Skip to content

The Problem with Dashboards: Show Me the Spikes!

Being a marketer, I love it when something spikes: clickthrough rates using new copy or creative, conversions from a specific channel or on a new campaign landing page, time on page. I also like to know right away when a bad spike happens: increased CPL, abandonment rates, opt outs, and more. The challenge I have as a marketer is that BI dashboards are not effective at surfacing and helping me drill into good or bad spikes. And I need that capability to identify new opportunities to increase revenue or marketing spend performance.

I joined Outlier because of this challenge. 

BI dashboards just didn’t give me what I needed as a marketer, and the Outlier automated business analysis platform does. Big time. I can’t tell you how many times I’ve worked with an ops team or analyst to spec and build dashboards, schedule a weekly review with the team, and then waste time each week staring at answers to questions we already knew. My ongoing pain included:

  • Getting frustrated at our collective inability to quickly drill in to understand what’s really happening beneath the surface rollup. 
  • Feeling anxious heading into an executive review and not having any meaningful insights to share. 
  • Worried that my team doesn’t really have a grasp on what’s happening across all our paid media channels, website initiatives, and outbound campaigns.

And I’m just a B2B marketer. 

I can only imagine how much worse this problem is for large brands with hundreds or thousands of product lines and campaigns, large web properties, multi-million dollar paid media budgets, and complex ecommerce funnels. Actually, I know how bad the problem is, because we have iconic brands using our platform every day to solve the fundamental problems with BI dashboards.

The core problem is that the best dashboards and analytics teams can maybe provide high level insights into a fraction of the available data, like maybe 10-20%. These dashboards can only answer questions brands know to ask, and typically aggregate data, which means critical insights, one or more levels down, go unnoticed. The result is that: 

  • 79% don’t use all their customer data
  • 66% say they don’t get insights fast enough 
  • 73% catch new trends by surprise (preliminary findings from a survey we’ve commissioned and will formally publish in August).

The Outlier platform solves this seemingly intractable challenge. 

It quickly integrates into all major data repositories—Google Analytics, Google Ads, Adobe Analytics, transactional databases, and more—and uses advanced AI each day ask every possible question of all the data, generate thousands of potential insights, identify a handful of potentially important insights, and share these select insights as stories so marketing and analytics teams can collaborate and take appropriate action.

Outlier has over a dozen story types, one of which is a spike story. A spike identifies when a metric value jumps above the modeled, expected range. Here’s a screenshot of an example Outlier spike story:

Outlier Spike Story
Outlier Spike Story

Outlier monitors for these deviations and changes over daily, weekly, and monthly spans. And again, it models expected range and looks for spikes across all data sources and metrics without needing any configuration. That’s why we call it automated business analysis. Once a single data point moves above the expected range, a spike story is automatically created. This allows you to take quick action if the spike in your data reflects something urgent in your business.

Also, of note, if the spikes continue for multiple days, then a new trend story is automatically created. More on that story type in a future post.

How do you read a spike story? 

In the example above, we see that session duration (metric) has spiked for ages (segment): 25-34 (segment value). We see that our data point for this particular day on April 22 has spiked 66% over our model and affected 0.7% of our impressions. The light green shading represents the expected range, which is calculated using historical data points for this time series.  The orange dot to the far right of the graph shows you the spike. This entire story, along with any related movements and suggestions for further investigation, is generated automatically.

Any metric can be tracked based on its increasing or decreasing movement. Thus, spike and the related drop story types, can alert you of sudden changes quickly, helping you monitor changes that may become new trends over time. 

The potential use cases are nearly endless. 

Spike stories help marketing teams track ad performance to see if CPM is increasing significantly. Finance teams can spot sudden changes to cash flow. Sales teams can use this story type to track performance by account executive and territory. Customer success teams can be quickly alerted about a sudden increase in service tickets, following a new product release. Operations teams get immediate alerts about issues that may impact their supply chain, as shown in this example:

Average Time to Dispense Spike Story
Average Time to Dispense Spike Story

An enterprise dashboard can show you increases and drops too, but it only shows you the metric or KPI that you know to look for and choose to track. Outlier automatically monitors for increases and decreases of a metric and all of its related segments over days, weeks, and months, comparing values to their expected range. 

Not only does Outlier show you the spikes and drops that you might not be expecting and aren’t tracking in your dashboard, but it also alerts you to those small changes that could strongly impact your business. 

And again, Outlier looks for spikes across all your data, curates a small list of stories of likely importance, and automatically generates these stories using natural language and graphs that any marketer can understand and take action on…every single day.

To learn more about Outlier AI please contact us or schedule a demo.

Tagged: