Custom garden center and nursery software in Baltimore: the plant you may still sell and may no longer buy

Maryland has 145 garden centers, and not one of them is in Baltimore

There is a version of this article that opens with the global horticulture software market and a compound annual growth rate quoted to two decimal places. It would tell an independent garden center on the edge of Baltimore County precisely nothing. So here is the local picture instead, counted rather than modeled, taken from the Census Bureau’s County Business Patterns file for 2023 — still the most recent release with county detail — which we downloaded whole and filtered ourselves on 6 September 2026.

NAICS 444220 is the code the Census Bureau uses for nursery, garden center and farm supply stores. In Maryland it counts 145 establishments with paid employees, 2,115 people and $91,502,000 of annual payroll. That works out at 14.6 employees per store, which is a strikingly large average for a trade this series usually finds atomized — more than double the 6.9 of the average Maryland meat market and the 6.4 of the average flower shop, both counted from the same file. Garden centers are big-footprint, staff-heavy retail, and the reason is the same reason everything else about them is hard: for a few weeks in spring they have to be able to serve a queue.

The file lists fifteen Maryland jurisdictions with at least one such store. Baltimore City is not among them.

The usual caveat applies and it matters more here than in most trades: County Business Patterns counts only establishments with paid employees. A one‑person nursery selling from a gravel lot on a county road is invisible to this file, and there are certainly some. But the absence of Baltimore City is not a rounding artifact of that caveat. The city has plenty of businesses with employees; it has seven fish and seafood retailers and thirteen florists in the same file. It has no garden center of this kind at all.

So the geography of this trade is a ring. Take the five jurisdictions that actually serve a Baltimore customer within a reasonable drive — Anne Arundel, Baltimore County, Carroll, Harford and Howard — and you get 60 of the state’s 145 stores, 41.4% of the count, 860 employees and $38,767,000 of payroll, which is 42.4% of the state total. Two out of every five Maryland garden centers sit in the ring around a city that has none, and their customers drive out to them. That single fact shapes everything downstream: the catchment is regional rather than neighborhood, the trip is planned rather than impulsive, the basket is large and awkward, and a meaningful share of the demand could have been a delivery if anybody had built the software to make it one.

The quarter that has to pay for the other three

Everyone knows garden centers are seasonal. Almost nobody puts a number on it, because the obvious sources do not publish one at the county level. But the County Business Patterns file carries first‑quarter payroll alongside annual payroll for every industry and every county, and the ratio between them is a usable, honest seasonality measure that costs one division to compute. A business with a flat year would show a first quarter at 25.0% of annual payroll. Here is what Maryland retail actually shows.

A garden center’s first quarter carries 20.6% of its payroll, which is 17.7% below a level year, and 4.5 percentage points below the home center down the road selling many of the same bags of mulch. The three remaining quarters average 1.29× the first. Put plainly: for every dollar of wages a Maryland garden center pays in January, February and March, it pays about a dollar and twenty‑nine cents in each of the quarters that follow, and the true spring peak is sharper still, because April and May are averaged in this measure with a slow August.

That shape is the reason so much garden center software feels wrong. Almost every retail subscription in this article is priced per month, evenly, forever. The revenue that has to pay for it is not. And as we will see, at least one of the platforms most likely to be running the online half of the business prices on a trailing measure of sales, which means a good spring quietly buys you a more expensive following year.

The vendors, and where they are honest

We do this in every article in this series, and it is the least glamorous and most useful part. We took thirty‑five products that a Maryland garden center might plausibly be sold — specialist garden center point of sale, hardware and farm‑supply systems, grower and nursery production software, general retail platforms, e‑commerce platforms, plant tag suppliers and shipping tools — and opened each one’s pricing page from one machine with an ordinary desktop browser user agent on 6 September 2026, recording the HTTP status code and the URL each request finally landed on.

A status code is a small thing, but it is an original observation with a date on it, and it converts “the review sites say this costs about a hundred dollars” into something a buyer can check. It also catches things nobody advertises, like a company that no longer exists under the name you were given.

