Media monitoring has come an awful long way since the days of the press clipping service. Much of that history is one of the time lag between publication and ‘consumption’ becoming ever shorter. And as it has done so, what can be achieved by media monitoring has changed also.
The truth is that back in the day, when ‘media monitoring’ meant flicking through a folder of newspaper cuttings that told us what we knew already, it was more an exercise in curiosity than anything seriously influencing day-to-day business decisions. The time lag was simply too great. So whilst feedback could be useful, interesting, and provide food for thought, it wasn’t embedded into business processes.
That has changed with time. The time lag is now hours or even minutes. A good media monitoring system can (and should) be used to tell us things we don’t know, and alert us to changes in our own organization and the environment that organization operates in. On the back of this information, we can change course and adapt our decision-making in something close to real time.
But we now find ourselves in a new position again. With the advent of media monitoring that scans the global news media in close to real time, and automatically alerts analysts as soon as anomalies in coverage are detected, we are able to use that monitoring to proactively identify potential issues and events: almost before they have officially ‘happened’.
What does this mean, and how would we implement such a system? Read on.
Looking for signals in ‘big media data’
If one fact characterizes the modern media landscape it is this: there is a lot of it. Conventionally, this fact frustrates media monitoring organizations and the analysts who work for them: more content means more work after all.
But for data scientists, lots of data is an opportunity. It brings with it the ability to scan that data for patterns and identify things that would otherwise go unnoticed. In the context of the media specifically, it enables us to see that something is happening, even before we might be aware of precisely what.
Consider this example: a customer of a financial institution has been linked with a terrorist organization in media reports within a specific country. Although nothing clear cut has yet been established, there is enough ‘smoke’ to cause an uplift in references to this individual in local media.
A truly proactive media monitoring service does the following:
- Enables the analyst to set up monitoring of particular entities (a person, organization, place etc), and topics that they wish to actively monitor
- Constantly scans the extended media landscape (in multiple languages) for mentions of that entity
- Reports changes in the volume or nature of coverage, thus enabling a proactive alert to be delivered not because coverage was reported (something that happens all day every day for high profile entities) but because the pattern of coverage changed.
That last point is key. It is relatively straightforward to be alerted when coverage is published. But alerts like these are not helpful. Effectively they drown analysts in irrelevant notifications and fail to identify what really matters: that a change in the nature of coverage suggests that something needs investigating further.
After being notified, the analyst can then investigate further: read the latest coverage relating to the individual and make a decision as to what action to take.
What is of particular interest here is that changes in the pattern of coverage can be the single fastest way to understand that something is about to happen, or that a story is about to break. Essentially we are aggregating speculation (or rumour, if you prefer) in order to signify that a potential entity needs to be looked into - now. Because as we have said many times on this blog, the sooner we are aware of a potential issue, the better the chances of avoiding or minimizing the subsequent damage to the organization.
More thoughts on proactive alerts
Of course alerts have a broader relevance in media monitoring. They enable the method of work to be turned on its head: from going out looking for information (which is a time-consuming business, even with all the automation we take for-granted today), to information coming to the analyst.
For this to work, however, we need two things: the ability to specify fine-grained alert criteria, and - most importantly - accuracy. Let’s talk about each in turn.
Alerts mean notifications, and the golden rule when it comes to notifications is relevance. Being alerted to something that is not relevant is costly and irritating. Perhaps worse, it can teach analysts to ignore their own systems: there’s only so many times a boy can cry wolf after all.
To maximize relevance it is necessary to ensure that the analyst can configure the media monitoring platform to alert when a specific set of circumstances are met, or only for specific types of media coverage (relating to a particular entity, whilst also being within a particular subject category, from specific sources and when sentiment is negative for example). Focused alerts such as these are helpful when we know what we are already looking for, whilst those described above help us find potential events to look into in more detail.
Lastly, and perhaps most importantly, is accuracy. The specific issue here is false positives, the bane of any media monitoring system. Already a challenge that most will be familiar with (try running a simple keyword search for Square, Stripe or Zoom), the issue is doubled with considering proactive alerts. Put simply, a media monitoring platform has to be close to bulletproof when it comes to delivering results solely relating to the entity that an analyst is interested in.
Get over that hurdle, however, and proactive alerts can be an incredibly effective way to monitor the world in real-time, for minimal effort.
If this sounds good to you, sign up for a RADAR trial today, AYLIEN's new end-to-end analyst research platform.