Choose your language
Do you know that we are indexing all major social networks in 187 languages
Customer Case Studies
How image recognition technology uncovered business opportunities for OpenKnowledge

How image recognition technology uncovered business opportunities for OpenKnowledge

A pizza restaurant looking to expand, chose to analyze social data with image recognition technology. It was able to identify and categorize consumers’ consumption habits. Would you like to eat inside or al fresco? What would you like to drink? Extra cheese? Let’s take a look!

Table of contents

  • Menu
  • The strategic value of measuring user generated content
  • A no-barriers approach to business expansion
  • A restaurant with a mission
  • Extracting insights step by step
  • Adapting business expansion plans to consumer behavior
  • The takeaway: strategic actions based on image recognition insights


A pizza restaurant looking to expand to new locations? Easy. Everybody loves pizza, right?

Venice-based Bella&Brava is no ordinary pizzeria. When it came to market research for their business expansion, they decided to go where their audience is - social media.

Think user-generated content. Think strategic insights from millions of images of pizza. And the best thing? The most delicious results have a measurable impact.

Bella&Brava pizza restaurant logo

Strategic value of measuring user-generated content

Bella&Brava is a startup in the food & beverage industry. The restaurant reimagines the centuries old tradition of Italian pizza for a young audience that favors sustainable ingredients and healthy, vegetarian or vegan diets.

Planning a business expansion, Bella&Brava collaborated with consultancy company OpenKnowledge. They took a digital-first approach to research ten potential new locations across Europe. Keen to match the expectations of their young, dynamic target audience, the company tapped into social data from Talkwalker to find out in which European cities people would be most open to their restaurant concept.

OpenKnowledge knew that Bella&Brava could get an edge if they used strategic insights from image analysis for the expansion of their business plans.

Pizza has always been a social connector - not just in Italy, but all across the world. People take millions of pictures containing pizza, amounting to a giant database of valuable information. Where people eat pizza. Which ingredients they favor. What they drink with it.

Together, the two companies targeted the following objectives for the measurement study.

  • Extract strategic insights from user-generated content. Identifying ten European locations suitable for business expansion based on citizens’ affinity to Bella&Brava’s brand promise and their desire to eat pizza.
  • Minimize business expansion risks by recognizing local and cultural circumstances that could affect the business.
  • Through qualitative image analysis methods, evaluate if and how the menu should be adapted for local markets and preferences.

A no-barriers approach to business expansion

Image recognition technology has gained more popularity with brands and agencies in recent years. The internet is becoming more visual by the minute, with ten percent of pictures in the entire history of photography taken in the last year. More than 95 million images are shared daily on Instagram.

Talkwalker’s proprietary image recognition technology detects brand logos even if the brand isn’t referenced in the accompanying text. It’s also capable of finding and analyzing images based on objects or scenes within them. Because audiences use social media to connect and interact - with each other, brands, and products - brands can make use of these unsolicited conversations and pictures to base decisions on the true voice of the consumer.

A single image containing a pizza might not be particularly insightful to a brand like Bella&Brava. However, when analyzing thousands or even millions of pizza images, a picture begins to emerge. Where do people eat it? What do they consume with it? What are their likes and dislikes? OpenKnowledge proposed using image recognition for their study as images found would bring qualitative insights at quantitative scale.

That’s where image recognition technology begins to open up opportunities - going beyond traditional market research. It’s one reason the company turned to social data to identify and categorize product consumption habits. Bonus - image analysis overcomes linguistic barriers and semantic specificities.

image recognition technology heat map

Talkwalker heat map showing where pizza images were published in Europe, excluding Italy and the UK.

A restaurant with a mission

Bella&Brava’s mission is rooted in the belief that there’s a universal right to a healthy diet. The restaurant offers only six varieties of pizza - each with carefully chosen, sustainable ingredients and recyclable packaging.

The restaurant worked with OpenKnowledge to make sure the study was generating meaningful insights based on their target audiences’ preferences. That’s to say, not all images would be treated with equal importance for the brands’ decision. To achieve this, OpenKnowledge worked on four keyword clusters, that brought together brand promise and customer values.

Cutting pizza

Eating pizza

Extracting insights step by step

As the initial database of images containing pizza was huge, OpenKnowledge took information from the user-generated images step by step.