Three of those rows are worth a sentence each. Advanced Grower Solutions — a name that still appears in grower software round‑ups — sends every request to silverfern.com, a different company’s front door. If you are working from a list somebody emailed you, that is a live example of why you check. Rapid Garden POS, the single most frequently recommended garden‑center‑specific system in the review directories, blocked us; we cannot tell you what it costs, and neither can the aggregators, who mostly quote each other. And Rundoo, which markets a Lawn & Garden vertical explicitly, publishes no price at all but does publish a savings claim on the same page, naming a customer and a figure: Country Gardens & Nursery saved $11,000 by switching to Rundoo. A dollar figure with no denominator is a marketing number, not a price. It might well be true. It is not something you can put in a budget.

None of this is a criticism of the products. Enterprise software has always been sold by conversation and there are real reasons for it — configuration, hardware, payment processing, migration effort. But a garden center owner trying to plan a February software decision is entitled to notice that of thirty‑five products, twenty‑five will not tell them a number without a phone call.

What a counter and a storefront actually cost

Here is what the eight honest ones published, on that day, in US dollars.

From those numbers you can build two honest stacks. Neither is a recommendation and neither is complete — payment processing, hardware, an extra register and any add‑on module sit outside all of them — but both are floors made only of published prices, which is more than most quotes you will be shown.

Stack B costs $2,184 a year more than Stack A, which is 2.08× the price. It also includes five back‑office users where Stack A includes one register and says nothing published about a second, and it puts the till and the website in one database rather than two, which is worth real money in reconciliation time. We are not going to pretend the cheaper stack is obviously right. What we will say is that the difference between those two annual figures is roughly what a garden center pays one seasonal employee for six weeks, and that both of them are small next to what the next section is about to describe.

The plan that upgrades itself

BigCommerce’s card carries something that deserves more attention than it gets from a seasonal retailer, and it is stated plainly on the page rather than buried: Core is “up to $30K TTM GMV; plan auto‑upgrades to Growth,” Growth is “up to $100K TTM GMV; plan auto‑upgrades to Scale,” and Scale carries a cap of “$33,333/month GMV” with 0.9% charged on anything above it.

TTM means trailing twelve months. So the plan you are on is a function of the sales you already made, it changes without you doing anything, and the measurement window is a year long. For a business with a level year that is unremarkable. For a business with a first quarter running 20.6% of the year, it has a specific and slightly unpleasant consequence: the cost of a good spring is paid over the following twelve months, automatically, whether or not the next spring arrives.

The dollar at which trailing sales cross $30,000 costs $600 over the year that follows it. That is not a criticism of BigCommerce, whose disclosure here is better than most; it is an observation about what happens when a pricing model built for steady direct‑to‑consumer brands meets a business that earns most of its money in eight weeks. And note the third row, which is monthly rather than trailing: a store selling $400,000 online across a year sits within $4 of the annual equivalent of the Scale cap — twelve months at $33,333 is $399,996 — and still pays an overage, because in April and May it is far above $33,333 and in January it is far below, and the cap does not average. Distribute $400,000 across the year in the quarterly shape the Census file implies and you are over the monthly cap in nine months of twelve by $1,965 each, which is $17,687 of excess and $159 of extra fee. That is a small number. The reason to know it is not the size; it is that nothing in the sign‑up flow tells you that your calendar is being priced.

The ladder that stops going down

One more piece of vendor arithmetic, because it is a nice reminder to always divide. ShipStation prices in shipment bands rather than seats, and its higher tier runs from 50 shipments a month to 10,000. Divide each published price by the shipments it buys and the unit cost falls beautifully for seven rungs and then turns around.

The last rung costs 50% more per shipment than the one below it. There is very likely a sound commercial reason — support load, or a deliberate nudge toward a conversation at that volume — and a garden center will almost certainly never reach it. We include it because the habit is what matters: every published price in this article was divided by something before we believed it, and that habit found a $0.02 discontinuity nobody advertises.

