Two states where we refuse to give you one number
For forty-six of the forty-eight jurisdictions we monitor, the question "do my Amazon sales count toward my own registration threshold here?" has an answer.
For Massachusetts and Louisiana, it has two.
We show you both. Two numbers, two verdicts, two crossing dates, side by side, with the disagreement named. This page explains why that is the honest output rather than a cop-out, because it is the design decision this product gets challenged on most.
The two states are not disputed for the same reason, and the difference matters: Louisiana is two credible authorities in genuine conflict. Massachusetts is one source we stand behind and a set of guides we will not.
What "disputed" means here, precisely
It does not mean we are unsure what the rule says. It means the available sources say different things, and no amount of care on our part resolves that. Here is each case.
Massachusetts
The reference chart we seed every rule from records marketplace-facilitated sales as excluded from the seller's own threshold. Other published guides describe that exclusion as conditional, and do not agree with each other about what the condition is.
We do not cite those other guides, and the reason matters: they sell competing services, and they disagree with one another on exactly this column. Naming one of them as corroboration would be borrowing authority we do not think they have here.
So Massachusetts is the weaker of the two disputes, and we say so rather than letting the word "disputed" imply more symmetry than exists. Our own second-sourcing pass on 2026-08-03 found a second source that agreed the exclusion applies, subject to a condition it stated cleanly. We kept the row disputed anyway, because a condition that other guides state differently is exactly the situation where quietly picking one reading is how a merchant gets a confident wrong answer.
Massachusetts' rule row carries a verified date of 2026-07-31; Louisiana's carries 2026-08-03. Each state's own date is shown beside its verdict in the app, and they are not the same date.
Louisiana
Until 2026-08-03 we had Louisiana down as counting marketplace sales. A second-sourcing pass across all forty-eight jurisdictions found that the Louisiana Sales and Use Tax Commission for Remote Sellers, which is the body that actually administers remote seller registration, says a remote seller "should only consider direct sales for delivery into Louisiana" and that marketplace sales "may be excluded". Our reference chart says include.
We moved Louisiana to the Commission's reading, because it is the controlling authority. That change moved us off the over-warning side, which matters more than it sounds and is the subject of the next section.
And we do not follow that same page on everything, which you should know before you click it. The Commission's FAQ still lists "200 or more separate transactions" as a live threshold. Two independent sources record that test as repealed on 2023-08-01, so our rule table does not apply it, and a merchant reading that FAQ will find a test we tell them does not exist. We follow the Commission on marketplace treatment, where it is the administering authority speaking about its own registration rule, and not on a transaction line the evidence says is stale. Saying which parts of a source we take and which we do not is the same discipline as showing the fork.
That pass checked every dollar threshold, transaction test and combinator across all forty-eight. Every one survived unchanged. Exactly one marketplace flag did not.
Why we do not just pick
The obvious product decision is to choose the more cautious reading and show one number. It is cleaner, it is easier to explain, and it is what we would do if the two mistakes cost the same.
Both mistakes cost real money, in different currencies, and neither one is the safe default. This is the same sentence the app puts beside every disputed verdict, and it is worth stating in full:
Registering in a state you did not owe buys permanent filing obligations there, including returns for months you sell nothing. Not registering in a state you did can mean back tax, interest and penalties, and where no return was ever filed, some states have no lookback limit at all.
Take each side seriously for a moment.
Over-registering is not a harmless precaution. It is returns due on a schedule in a state where you might sell four thousand dollars a year, penalties for missing them, and an unwinding process that is real work. A tool that resolves every ambiguity toward "you have probably crossed" is not being careful with you. It is being careful with itself and sending you the bill.
Under-registering is not harmless either. Quietly picking the exclude reading would tell a seller with $400,000 of Amazon sales into Massachusetts that they are approaching a line they may already be well past, and the bill for that arrives later and larger, with interest on it.
There is no safe direction to guess in. So we do not guess, we show the fork.
