search and you’ll find a ton of reading material. We had a look at some of these posts and wanted to share how our thoughts – our unique take on the industry of business data, really – differs.
More than anything, you’ll find that online publications believe that high-level plans should be based around collected, factual data. Other publications focus more heavily on data trends and anomalies; another useful tool in determining the next necessary steps needed.
Because trailblazers like being different, we’re not going to focus on any of those notions. With our own unique concepts of success comes our own unique methods to reaching that success – we believe the there are three keys to making powerful data-driven decisions:
Grouping, Comparing, and Sequencing.
Grouping is a key step in understanding data – before you even begin going through numbers, it is essential to group your users into sections that can be tracked and measured. If making your customers successful is your primary goal (it’s certainly ours), you must first be able to see which groups of customers are doing what, when they’re doing it, and where it’s happening. This is where specificity becomes crucial – many companies make the mistake of tracking too broadly (weekly or monthly), ultimately missing out on more specific insights about group behaviors. Don’t get us wrong; making high-level groupings can be useful and telling. But without digging into more specific instances, you run the risk of losing customer loyalty by missing more minute analytics buried in the data.
Piggy-backing on the previous paragraph, it’s important to distinguish between broad and specific groupings – high-level grouping includes categories like age, gender, location, and income level. More specific conditions can than be scrutinized for more minute data; like actions, interactions, transactions, customers that purchase multiple items and customers who spend, say, 100 hours browsing the site. Again, these are just examples – there are thousands of specific questions and data points just like these. Further complexities are added when combining specific points with general ones, like: “all women who are over 40 years old who have purchased something more than once via Twitter or Facebook.” Now we’re getting more specific.
Yet the rabbit hole gets deeper. Adding a time-range is one of the quickest ways to analyze specific data and compare patterns that occur during different points throughout the year (particularly useful for online vendors). So on top of the aforementioned example, you could add “in the last 30 days,” or, “within any 30 minute time span.” The latter example showcases our flexibility – you can literally monitor the analytics of a new product only minutes after its release, then compare that data with longer-running stats later. Comparison is a crucial element of making data-driven decisions; so let’s talk about why.
At a fundamental level, we make the best decisions by comparing our options – 100% of the time, making a decision means weighing the potential benefits and downfalls of two or more options.
Therefore, comparing two or more things allows us to properly assess the similarities and differences between all of our options and then base our decision on specific data points that come out of that comparison. Now, the most common comparison to make is between the same group of users over different periods of time. For example, we can compare results of a group this week versus last week, or today versus this same day of the week last month, etc. This has proven very useful in making high-level decisions that affect the business and direction of the company.
In the context of achieving successful user experience and customer journeys, however, this comparison becomes even more insightful and revealing when we consider two groups of customers that are specific, yet slightly different.
Let’s look at an example of this notion. Let’s say we’ve grouped our customers into two sections: “California residents over 30-years-old who have made a single purchase in the past week,” and “Texas residents over 30-years-old who have made a single purchase in the past week.” Because most of the parameters are identical, we know to compare the differences in locations. If the results were roughly similar, we could deduce that location has little to do with the performance of this particular hypothetical. This is a simple example, but clearly illustrates how these sorts of analytics can be used to remove or disregard factors that don’t cause impactful change.
Similarly, let’s posit another hypothetical. Let’s make the same comparison, only this time, let’s change the age group – once again, if the results remain similar, we can rule that factor out as well.
Here’s where it gets tricky – let’s take the first example, and hypothetically surmise that the change in location yielded divisive results. This would automatically tell the user that the purchase location is an integral factor in determining the success of the campaign, meaning more questions and analysis should be focused on that particular aspect is not only accurate, but defensible.
The point is that there’s a fine line to walk when grouping your customers – you want to group them specifically, sure, but also make sure to maintain some level of flexibility for the sake of comparison. By continually making nuanced comparisons, you can find the single group property, attribute, or condition that ensures your data-driven decision. Let’s touch on the final key technique needed to make data-driven decisions – sequencing.
Any user event, by default, is defined by the following three required elements:
- Actor – the user or device that completes an action.
- Action – something that an actor completes.
- Timestamp – the point in time when an actor completes the action.
Though these are the three fundamental aspects of any given user event, there are supplemental attributes that sometimes come into play.
Let’s talk about grouping actors. As we discussed previously, actors can be grouped based on specific properties or conditions, then studied over a chosen period of time. As such, the user event timestamp is always referenced for inclusion or exclusion purposes. Put plainly: user events only exist within the specified time range of the comparison – otherwise the results are simply excluded, making for useless data. This is important to keep in mind.
Let’s look at an example: “How many users over the age of 18 have signed up for our newsletter this month?,” is a common and simple question to analyze. Replacing “…this month” with “in the last month” will give us a new set of numbers, and a basis for a fair comparison. This is the quickest way to answer a simple question: have the numbers increased or decreased since last month?
This is fairly standard practice, and though it offers a general sense of direction, it’s also a simple comparison in the world of analytics – there’s more insight to be gained by digging into that timeframe, like user actions, interactions and transactions. Simply put, we now know how much one set of numbers changed in that given time, but we don’t know why.
We can certainly guess, (or derive some theory or hypothesis), and base our decisions on guesses fueled by broad data – though this hardly seems data-driven. Even with a general framework, a guess is still a guess. Luckily, each event has a timestamp, meaning we have the option of laying out each unique event sequentially (hence the term ‘sequencing’). Once the events are laid out in a timeline that makes sense, we can then retroactively rebuild the exact paths and flows our actors took. In other words, we can study their behavior every step of the way, then compare those steps to other actors in a different grouping.
The result? An acute ability to spot patterns that emerge from event sequencing that helps us determine why or who did what, and when and where they did it. Eventually, we can explain and defend our data-driven decisions based on the culmination of these four indicators – which leaves us with how our customers did what they did.
As you can see, a whole lot of analytical concepts go into making sound, data-driven decisions. Typical methods of gathering and parsing metrics aren’t deep enough, and will likely point you in the right direction but leave you guessing when it times come to make important decisions.
And at the end of the day, that’s what this is all about. Making a data-driven decision is dependent on how users group, sequence, and compare user events – this is the most concrete way to reveal, in detail, all facets of customer actions and interactions. This means that users can ensure their customers have a successful experience, while simultaneously establishing trust and brand loyalty every step of the way. When it comes to data, we want users to stop guessing, and start knowing – a mantra we repeat at Interana that is music to our ears. Speaking of which, try it for yourself. Turn your team into a true orchestra of productivity and collaborative thought, and hear the music too.