Why a PI intake call has three different clocks running in DC, Maryland, and Virginia

It's 9:14pm. The office has been dark for three hours. Your cell buzzes with the forwarded intake line, and the voice on the other end is doing the thing every after-hours caller does — starting in the middle. "...so it was back in March, my sister got rear-ended, and the other guy's insurance keeps calling her and we don't know what to..."

You write down the date. March something. You'll get the exact day tomorrow. And somewhere in that sentence, without anybody in the call noticing, a clock started running — one of three possible clocks, and which one it is has nothing to do with the fact that the sister lives in Silver Spring and is calling you from a Maryland cell number.

It has to do with one thing your intake didn't ask yet: where did the crash actually happen?

The one field that decides everything

Here's the part that catches solo and small PI firms practicing across the DMV, and it's worth saying plainly before anything else: this isn't legal advice, and nothing here substitutes for pulling the controlling statute on a specific matter. This is an intake-design piece — about the question your script asks, not the conclusion your script draws. Your script should never draw the conclusion. That's the whole point.

The statute of limitations on a personal-injury claim runs from jurisdiction-of-incident — the place the injury happened — not from where the client lives, not from where they're calling, not from where your office is. And in the DMV, those three things are almost never the same. Your caller lives in Maryland, works in DC, got rear-ended on the GW Parkway in Virginia, and dialed a firm whose office is in Arlington. Four jurisdictions in one sentence, and only one of them starts the clock.

So the clocks. Confirm each against the controlling statute for the specific claim before you rely on it, but the general PI civil limitations periods are:

• Virginia: two years. Every action for personal injury is brought within two years after the cause of action accrues (Va. Code § 8.01-243). Note the trap hiding in the same statute: injury to property in Virginia gets five years, not two. Same accident, two different clocks depending on whether you're talking about the person or the car.
• Maryland: three years. A civil action is filed within three years from the date it accrues unless a different period is specified (Md. Code, Cts. & Jud. Proc. § 5-101).
• DC: three years. For a personal-injury action (D.C. Code § 12-301(8)) — with the wrinkle that DC courts apply a discovery rule, so accrual can start not when the injury happened but when the client knew or reasonably should have known they were hurt and that someone else caused it.

Two years versus three is not a rounding difference. A Virginia incident has a full year less runway than a Maryland or DC one. If your intake note says "MD client, March accident" and nobody flagged that the crash was on I-495's Virginia stretch, you've just quietly filed that matter under the wrong clock in your own head — and the person most likely to catch the error is the one who won't look at the file for another six weeks.

What actually breaks when address and incident get conflated

Walk the failure through. It's not dramatic. That's what makes it dangerous.

The caller leads with where they live, because that's how humans talk. The intake note captures an address — the caller's address, because that's the field the script asked for. Incident location, if it's captured at all, lands in a free-text "tell me what happened" blob that nobody parses for a jurisdiction until much later.

Now the matter sits in the callback queue tagged, in everyone's mental model, as a Maryland case, because the client is a Maryland client. Three years feels like plenty. Except the wreck was in Virginia, the real clock is two years, and the gap between "felt like plenty" and "was actually tight" is the kind that doesn't announce itself until it's a problem.

The fix is a script change, not a judgment call: jurisdiction-of-incident is its own field, captured on its own, distinct from caller address, on every PI intake. Not "where do you live" with incident location inferred. Two separate questions, two separate fields:

1. Where can we reach you? (caller address — for the callback)
2. Where did the accident actually happen — what city or county, or the nearest road and cross-street? (jurisdiction-of-incident — for the clock)

If your current intake form or phone script only has one location field, that's the gap. That's the whole audit. Go look right now: does your intake capture where it happened as a field an attorney can see at a glance, separate from where the caller is? If a rushed after-hours note could conflate the two, it will, on the night it matters most.

And a related discipline: capture the incident date precisely, or capture "caller unsure, needs confirming" precisely — because a discovery-rule question in DC, or a wrongful-death claim, or a claim involving a minor, or a matter against a government entity with its own notice deadline, can run on a different clock than the general PI period. The intake step's job isn't to resolve any of that. Its job is to capture the facts cleanly enough that the attorney can. Incident location, incident date, claim type, who else is involved. Facts, not conclusions.

The honest part about how software fits here

I'll say the thing a vendor usually buries, up front: an intake tool should not tell your caller which statute applies to their accident, and ours is scripted so it never tries. That's not a limitation we're apologizing for — it's the correct behavior. The moment a phone system starts giving a caller a read on their own deadline, you've got a problem no feature list makes up for.

What Dialkeep is built to do is narrower and, on a 9pm intake call, more useful: pick up, and ask the questions in the right shape — case type, the incident date, the incident jurisdiction as a field distinct from the caller's address, urgency, and how to reach them back. Asking where did it actually happen out loud, on the call, while the caller can still picture the cross-street, is the entire value here. A voicemail never asks anybody anything. It's scripted to never give legal advice, either — that's a hard rule, not a judgment it makes on the fly.

Where that ends, and it ends earlier than a vendor would normally admit: the answers stay in the call. There's no packet at the end of it, no summary, no transfer to a lawyer mid-call, nothing that puts those separated facts in front of you afterward — that's not built. So the discipline this whole piece is about still has to live in your script and your form, which is the point today rather than the sale. (Dialkeep isn't live yet — the demo line launches soon; the field has to exist whether a human or a script is the one asking.) If you're also weighing a metered virtual-receptionist plan against a flat one for a PI intake line that spikes unpredictably, I worked through that exact overage-month math for a family-law caseload here — the billing mechanics translate directly: Smith.ai vs. Dialkeep for a solo DMV family-law practice.

That's the honest fit: it doesn't draw the conclusion, it just refuses to lose the fact that lets you draw it.

The bottom line

Three jurisdictions, three clocks — two years in Virginia, three in Maryland, three in DC with a discovery-rule wrinkle — and the one running on any given call is decided by where the injury happened, not where the caller lives, works, or dials from. That means your intake has exactly one job it cannot get wrong at 9pm: capture jurisdiction-of-incident as its own field, separate from caller address, on every single PI call. Audit your script for that gap tonight. And confirm the controlling statute for the specific claim before you rely on any number here — because the reason intake captures facts instead of conclusions is that the conclusion is the attorney's, and the DMV makes sure of it.

For the broader math on what a missed after-hours call actually costs a small firm, see how many calls a small operation actually misses.