The part no national platform models

Every article in this series arrives at the same place: somewhere under the trade there is a rule that decides what a record has to contain, and the national platform does not know it exists. Garden centers have the sharpest example we have found. It is not a labeling rule or a tax boundary. It is a rule that took a column that already existed in every plant catalog in Maryland and changed what the values in it mean.

A field that used to mean “print a sign” now means “start a clock”

Since 2016, Maryland classified invasive plants in two tiers. Tier 1 plants could not be propagated, imported, transferred, sold, purchased, transported or introduced at all. Tier 2 plants could be sold, but only if the retailer displayed required signage beside them, and a landscaper who planted one had to hand the customer the list. Three species were tier 1. A handful were tier 2. Every Maryland garden center that took the rule seriously encoded it the obvious way: a field on the plant record with three possible values — nothing, tier 1, tier 2 — where tier 1 meant do not stock it and tier 2 meant print a sign.

The Biodiversity and Agriculture Protection Act, passed in 2024, deleted that vocabulary. The replacement is in Agriculture Article §9.5–301.1, and it is worth reading as a migration script rather than as legislation, because that is what it is.

“(a) The regulations adopted under § 9.5–301 of this subtitle shall classify as prohibited invasive plants all terrestrial plants classified by regulation before January 1, 2024, as a tier 1 invasive plant. (b) Each terrestrial plant classified by regulation before January 1, 2024, as a tier 2 invasive plant shall be assessed in accordance with § 9.5–301 of this subtitle by December 31, 2025, and shall be: (1) If the results of the assessment determine the plant is an invasive plant, classified as a prohibited invasive plant …; or (2) If the results of the assessment determine the plant is not an invasive plant, placed on the Watch List.”Md. Code, Agriculture §9.5–301.1

Read what that does. Subsection (a) is a straight rename: every tier 1 value becomes prohibited, one to one, no judgment involved. Subsection (b) is not a rename at all. It is a fork. Every tier 2 value in the state had to be recomputed by a fixed date, and each one would land in one of two entirely different places — a list that ends the sale, or a list that only obliges you to hand somebody a piece of paper. From 1 January 2024, the tier 2 column in every catalog in Maryland held values whose meaning was scheduled to be replaced, and nobody could tell you in advance which replacement any given row would get.

The fork resolved on 15 January 2026, when the Maryland Department of Agriculture’s Consolidated List of Maryland Invasive Plant Species took effect with thirteen species on it: Japanese angelica tree (Aralia elata), Japanese barberry (Berberis thunbergii), Scotch broom (Cytisus scoparius), burning bush (Euonymus alatus), border privet (Ligustrum obtusifolium), nandina (Nandina domestica), golden bamboo (Phyllostachys aurea), yellow groove bamboo (Phyllostachys aureosulcata), Callery or “Bradford” pear (Pyrus calleryana), bee bee tree (Tetradium daniellii), Japanese wisteria (Wisteria floribunda), Chinese wisteria (Wisteria sinensis) and their hybrid (Wisteria × formosa).

Two of the best‑selling ornamental shrubs in the Mid‑Atlantic are on that list, and one of the most planted street trees of the last forty years. This is not an obscure rule about a plant nobody stocks. Burning bush and border privet were tier 2 under the old scheme, which means they stood on Maryland benches lawfully, with the required sign beside them, right up until the day they did not. Japanese barberry and Callery pear were never on the old tier 2 list at all; they reached the prohibited list through the new assessment process, with no signage era in between. Both routes end in the same place, and neither of them is visible in a product record that knows only a plant’s name and its price.

We should say plainly that we could not read MDA’s own list page from this machine. mda.maryland.gov returned 403 Forbidden to an ordinary desktop request on both 5 and 6 September 2026, through three separate clients. That is very likely a blanket block on automated traffic rather than anything a customer would hit in a browser, and the species list above comes from the Maryland Native Plant Coalition’s report of the announcement rather than from the Department. The regulatory text quoted throughout this article, by contrast, comes from the primary sources directly: the statute from the General Assembly’s own database and the regulations from the Library of Maryland Regulations. Check the Department’s list yourself before you act on any of this; it is the authoritative version and it is the one that changes.

