Interview: Exploring Data Science at Brandwatch with Hamish Morgan
By Olivia SwainAug 23
At Brandwatch, we work closely with agencies and we have had the opportunity to help our clients improve across the board and we want to share those experiences and things we’ve learned along the way with you.
Though part of the story, these metrics alone don’t tell us about overall engagement or affinity.
If you want to highlight Likes, explain what drove that action: was it a post, was it a image or was it a promotion? Can you correlate this to an increase in sales, a lift in leads?
If you ask yourself the hard questions first, your client won’t.
I can’t believe how often this one comes up with clients when highlighting examples of reports!
Take your time when grabbing an image – check that it’s clear and large enough. Always remember most listening/analytics platforms have an export to excel/.csv function where you can resize your chart if your platform doesn’t support it.
Though we all want to paint an amazing picture of engagement for our clients, sometimes bad things happen. Too many times I’ve seen client reports avoid the hard issues and/or simply not highlight key insights. Maybe Twitter isn’t the right channel for your brand, or your images aren’t getting retumbled or pinned enough. Let the data set you free.
Though screen shots of engaged tweets or popular posts are helpful, not building a dataset that allows the client to see what is driving engagements, the audience archetypes, interests or drivers simply doesn’t allow for actionable insights which in turn drive recommendations.
The best reports will always highlight how campaign X may have succeeded based on email, search, paid, experiential or even broadcast to understand the complete picture of audiences and effectiveness of engagements. If the opportunity exists to report on a complete dataset do take advantage.
Though you want a concise “short” report, most modern tools will allow you to grant some sort of dashboard access. Rather than just showing the client slides and addendum, give access to a structured dashboard with customer data.
This one is a personal pet peeve of mine and comes with experience. Take your time and make sure your fonts, formatting and colors are on brand. It takes a few moments and makes the difference between a well received deck and a memo asking to fix X, Y & Z before forwarding.
This metric, more than any other, will immediately lead a client to question the validity of your analysis and though it takes time, validation is key. Most modern listening and analytics platforms will allow human overrides to ensure accuracy.
Even the most advanced analysis technologies are still in just robots. Topic extraction, peak detection and signal identification all rely on algorithms and textual analysis to find the most important parts of your data set.
The temptation may be there to simply trust your analysis to this, but in truth there will always be the need for a human analyst – sometimes manually diving into the data will surface hidden insights that a technology can’t find automatically, and any false positives churned out by the machine can also be isolated and dismissed. So, while there’s definitely great value in using sophisticated technology to aid users in the social data analysis process, be prepared to get your own hands dirty too.
This one still makes me giggle from time to time when migrating clients from competing platforms – often when we discuss conversions, sometimes they might not even know why they are showing data to a client and what it means.
In conclusion, some platforms are simply better in supporting the points above but most will allow you to blow your clients away with good insight driven reporting.
We’ve also written a complete guide to writing a social media report.