How Semantic Networks Can Reveal Why a Social Media Campaign Wasn't Working: Case Study

This post describes a fictional case study of a brand that launched an ambitious sustainability campaign six months ago. They hired influencers, created extensive content, earned media coverage, and invested heavily in messaging around their environmental commitments. Social listening showed thousands of mentions. Sentiment analysis reported 73% positive. By traditional metrics, the campaign was succeeding.

Unfortunately, sales weren't moving. Sustainability-focused consumers—the target audience—weren't converting. Focus groups revealed skepticism despite the positive social sentiment. The marketing team couldn't understand the disconnect.

Fortunately, semantic networks can directly measure brand perceptions to understand how audiences react to a campaign. While data from this case study is fictional, it is based on previous projects to show what findings from this approach can look like.   

The Methodology: Mapping Brand Meaning from Social Discourse

We collected 12,847 social media posts mentioning the brand across three platforms (Instagram, Reddit, Twitter/X), focusing on posts that included substantive content (not just brand tags or brief mentions). From this discourse, we constructed semantic networks:

• Each unique concept mentioned in posts becomes a node

• When two concepts appear together in the same post, they're connected

• Network structure reveals how consumers collectively organize brand meaning

We then calculated network-based metrics (centrality scores and cluster analysis) to identify core brand associations and discourse themes

Finding #1: "Sustainability" Was Present But Isolated

The campaign successfully increased mentions of "sustainability”: the term appeared in 42% of brand-related posts—up from 8% pre-campaign. But centrality analysis told a different story:

• "Sustainability": Eigenvector centrality of 0.19 (low—poorly connected)

• "Convenient": Centrality of 0.71 (high—well-integrated)

• "Works": Centrality of 0.68

• "Affordable": Centrality of 0.61

Sustainability appeared frequently but remained peripheral. It was mentioned in posts but not integrated into how consumers actually understood the brand. Meanwhile, associations of convenience and quality dominated the semantic core.

Finding #2: Cluster Analysis Revealed "Greenwashing" Narratives

Community detection identified distinct thematic clusters in the semantic network:

Cluster 1: "Basic Convenience" (51% of brand associations)

Connected concepts: easy, convenient, affordable, available, practical

This cluster dominated brand discourse—consumers primarily discussed the brand in functional convenience terms.

Cluster 2: "Marketing Claims" (23% of brand associations)

Connected concepts: sustainability, greenwashing, marketing, campaign

Sustainability clustered together with skeptical concepts. This shows that Sustainability was discussed as marketing claims rather than authentic attributes.

Cluster 3: "Product Quality" (18% of brand associations)

Connected concepts: works, effective, good, reliable

Sustainability was not connected to functional brand attributes and heavily connected to skepticism concepts. Consumers saw sustainability messaging as separate from actual brand identity.

Implication: The campaign had successfully increased sustainability awareness without integrating sustainability into brand meaning. Consumers acknowledged the claims but organized their brand understanding around different concepts.

Intended vs. Actual Brand Positioning

The company's campaign strategy emphasized three positioning elements:

1. "Sustainability without compromise on convenience"

2. "Affordable eco-friendly alternatives"

3. "Real environmental impact, not just claims"

Semantic networks showed how each element landed:

Sustainability without compromise: The positioning intended to bridge sustainability and convenience. Networks showed these as entirely separate clusters with minimal connection. Consumers saw them as competing rather than complementary attributes.

Affordable eco-friendly: "Affordable" showed high centrality (0.68) but connected primarily to the convenience cluster, not the sustainability cluster.

Real impact not claims: This positioning tried to preempt greenwashing concerns. But semantic networks showed "claims," "marketing," and "greenwashing" as high-centrality concepts within Sustainability. The discourse was precisely what the positioning tried to avoid.

Overall Assessment: The campaign was being received as greenwashing. None of the intended positioning elements landed as planned and the campaign created brand meaning opposite to strategic intent. Sustainability claims existed as a separate narrative about marketing rather than integrated into authentic brand meaning.

Strategic Recommendations

Semantic network analysis shows that a strategic reboot is needed: sustainability should be paired with core concepts like  “convenience”, “affordable, and "works" to build off associations that the brand already owns.

·      Recommendation: Position sustainability as product improvement, not ethical add-on.

·      Positioning Example: "Sustainable ingredients that work better", “A sustainable product you could rely on”, “Being environmentally sustainable has never been so easy”

Learn More

This case study demonstrates how semantic network analysis can be applied to social media discourse to reveal brand perceptions often overlooked in traditional social listening. While standard tools count mentions and calculate sentiment, network analysis reveals meaning structure—whether associations are central or peripheral, integrated or fragmented, coherent or confused.

See our white paper for a detailed description of our methodology, additional strategic applications, and how semantic networks transform both survey data and social listening into quantified brand intelligence.

Contact HDI to discuss how semantic network analysis can reveal positioning insights hidden in your social media data. Visit hauptdatainsights.com or email Michael@hauptdatainsights.com.