Project 03 · Applied · Greensboro Convention & Visitors Bureau

Diagnosing an 89% drop in Instagram reach

Impressions had collapsed year over year, and the easy conclusion (“the content got worse”) turned out to be wrong. The data told a different, more useful story.

The problem

On paper, the Bureau's Instagram looked like it was falling apart. Comparing the same window year over year (January 1 to April 17), impressions were down 89% and total engagements were down 83%. The instinct in the room was that the content had simply gotten worse. Before anyone rebuilt the content strategy around that assumption, I wanted to check whether it was actually true.

How I approached it

I pulled the post-level exports from Sprout Social and rebuilt the year-over-year comparison from the ground up, not just the headline impressions number but the metrics underneath it: posting volume, engagements, engagement rate, and impressions per post. The goal was to separate two very different explanations that produce the same drop in reach: a content problem (people see it and don't engage) versus a distribution problem (fewer people see it at all).

The finding

The bottom row changed the whole story. Engagement rate hadn't fallen at all. It had improved 58%. The people who saw the 2026 posts were actually engaging with them at a higher clip than the year before. And the 89% impression drop wasn't explained by posting less, either: impressions per post were down 85%, so each post was reaching a fraction of the audience it used to.

The content wasn't the problem. Distribution was. The account had stopped reaching new people; it just kept engaging the ones it still reached.

Digging into what changed, the clearest difference was the disappearance of collaboration posts. The 2025 breakout posts were collaboration-heavy Reels that borrowed a partner's audience and reached 70,000–200,000 accounts each. The 2026 posts were strong creatively (several cleared a 7–10% engagement rate), but without collaborators they were capped at the Bureau's own follower base.

Benchmarking it

To pressure-test that read, I benchmarked the account against two regional peers, Visit Raleigh and Visit Greenville SC. Greensboro had a far smaller follower base, but it was the only one of the three actually growing:

That reinforced the diagnosis: the audience and the content were healthy. What had gone missing was the distribution mechanism, the collaborations, that put the work in front of new people.

What I recommended

• Reintroduce a recurring collaboration series (a “New Business Alert 2.0”) to rebuild partner-driven reach.
• Keep Reels as the discovery format, since they were still the top engagement driver at ~340 engagements per post.
• Use carousels for saveable local guides and itineraries, and reserve photos for when the image or announcement genuinely carries it.
• Tighten hashtags to 3–5 per post focused on local discovery, and drop generic filler tags.
• Track collaboration lift and save/share rate as the real KPIs (reach borrowed and content worth passing along) instead of raw engagement counts.

What I took away

The number everyone reacts to is rarely the number that explains what's happening. Splitting reach from engagement rate, one metric measuring distribution and the other measuring quality, turned a panic about “bad content” into a specific, fixable problem about how the work was being distributed. The recommendation was cheaper and more confident because the diagnosis was right.

Deliverables

• Year-over-year Instagram performance report with executive summary
• Format and collaboration analysis (Reels, carousels, photos)
• Competitor benchmark vs. Visit Raleigh and Visit Greenville SC
• KPI framework and an 8-week collaboration pilot plan

Tools

• Sprout Social, Instagram Business Profile analytics
• Microsoft Excel, eMarketer & Socialinsider benchmark research

This was one strand of a broader internship. See the CRM rebuild case study for the rest.