Step 1 Developing keyword clusters

In line with Bella&Brava’s offer and market positioning, OpenKnowledge defined the four keyword clusters to ensure the recommendations were in line with the voice of the right audience. The four clusters were:

  • Healthy
  • Made in Italy
  • Vegetarian
  • Vegan

image recognition technology keywords

Fetta cheese pizza

Step 2 The visual content analysis funnel

Based on the keyword clusters, the company analyzed an initial 454,500 images containing pizza - posted between January and August 2017. The goal was to find European cities suitable for business expansion so the search was narrowed down to 66,400 images with an exact, street level geo-localization tag. The images were then grouped according to the city they were taken in. Cities with fewer than 50 results were discarded. Leaving 30,165 images from 82 cities.

image recognition technology content analysis funnel

Compatibility index

OpenKnowledge created an index, ranking each city on a scale from 0-100 for their desire to eat pizza. This was calculated comparing the volume of user-generated content with the number of active users on social media in each city. Next, the consultancy connected desire to eat with individual scores relating to the relevance of the four clusters.

They ended up with ten European cities (and 2156 images) that would be good candidates for business expansion.

Step 3 Contextual image analysis

To gain deeper, more contextual insights, OpenKnowledge further analyzed images containing pizza posted from the top ten cities that were identified in the index. They used Talkwalker’s proprietary image recognition technology to recognize brand logos, objects and scenes present within the pictures.

Adapting business expansion plans to consumer behavior

"In the digital age, there are two types of organisations: those that collect data and those that transform it into opportunities."
Ilaria Baietti, Director - Brand Interaction @ OpenKnowledge

Objective 1 - city ranking

In line with the study’s objectives, OpenKnowledge delivered ten top cities that had scored highly in the measurement study to Bella&Brava. Some were surprising but strong candidates the company wouldn’t have taken into consideration otherwise.

Objective 2 - mitigating business risks

Based on a qualitative and quantitative analysis of the images, OpenKnowledge found several local and cultural circumstances that could affect the business expansion.

  • Where getting a terrace is essential. 72% of pizza is consumed in bars and restaurants. But in some cities like Nice, there is an above average level of consumption in outdoor areas (+38%) due to socio-demographic variables - e.g. climate, tourist flows, number of events in the city, etc.

  • International footprint vs local heroes. Research showed that some cities had attracted different business models. Cities like La Valletta or Nice are home to small, "boutique" restaurants that communicate their Italian roots and family heritage. While, places such as Leipzig and Mainz boast several franchises (Vapiano, L'Osteria), which come with a more international audience - an important decision factor for Bella&Brava.

Objective 3 - menu options

The company also evaluated if and how the menu should be adapted for local markets and preferences, which returned the following audience insights.

  • ‘Healthy’ is flexible. Analysis showed that what’s considered “healthy” food is interpreted differently in every country. Consumers in cities with traditionally calorie-heavy food cultures tend to see traditional Italian ingredients -buffalo mozzarella, parma ham - as light and healthy. Something the company hadn’t foreseen and which could influence their local menu significantly.

  • ‘Made in Italy’ doesn’t mean the same everywhere. Their Italian heritage is one of the cherished pillars of the Bella&Brava brand. It was crucial for the company to evaluate how this would play out in different markets. Many images in different European countries showed non-traditional pizza ingredients - avocado, pumpkin, peas, corn, peach, etc. - but were still perceived as typically ‘Made in Italy’. This was new to the company and could be used to their advantage.

  • Putting the right drink choices on the menu. Without any filtering restrictions, the beverage brands most associated with all pizza images were Coca-Cola and Pepsi. However, when looking at images in the four Bella&Brava clusters, results showed that people who consume healthier pizzas tend to favor San Pellegrino mineral water. In some cities, cocktails are the favored accompaniment of pizza - typically, Italian Aperol Spritz. Should these locations be chosen for expansion, an Aperol Spritz would be a great addition to the drinks menu.

Takeaway? Find strategic actions from image recognition technology insights

The digital and social world is playing a decisive role in the business decisions of startups and big brands. With social listening, being one of the most time and cost effective tools to collect consumer insights.

Pietro Peccenini, CEO of Bella&Brava, stressed the importance of integrating digital methods and new data sources into their decision making process, in order to stay one step ahead of the competition.

Based on this data, his team set out to evaluate the proposed locations. Including several strong contenders that would not have been considered using traditional market research methods.