Brand Monitoring Tool

Tool for the online-marketer:
Brand Monitoring

2018

Project for the Semrush company

Project goal

To improve monitoring of published content about brand

Filters

Analysing and improving work with filters

Sentiment

Helping to understand with which mentions need to work firstly

Overview

Showing summary information about mentions

New data

Implementing new kinds of data in the tool

NDA

Any metrics related to users or money under NDA. They can't and won't be used in this case descriptions.

Brand monitoring tool

Filters

Goal: analyse and improve work with filters

What was done

1
Research
Analysed how users work with filters using GA, BQ and inspectlet tools. Found insights which filters was more usable and which was less.
2
Wireframes
Created prototypes and tested them on users. Worked with a UX writer.

3
Development
Checked UI design, how it works in real life. Found bugs on a stage.
4
Check result
Analysed by GA how users changed interactions with the new interface. Checked metrics.

Old filters

New filters. Key changes: lower height, actions and filters became separately, most usable filters firstly, not applied filters visualisation changed.

Brand monitoring tool

Sentiment

Goal: help to understand with which mentions need to work firstly

What was done

1
Wireframes
Created prototype and tested them on users. Worked with a UX writer.
2
Development
Checked UI design, how it works in real life. Found bugs on a stage.
3
Check result
Analysed by GA how users changed interactions with the new interface. Checked metrics.

About choosing visualisation

There was a different variation of visualisation. Some of them were emoji (smile/angry) and text with color markers. Our competitor had similar functionality with emoji. Research showed that both variations are equally understood. But some users said that they don't want to take screenshots and show them to their bosses or clients with emoji. We decided to not use emoji in interface.

Sentiment. Sentiment can be filtered or changed in mention if our algorithm recognized incorrectly.

Checking result

In the first version of the algorithm was many changes sentiment in mention by users.
We decided to improve the algorithm.

Brand monitoring tool

Overview

Goal: show summary information about mentions

What was done

1
Information Architect
Analysed which data we can use and how we can show it graphics.
2
Wireframes
Created prototypes with different visualisation. Worked with a UX writer.

3
Tests
Using series of tests was identified what visualisation was more clear for every type of data.
4
Development
Checked UI design, how it works in real life. Found bugs on a stage.

Full Overview

Estimated reach graphic

The hardest task was to create a graphic for Estimated Reach. Competitor research didn't show any examples. We needed to create the graphic by yourself and test it. First interaction of tests showed that the best to understood was 3rd and 6th graphic. Only the second test series showed the winner that was the 3rd.

Variations of Estimated Reach graphic

Brand monitoring tool

New data

Goal: implement new kind of data in the tool

What was done

1
Information Architect
Analysed new data. Described which data where take place in structure. Found all the places in the service that may be affected by the data implementation.
2
Wireframes
Created prototype and tested them on users. Worked with a UX writer.
3
Development
Checked UI design, how it works in real life. Found bugs on a stage.

Data analysis

The biggest problem of the task was to merge different type of data in one feed. I was finding similarity in data and create new structure for future visualisation.

Article

Forum

Instagram

Brand monitoring tool

Bonus track

A little cool things that I did with the UI designer

At the end of the feed we show notification. Because early users didn't understand that feed ended. Users waited and thought that something was still loading.

About tool. On old version of the about tool was only poor marketing text. We wanted to teach our users how our features work.

Empty state. We found all cases when users can't see any result and now show tips when this happens.

Set up. On old version of the set up was many fields and and the final request wasn't obvious. Campaign name we create automatically from the user's keywords and it can be changed. Not necessary fields shows when the checkbox is on. Summary information explains what combinations of keyword tool will collect.