Seven Things a Retail Owner Should Not Automate

We spend most of our time here making the case for these tools. This piece is the other half, and it exists because "keep a human in the loop" is advice so general it changes nobody's behavior.

Here is the specific list.

1. The judgment calls inside CAM

Automate the extraction. Automate the invoice classification. Do not automate the calls.

Whether a $40,000 parking lot repair is a capital improvement or an operating repair is a question with money attached and a lease argument on the other side. Whether a management fee survives a particular exclusion clause is the same. CAM is a recurring source of landlord-tenant disputes precisely because the provisions are ambiguous, and ambiguity is not something a model resolves. It is something a model papers over.

The right output is a flag: these two leases treat this expense differently, here is the language from each. Then you decide.

2. Anything touching co-tenancy

Co-tenancy is among the most heavily negotiated provisions in a retail lease, and the negotiation reflects real leverage between you and a tenant with drawing power.

The status of a co-tenancy clause depends on the drafted condition, your current rent roll, and a calendar, and the remedy usually escalates in stages. Getting it wrong in your tenant's favor costs rent. Getting it wrong in your own favor invites a fight you will probably lose, because they have the clause too.

Extract the language. Track the condition. Never let anything auto-conclude that a tenant is or is not entitled to reduced rent.

3. The decision to decline a guarantor

If a consumer report on an individual guarantor contributes to declining a deal or changing its terms, you have obligations, including notifying that person. The obligation is yours, not your vendor's.

More to the point: the direction of regulatory thinking, clearest in the residential context but sound generally, is that you do not shed responsibility by pointing at a third party's algorithm. If a score declines someone, you declined them.

Score as an input. Decide as a person. Write down why.

4. First contact with a tenant who is struggling

A tenant who has missed a payment or whose sales are sliding is a business problem with a person attached. Automated collection sequences are efficient at converting a solvable situation into a vacancy.

The economics favor the conversation. Re-tenanting a space costs downtime, TI, commissions, and often a worse deal than the one you were trying to enforce. That calculation almost never appears in the tool that sends the notice.

5. Anything a tenant will read as coming from you personally

Renewal proposals, responses to a complaint, notices that affect someone's business. Retail at this scale is a relationship business, and tenants can tell.

This is not sentiment. A tenant who feels handled rather than dealt with negotiates differently at renewal, and you will pay for the efficiency later at a worse rate than you saved.

6. The final read on an acquisition

Use every tool available to accelerate diligence. Parse the rent roll, spread the T-12, extract the leases, surface the anomalies. Then read the leases yourself, at least the anchor and the three largest inline tenants.

The failure mode is not that extraction misses a date. It is that it produces a clean, confident summary of a lease with a problem in it, and the summary is so tidy that nobody opens the original. Deals are won and lost on exactly the clauses that are hardest to extract.

7. Your own understanding of the portfolio

The subtlest one, and the one nobody warns about.

Every system you install moves knowledge out of your head and into a tool. Mostly that is good, because knowledge in one person's head is a real operational risk. But there is a threshold past which you stop knowing your own centers and start querying them, and owners who cross it lose something they do not notice losing until a negotiation goes badly.

The tell is when you cannot answer a question about your own asset without opening software. Keep enough of it in your head to argue.

What this list has in common

Every item is a place where the cost of being wrong is asymmetric, where the counterparty is a person with their own interests, or where ambiguity is the actual subject matter.

The pattern is simple and it holds up better than any general rule: automate the reading, keep the deciding. Retail ownership involves an enormous amount of reading, and almost all of it can be handed over. The deciding is the job.

If you are automating something and cannot say who made the decision, you have gone past the line.