Working in marketing, you’ve probably heard that “data should tell a story.” It sounds like a hacky business cliche, but thinking about data in terms of storytelling can help you be a lot more effective in creating and delivering reports, dashboards and other data-driven recommendations to your clients and stakeholders.
Simply put, “data should tell a story” means that you should never be reporting on information without having a clear idea of what it means, the context in which it should be interpreted, and the actions you’d expect to be taken as a result. Anyone can pull a report – the real value is in using that report to effect positive change.
If you’re taking time to track a given metric, you should have a good idea of what you’re planning to do with that information. Your data collection strategy should be founded in clear goals for your overall site. Ask yourself:
- What is this website intended to do? This might be things like selling products, generating leads, building a company brand, etc.
- What will we do to achieve these goals? Activities like marketing campaigns, site optimization, and content creation should all be mapped back to concrete goals for your website and your business.
- How will we know if our efforts are successful? This is where you outline the metrics that you want to track.
- What action will we take as a result of this data? You should have a clear idea of what you’d want to do next if your chosen metrics look good and/or if they’re not performing as well as you’d hoped.
When setting up e.g. custom Events in Google Analytics, it can be very tempting to track every possible engagement point for your site’s users (especially now that Google Tag Manager has made doing so much easier), but too much data becomes unwieldy and not useful. “We want to see how people are using the site” seems like a good goal, but it’s hard to put general site usage data to use in a meaningful way. Instead, form a series of hypotheses you’d like to test (“Our blog has a high bounce rate because there’s no call to action at the end of each post”) or questions you’d like to answer (“Which items in our navigation could be culled out without sacrificing user experience?”) and set up your data collection around those.
Now that you’ve defined how you’d like to use your data, it’s time to make sure you’re presenting it in a way that will support that use. What’s going on with a given metric should be clear at a glance, so you can quickly determine whether you need to take action; think about the ways in which your data visualization may be obscuring that need for action.
One classic example is Google Analytics’ defaulting to charting metrics over time by day. Measuring by day can be useful if you’re only interested in performance over a short time period (say, a week or two), but natural fluctuations by day of the week can make it harder to observe overall trends on a longer time scale.
This is especially true for B2B businesses, who typically see a sharp drop in traffic over weekends; the natural up-and-down spiking of a day-by-day graph can make it harder to discern subtle overall trends that become clearer when charted by week or month.
Effective data visualization should also include, wherever possible, the necessary context to interpret the data. For example, comparing traffic to the previous year’s data will help you quickly determine whether a drop in traffic is seasonal in nature, or whether more immediate action needs to be taken.
Reports should always be tailored to their consumers; different stakeholders in the business will need different pieces of information in order to decide whether or not to take action. Wherever possible, create separate dashboards and reports for each person to whom you’re reporting. It can be tempting to contain everyone’s metrics in one mega-report, but in my experience, this can reduce the likelihood that the right information will reach the right person. Someone who receives a giant report that requires them to dig to find the information they need will be disincentivized to look at the report at all – you want to make information effortless to consume, so the people viewing it can define a course of action as effortlessly as possible.
The last piece of effective data storytelling is analysis. When presenting data to a client or an internal stakeholder, you never want to just hand over the raw numbers. You should already have taken the time to dig into the data and provide some analysis along with your report. In many cases, this will mean recommending a course of action (since we’ve already set up our data collection and presentation to drive action).
I also like to take a look at the data I’m presenting and find places where I think “What happened there?” or otherwise have questions about what I’m seeing. Put yourself in the shoes of someone who is seeing this data for the first time, and try to anticipate the questions they’ll ask. Then, include the answers to those questions as part of your analysis.
Doing so will not only help your data consumers get to their course of action faster, it will also build trust with them, because they see that you’re attentive to their needs and conversant with the data. It’s OK to answer questions with “I don’t know” from time to time, but the more you can anticipate questions ahead of time, the more you can come to the table with a definitive answer, which makes you look like a rock star.
Over time, reports have a tendency to warp out of shape. Data points are added but never removed, priorities change without the report changing with them, additional stakeholders are added to the mix, and before you know it, your reports and dashboards look pretty different from how they started.
The best way to fight report bloat is to revisit your reports every 6 months or so, and ask:
Do we still know what actions this data was intended to drive?
Has this data been successful in driving action?
Is this being delivered to the right stakeholder?
Are there any additional data points connected to our goals that aren’t represented here?
This periodic pruning and reworking will ensure that your data stays lean, mean, and totally actionable.