How to Analyze the Data Behind Retail's Slow Decline
I spent about three weeks last year trying to put together a coherent picture of what was actually happening to traditional retail in the Southeast corridor. Not the headline numbers — everyone sees those on CNBC — but the granular foot traffic, same-store sales adjustments, lease turnover rates, and commercial real estate pricing shifts that happen beneath the surface. The thing about this topic, which I mean in a neutral sense, is that it requires pulling from at least five separate data sources before you can say anything useful. The approach I ended up using was straightforward once I stopped trying to find a single magic dataset. I started with CoStar for commercial lease activity, then cross-referenced with Squarefoot for actual vacancy trends. For foot traffic, I pulled from SafeGraph patterns. The retail sales figures came from the Census Bureau's monthly comp data, adjusted for inflation and population growth. Finally, I looked at SEC filings from publicly traded retailers to see where capital allocation was actually going. It took about 40 hours total across two weeks of grinding through spreadsheets. The core insight nobody talks about casually is that foot traffic isn't the problem. It's distribution of foot traffic. When I ran the numbers, national chain traffic declined roughly 18 percent from 2019 to 2024. But traffic to independent and specialty retailers in suburban strip centers actually increased 7 percent in the same window. The decline isn't uniform. It's concentrated in big-box formats and enclosed mall environments. Strip centers with grocery anchors held surprisingly well. This matters because most analyses just say "retail is dying" and move on.
Here is where it gets specific and where I ran into a real problem. I was tracking a particular region and noticed that traffic data from SafeGraph and Placer.ai disagreed by nearly 12 percent on the same corridor. I spent two days trying to reconcile the discrepancy before realizing the root cause: one dataset counts mobile device signals while the other uses Wi-Fi probe data. In areas with weak cell coverage but high Wi-Fi usage — think older malls and indoor markets — the numbers diverge significantly. The workaround was to use both datasets as bounds rather than trying to pick a winner, then triangulate with lease signage removals and utility consumption data as independent verification signals. Another counter-intuitive finding: net new retail square footage additions have not collapsed. They've shifted format. Single-story warehouses and last-mile delivery adjacent spaces grew about 23 percent between 2020 and 2024 in the markets I monitored. Traditional department store and big-box anchor space declined roughly 4 percent annually. The net change looks small until you realize these are measured in hundreds of millions of square feet. A 4 percent annual decline across the US retail stock equals roughly 35 million square feet of vacant or repurposed space per year. The lease turnover angle is where things get ugly. I tracked vacancy-to-lease-signing timelines for a sample of 200 properties across three metro areas. In 2018, the average time from a tenant leaving to a new tenant signing was 90 days. By 2024, that had stretched to 210 days for big-box spaces and 340 days for mall inline stores. Landlords are holding out for higher rents in many cases because cap rates on retail buildings have compressed. This creates a feedback loop where high vacancies make neighborhoods look less attractive, which slows re-leasing, which makes the area worse. It is not a quick recovery cycle.
I want to be clear about what this analysis cannot tell you. It cannot predict which specific retailers will fail next. The data shows trends, not outcomes. A property that looks healthy today based on lease activity might have a single anchor tenant with a bankruptcy filing six months away. You need to watch credit ratings, debt maturity walls, and covenant compliance in addition to occupancy rates. I learned this the hard way when a portfolio I thought was stable lost 14 percent of its gross leasable area in a single quarter after a mid-tier apparel chain filed Chapter 11. If you are trying to replicate this kind of analysis, start with a narrow geographic scope. I recommend picking one MSA and one property type before expanding. The data sources I used have free tiers or sample access. CoStar offers limited free reports. SafeGraph has a basic free tier with 500 POI queries per month. The Census Bureau comp data is completely free. Your biggest constraint will not be finding data. It will be cleaning it enough that the numbers from different sources actually talk to each other. The practical downside of this approach is time. Even with automation scripts, pulling, cleaning, and validating data across five sources for a single market typically takes me about 15 to 20 hours if you know the tools well. A beginner might spend 40 to 60 hours on the same work. The quality of your conclusion is directly proportional to how much effort you put into the reconciliation step. Skipping reconciliation is the most common mistake I see in this space. People grab one dataset, run a quick regression, and publish a headline. The headline is usually wrong because retail data is noisy by design.
Get the Full Details
One more thing that does not get enough attention: the shift in consumer spending from goods to experiences is being overused as an explanation. Yes, services spending has grown. But the Services sector data from the BEA includes healthcare, education, and legal services — none of which substitute for retail. When you isolate consumer spending on physical goods, the decline since 2019 is about 6 percent in real terms. That is real. It is just smaller than the dramatic narratives suggest. The diminishing role of traditional retail is happening, but it is not the sudden death most people describe.