The same plant, still legal to sell and already illegal to buy

Here is where it stops being a compliance chore and becomes a software problem. The phase‑out rules live in COMAR 15.06.04.06B, and they do not say “you have a year.” They say four different things, and two of them point in opposite directions on the same product record.

“(3) A person may not acquire a new prohibited invasive plant after that plant’s effective date on the prohibited invasive plant list. (4) Sales Limited to Existing Stock. During the phase‑out period, a person may only sell the existing stock of a prohibited invasive plant that person has in inventory as of the date the plant was placed on the prohibited invasive plant list.”COMAR 15.06.04.06B(3)–(4)

Acquisition stopped on 15 January 2026, with no phase‑out at all. Sale continued. So for the whole of the phase‑out window, a Callery pear on your lot is a product you may lawfully sell and may not lawfully reorder. Every point of sale we have ever worked on treats those as one property. There is a checkbox, and it is called active, or available, or discontinued, and turning it off stops both the sale and the purchase order because in every other retail trade on earth those two facts are the same fact.

They are not the same fact here, and the failure is silent in both directions. Leave the item active and your buyer can raise a purchase order for a plant it is unlawful to acquire — and the buyer will, because the sales report says it is still selling. Mark it discontinued and your staff cannot ring up the thirty perfectly lawful shrubs still standing on the bench, so they will ring them up as something else, which destroys the one record that could later prove the sale was lawful. There is no setting that produces the right behavior, because the shape of the rule is not the shape of the field.

The clock is set by the pot, not by the plant

The length of the window is decided by something your catalog almost certainly does not store, and the two sentences that decide it were not written to fit together.

“(1) A person in possession of a prohibited invasive plant, except a person who is growing a woody plant in ground, shall comply with the prohibited invasive plant provisions of this chapter within 1 year of that plant’s effective date on the prohibited invasive plant list. (2) A person who is growing a prohibited invasive plant in ground shall comply with the provisions of this chapter within 2 years from that plant’s effective date on the prohibited invasive plant list.”COMAR 15.06.04.06B(1)–(2)

Two barberries of the same species, the same age and the same supplier, standing four feet apart, have different deadlines if one is in a container and the other is growing in the ground. The discriminator is not an attribute of the plant. It is an attribute of how you happen to be holding it on a given morning, and it can change — a shrub lifted from the field into a pot in March has, on the face of the rule, just moved between two categories with a year between them.

So the field you need is not invasive status. It is a small composite: which list the species is on, the effective date of that listing, whether this particular unit is in a container or growing in ground, and whether it is woody. Four values, three of which live on the unit rather than on the species. No retail product catalog is built that way, because no other trade needs it.

Two of the thirteen sit on a line the regulation draws twice, in two different places

Read those two subsections again, side by side, and notice that they do not describe the same set of people.

Subsection (1) gives one year to everybody except “a person who is growing a woody plant in ground.” Subsection (2) gives two years to “a person who is growing a prohibited invasive plant in ground” — with no mention of woodiness at all. The exception in (1) is narrower than the grant in (2). Anyone growing a prohibited invasive plant that is not woody, in the ground, therefore falls inside both: not excepted from the one‑year rule, and squarely inside the two‑year one.

For twelve of the thirteen listed species this is a curiosity and nothing more, because they are unambiguously woody — shrubs, trees and woody vines. It matters for exactly two of them. Golden bamboo and yellow groove bamboo are Phyllostachys, and bamboo is a grass. Its culm is lignified and every gardener in Maryland would call it woody without hesitating, but “woody” is a botanical term, and in botany a grass is herbaceous however hard the stem gets.

So the botanical question does not merely change the deadline. It decides whether the regulation agrees with itself. If bamboo is woody, (1)’s exception applies, (2) governs, and an established in‑ground stand has until 15 January 2028. If bamboo is not woody, (1) says 15 January 2027 and (2) says 15 January 2028 for the same plant on the same lot, and a grower has to choose which sentence to obey.