What this actually looks like
A seller with $50,000 direct and $400,000 through Amazon, all shipped to Massachusetts, against a $100,000 threshold measured over the previous or the current calendar year, whichever triggers first. The period is not decoration: it is half the rule, and it is the part most charts leave out.
| Total into Massachusetts | Counts toward the threshold | Verdict | |
|---|---|---|---|
| Marketplace excluded (our default) | $450,000 | $50,000 | Approaching (50%) |
| Marketplace included (the other reading) | $450,000 | $450,000 | Threshold met |
Those two verdict words are the ones the app actually prints, and they are deliberate. It says "Threshold met" rather than "you have nexus", because whether you have an obligation is a legal conclusion your accountant draws, not one a screen announces. It says "Monitoring" rather than "no nexus", because we watch 48 jurisdictions against the history we can see and cannot certify the absence of an obligation.
Notice also that the excluded reading does not read as comfortable. At $50,000 against $100,000 it is halfway there and says so. The gap between the two rows is not "fine" versus "not fine", it is "halfway" versus "past it", which is a more useful thing to take to an advisor.
Both rows are rendered. Both verdicts are rendered. Each reading's crossing, where that reading has one, is dated on its own terms; a reading that has not crossed shows no date rather than a borrowed one.
This is not a footnote on a screen that has already picked. It was, once, in pre-release review and before any merchant ever saw it. An early version of the engine showed the two dollar figures side by side and then computed the verdict, the progress band, the map colour and the crossing date from one reading only. That is a worse failure than showing one number honestly, because it looks like disclosure while quietly deciding the thing the merchant actually acts on. A reviewer caught it, and the engine now re-runs every measurement window and the entire crossing replay with the marketplace flag flipped, and returns both.
One precision, since this page is about not overstating: the map is the exception, and deliberately so. A state gets one fill colour, from the default reading, because a single tile cannot honestly show two verdicts at once. A disputed state carries a "sources disagree" marker instead, and the two readings appear in full the moment you open it. The map is an index, not the answer.
The disagreement is shown, not just the numbers
Next to those two rows, the app renders what the disagreement is: who says what, which source we default to, and why. For Louisiana that includes the Commission's own wording. For Massachusetts it includes the fact that we decline to name a second source and our reason for declining.
You should not have to take our word that a state is contested. You should be able to see the shape of the argument and take it to someone who can settle it.
What we are asking you to do with two numbers
Not to average them. Not to pick the one you prefer.
Ask the state. The Massachusetts Department of Revenue and the Louisiana Sales and Use Tax Commission are the only authorities that can settle their own rules, and both answer questions from sellers. Two numbers and a named disagreement is exactly what makes that a short conversation instead of a vague one: you are not asking "do I owe you tax", you are asking one specific question about one column.
Ask them before you register, and before you decide not to. Both of those are decisions, and the second one is a decision too, even though it feels like doing nothing.
The other forty-six
Worth saying plainly, so this page does not leave the wrong impression: forty-six of the forty-eight are not like this. Their treatment is recorded, sourced, dated, and rendered as one answer with its citation, because that is what the evidence supports.
Twenty-eight of the forty-eight count marketplace sales toward your own threshold. Twenty do not. Two of those twenty are the ones on this page, and they render both ways.
A single confident number that is wrong is worse than two honest ones, because you cannot act on what you cannot see. But two numbers where one would do is its own kind of failure, and we do not do that either.
Sources
- Marketplace treatment for all 48 monitored jurisdictions: Sales Tax Institute, Economic Nexus State Guide (chart dated 2026-05-04), verified 2026-07-31 and re-verified against a second source 2026-08-03.
https://www.salestaxinstitute.com/resources/economic-nexus-state-guide - Louisiana Sales and Use Tax Commission for Remote Sellers, quoted for the "direct sales" and "may be excluded" wording.
https://remotesellers.louisiana.gov/FAQ - Massachusetts Department of Revenue is named as the authority that can settle its own reading; no second published source is cited for Massachusetts, deliberately, for the reason given above.
We show you the arithmetic and the rule behind it so you can have an informed conversation with someone qualified to give you advice. The not-tax-advice notice below is the full version and it is rendered on every page here, so no article can be published without it.