Using Artificial Intelligence to Uncover Hidden Insights

Using Artificial Intelligence to Uncover Hidden Insights

“Alexa, how did my ad campaign perform last month?”

This is how we might receive advertising insights soon thanks to advancements in Artificial Intelligence (AI). Maybe we can’t get an entire analysis yet, but access to repetitive, month over month, percent change graphs are on the horizon.

You may think that using AI for business is not quite here yet, or that the analysis needed is too complex and would require a specific type of data that would only work for some companies. On the contrary, AI is only getting easier, faster, and more accessible to use than ever.

The Reporting Process

More organizations use data visualization tools, such as Tableau, to help present data in an understandable format through charts and maps. A simple reporting process usually includes three steps.

Step 1. Gather data and build charts

Step 2. Highlight what’s in each chart

Step 3. Write how the results are positive or negative

In the end, the step that matters most is step 3 – the results.  AI gives companies the opportunity to move through step 2 more quickly and efficiently.

Natural Language Generation

Companies such as Narrative Science use a version of AI to write about what it sees in data. This is called Natural Language Generation (NLG). This “analysis” is an explanation of the data breakdown when compared to itself, so it literally spells out what is shown in chart form. They partner with Tableau’s visuals to enhance the user experience of reporting.

Here’s an analogy to help explain NLG: Natural Language Generation is to a report what autopilot is to an airplane. So, when used, autopilot essentially covers 95% of what is involved in flying an aircraft. However, the pilot still covers the most critical 5% of the flight, which includes the takeoff and the landing. In a similar way, the NLG system writes about what’s in the collected data, but a research analyst has the important job of determining how that data is valuable and relevant to marketing decision makers.

Efficiency Rules

In short, these systems can be used as data proofreaders to improve current reporting. For example, in every report there may be charts or data that do not add value to the overall outcome or insights. Using this software, the computer can read the data in seconds and provide the analysis that, for example, the differences between week 5 and week 6 in a campaign were negligible. The analyst can then use this info to make the decision to show or expand on another piece of data more relevant or interesting – and this process would be completed much faster utilizing the software rather than solely by human analysis.

We hear things like ‘big data is all around us’ or ‘leveraging your mountain of data is key’ – but what does that mean really?  Data can be used to make better decisions, but having access to data doesn’t mean it’s being used effectively. If data can be interpreted quickly, easily, and efficiently, then it becomes extremely valuable to the decision-making process, in our marketing example or beyond. Tools such as Narrative Science’s NLG and Tableau’s data visualization tools can be utilized together to increase the value of reporting for agencies and clients.