We are not going to tell you which reading is right. That is a question for the Department, it is exactly the kind of question a licensed dealer should put in writing, and the answer is worth keeping. What we will point out is that whichever way it resolves, somebody has to type a boolean into a field called woody for a grass, and the person doing it should know that the answer is contested rather than obvious — because a year of lawful sales hangs on it. That is what it looks like when a legal category and a data type disagree.

There is one more lever, and it belongs in the same field. Subsection B(5) lets a licensed nursery, broker or dealer ask the Secretary to extend the phase‑out for a specific species, weighing the harm the plant causes against “the economic impact that the phase‑out period has on the affected person” — and B(5)(d) caps any extension at one further year. So the true outer edge of a lawful sale is not 15 January 2028 but potentially 15 January 2029, and only for a business that asked and was told yes. That is a per‑business, per‑species date which exists only in a letter. If it applies to you, it belongs in the database beside the species, not in a drawer.

“Existing stock as of the date” is a query, not a flag

Return to subsection (4), because it contains the single hardest requirement in this article. During the phase‑out you may sell “only…the existing stock…that person has in inventory as of the date the plant was placed on the list.”

That is not a status. It is a point‑in‑time balance. To know whether a sale in October 2027 is lawful, you have to be able to answer: how many units of this SKU did this business hold on 15 January 2026? — and then, implicitly, whether the units sold since then exceed it.

Almost no small‑retail point of sale can answer that question after the fact. Inventory counts in most systems are a single mutable number per item: you receive, you sell, the number moves, and last year’s value is gone unless the system happened to write an immutable movement log and unless you kept it. Some systems keep a transaction ledger from which the balance could in principle be reconstructed; many roll up or purge, and a plain stock‑count field carries no history at all. The retailers who will be able to evidence this in eighteen months are the ones who, without knowing why, happened to be on a system that records movements rather than balances.

It is worth being concrete about what a defensible answer looks like, because it is genuinely modest: a dated opening balance for each affected SKU, captured once; every subsequent sale of that SKU tied to it; and a running remainder that cannot go below zero without somebody being told. That is perhaps four database tables and an afternoon of thought, and it is the difference between a lawful sale you can prove and a lawful sale you merely believe in.

Your records may expire a year before your permission does

Now put two unrelated regulations side by side, which is where the sharpest finding in this article lives.

The Plant Dealer License, in COMAR 15.06.02.04B(3), requires a dealer to “maintain records for 1 year of plant acquisitions, including certificate of inspection or another document indicating consignor and state of origin.” One year. That is a floor, not a ceiling — nothing stops you keeping more — but it is the number a dealer designing a retention policy around the regulation would naturally pick, and it is the number a records‑destruction schedule would be set to.

The invasive phase‑out, meanwhile, gives an in‑ground woody plant a sale window running to 15 January 2028, and makes every sale in that window lawful only if the unit was in inventory on 15 January 2026.

So a dealer who keeps exactly what the licensing regulation asks for, and no more, may lawfully destroy its January 2026 acquisition records in January 2027 — and then spend the following twelve months making sales whose lawfulness depends on precisely those records. The retention floor is one year. The evidentiary need runs to two. Nothing in either regulation cross‑references the other, and nothing in any point of sale we have seen would flag the conflict, because the conflict is between two documents that live in different chapters of COMAR and were written for entirely different purposes.

The fix costs nothing and has to be decided before the deadline rather than after it: for the thirteen listed species, retain acquisition and inventory records until at least 15 January 2029 — the outer edge B(5)(d) allows — and write down why. A one‑line note in a retention policy is cheaper than reconstructing a stock position from bank statements.

The list that governs your catalog arrives once a year, by post

How is a garden center supposed to learn that a plant it stocks has been listed? The regulation answers that directly, and the answer is a small masterpiece of mismatch between the speed of the rule and the speed of the business.

