If you are exploring a new self-storage development opportunity, a feasibility analysis is one of the first real tests of whether the deal makes sense. The challenge is that good feasibility work takes time, and hiring a consultant can be expensive. A solid study requires current market data, competitive research, demand signals, access and visibility considerations, and a clear way to make sense of a lot of fragmented information.
Which begs the question, what if AI can just do it for you?
Across commercial real estate, investors, developers, and operators are already using AI to summarize market reports, compare properties, surface patterns in demographic and pricing data, and move faster through early-stage analysis. At the same time, a new wave of startups is emerging that offer agentic systems designed to handle pieces of the feasibility process that were once left almost entirely to traditional consultants.
That does not mean the consultant-led feasibility study is going away, at least not yet. For major development decisions, a more robust analysis is often still necessary. But it does mean owners and operators now have a practical way to get to a preliminary answer faster and at little to no cost.
In this article, we will focus on how to use a basic large language model, or LLM, to conduct a preliminary self-storage feasibility analysis using publicly available information. The goal is simple: help you pressure-test an opportunity, identify your biggest open questions, and decide whether the deal looks promising enough to justify investing in a deeper, more formal analysis.
What a self-storage feasibility analysis is really trying to answer
At its core, a feasibility analysis asks a simple question: could this piece of land support a new self-storage facility?
That answer usually depends on several smaller variables. Is local demand strong enough? Is the area already oversupplied? Are nearby facilities well positioned or vulnerable? Is the site easy to access and visible enough to attract tenants? Are there zoning, entitlement, or development constraints that could slow the project down?
Feasibility is not just one number in a spreadsheet. It is a collection of signals that, taken together, help you decide whether to keep moving, slow down, or walk away.
Start with the right inputs
The biggest mistake people make with AI is assuming it will fill in all the gaps for them. It will often do a decent job of pulling information from available sources on the fly, but it will also be prone to accessing outdated or unreliable information, or just making things up. Clean, structured inputs lead to better output, while incomplete or inconsistent source material increases the risk of weak conclusions.
For a first-pass feasibility analysis, it is best to gather verified information to feed into your prompt before starting the analysis, such as:
- parcel size and location
- population and household trends in the trade area
- median income, renter mix, and housing growth
- nearby self-storage facilities and their locations
- advertised rates, promotions, and amenities
- public reviews that may reveal customer expectations or service gaps
- traffic counts, visibility notes, and access considerations
- zoning information or municipal development guidance
- any known pipeline supply or planned developments nearby
Compile as much of this information as you can into a single document that you can paste into the model when you are ready. As you build that research set, an LLM like ChatGPT or Gemini can also help you find and organize public information more quickly. Just make sure you verify important details against the underlying sources rather than trusting the summary at face value.
Structuring your analysis
A good approach is to treat AI like a research assistant.
Start by defining the exact business question. Instead of saying, “Analyze this site,” ask something tighter, such as: “Based on the market and site information below, how feasible does a new self-storage facility appear, and what risks should be validated manually?”
Next, paste in your source material in clearly labeled sections. Separate facts from assumptions. If some data is missing, note that directly.
Then ask AI for a structured response. For example, you might tell it to summarize:
- the strongest demand signals
- the biggest competitive risks
- any gaps in the data
- assumptions that should not be trusted yet
- the top three next steps before making a go or no-go call
If the answer comes back too long, ask for a tighter version. Better yet, ask the model to organize the output into a simple scorecard with columns like category, evidence, takeaway, and requires verification. That makes the result much easier to review and act on.
A few useful prompt ideas
You do not need fancy prompting to get useful output. You just need to be specific.
One prompt could ask for a market summary: review the demographic, competitor, and site data below and summarize the strongest indicators that support or weaken the case for a new self-storage facility.
Another could focus on competition: compare the nearby facilities based on price positioning, promotions, location convenience, online reputation, and likely customer appeal.
A third could focus on risk: based on the information below, what assumptions are too weak to rely on without manual validation?
The pattern matters more than the exact wording. Give the model a clear question, structured inputs, and instructions on how you want the answer organized. That is usually enough to get a useful first pass.
What AI still cannot do reliably
AI can help you process information faster, but it should not be treated as a substitute for local expertise or formal underwriting. It cannot visit the property, verify whether public data is current, predict lease-up with certainty, or account for every hyperlocal factor that may shape performance. It also should not be trusted to make high-impact decisions without human review.
That is why the best use of AI is early in the process. Let it help you get organized, identify where the opportunity looks promising, and surface what still needs to be checked. Then use your own judgment, local market knowledge, and professional advisors to pressure-test the conclusion.
Common mistakes to avoid
A few pitfalls come up again and again:
- trusting the summary without reviewing the source material
- feeding the model incomplete or inconsistent information
- asking for a conclusion before defining the business question
- mistaking a fast answer for a reliable one
- using AI to confirm a deal you already want to do
The goal is not to let AI tell you what to build. The goal is to challenge assumptions, spot blind spots earlier, and make better early-stage decisions with less wasted time.
Final thought
A self-storage feasibility analysis will never be completely automated, and it should not be. But AI can make the initial process faster, clearer and more flexible.
If you get a positive recommendation, that’s a strong signal that pursuing a self-storage project is viable and worthy of further due diligence. In the case of a negative recommendation, you probably saved yourself time, hassle, and paying a consultant a lot of money to get the same answer.
For a practical roadmap for putting AI to work in your business, download Storable’s e-book, A Practical Guide to AI Adoption for Self-Storage Operators.
