Big data and social listening with image analysis for Mastellone Hnos

Mastellone Hnos is one of the largest dairy groups in Argentina, with several mass consumption brands. What gets said about them happens across the whole web, all the time and in many formats.

Tracking brand mentions in text solves half the problem: a huge share of the images carrying a company logo never name it in the accompanying text, and those are exactly the ones revealing how customers, detractors and fans actually see the brand.

• The invisible half: text listening leaves out every visual piece that does not come with the keyword.
• A report for everyone is a report for no one: marketing, sales, sustainability and HR need different conversations.
• Volume against noise: collecting from millions of sources without filtering produces a report nobody reads.

The cost of stopping at text mentions is believing you are listening when you hear half.

A data intelligence layer was built adding visual recognition to text listening, delivering to each internal area the conversation that belongs to it.

Scope covered the whole company and all its brands, with outputs aimed at decisions rather than reports.

Collection with filtering

• Real time collection from millions of online sources, with noise filtering.
• Precision searches using logical operators, including comments and opinions from social networks and open forums.

Image analysis, the missing half

• Visual recognition layer collecting and analyzing images containing the brand identifier even when the text does not mention it.
• It widens the analysis base to a volume text listening leaves out entirely.

Listening by area, not one general report

• Conversations organized by topic and by internal department: marketing, sales, sustainability and human resources.
• Evolutionary tracking of sector decisions and own actions, with real time monitors and alerts per active campaign.
• Comparisons against other markets and against the sector, and identification of who shapes brand perception.

Architecture and technology

• Real time social listening with a visual recognition layer over the brand identifier.
• Indicator dashboards, comparative studies and monthly support on interpreting the data.

Learn more about artificial intelligence and digital marketing.

The output is not a mentions report: it is an alerting and decision system over real time data.

Starting point

• Listening limited to text mentions, with visual content outside the analysis.
• Information delivered generically, with no split by internal area.

Strategic impact

• Visual recognition brings in the logo bearing images text was never going to capture.
• Each department receives the conversations that concern it instead of a single report.
• Real time alerts per active campaign allow correcting while the campaign runs.
• Outputs are actionable opportunities, with monthly support to interpret them.