“(1) The Secretary shall send the Prohibited Invasive Plant List and the Watch List with annual renewal letters and with each new application for a nursery, plant dealer, or plant broker license. (2) The Secretary shall post the Prohibited Invasive Plant List and the Watch List on the Department’s website.”COMAR 15.06.04.06E(1)–(2)

The push channel is an annual letter. The pull channel is a web page you have to remember to visit. Meanwhile the list can be amended at any time — regulation .06C says the Watch List “may be amended from time to time,” and .06D sets out a standing procedure by which anyone may petition to add or remove a species. Your catalog changes weekly. Your buying decisions are made in autumn for the following spring. The notification arrives once, in an envelope, at renewal.

This is the clearest possible argument for a very small piece of software, and it is the one we would build first for any client in this trade: something that holds your live plant list, checks it against the Department’s published list on a schedule, and tells a named human when a row changes. It is not clever. It is a scheduled job and a diff. It is also the only mechanism that closes the gap between an annual letter and a daily catalog, and no point of sale on the market ships it, because no point of sale on the market knows Maryland has a list.

The Watch List is a document you have to hand over

The other half of the fork — the plants the assessment determined were not invasive plants, and which went to the Watch List instead — carries an obligation of its own, and it attaches to a service rather than to a product.

“A person in the business of providing landscaping services who plants or supplies for planting a plant on the Watch List shall provide to the customer requesting the service the Watch List.”COMAR 15.06.04.06E(3)

Plenty of garden centers in the Baltimore ring also plant. The moment they do, a line item on a job — not a product on a shelf — triggers a document delivery, and the trigger is whether any plant on that job appears on a list maintained by a third party and revised without notice. In software terms that is a rule evaluated at quote time against an external reference, producing an attachment on the customer’s paperwork. In practice it is a folded sheet somebody remembers to put in the van, and the memory is the control.

Note also who the duty falls on. Not the nursery that grew it, not the broker who moved it, but the person providing the landscaping service — which, for a garden center with an install crew, is you, on a plant you may well have bought in. The obligation follows the service, and services are exactly what retail software models worst.

Three fields, and one of them changes its unit

Underneath the invasive rules sits something quieter and, for anyone building an online store, more immediately useful: Maryland has already written down what a plant record must contain. It is in COMAR 15.06.02.07, it has been there for years, and it reads like a schema because it is one.

What the label has to say

“Each piece of nursery stock offered for retail sale shall be accompanied by a label containing the following information… (a) Plant Name. This name may be the scientific name or an approved common name, or both. The variety name may also be on the label, but may not be used in lieu of the scientific or approved common name. (b) Height or caliper or grade or diameter if sold by this measure. (c) Price of each plant.”COMAR 15.06.02.07A(1)

Three fields. Every one of them is more interesting than it looks.

The name field has a precedence rule built into it. You may show the scientific name, or an approved common name, or both, and you may add the variety name — but the variety name may not stand in for the other two. That is a validation rule, and it is the exact rule that garden center marketing tends to break, because the variety name is the one customers recognize. A bench sign reading only Bloomerang or Endless Summer is a beautiful piece of merchandising and, on the face of the regulation, an incomplete label. In a product database that means the cultivar cannot be the display name; it has to be an attribute hanging off a record whose identity is the botanical name. Almost every plant catalog we have looked at has it the other way around, because the catalog was built by whoever writes the marketing copy.

The size field is stranger, and it is the one that breaks templates. It is not “size.” It is “height or caliper or grade or diameter if sold by this measure” — four different units, one of which (grade) is not a length at all, and a condition that makes the whole field dependent on how the plant is being sold. A shade tree is sold by caliper in inches. A shrub is sold by height in inches or by container size. A ball‑and‑burlap specimen is sold by root ball diameter. A liner is sold by grade. In a spreadsheet that is one column with four incompatible unit systems in it; in a well‑made product record it is a pair — a measure type and a value — with the measure type driving both the label and the way the item is priced. Shopify gives you a text field called size. It will hold anything, including nothing, and it will never once tell you that a row is wrong.

The price field is the one nobody expects, because we are used to thinking of price as commercial rather than regulatory. Here the price of each plant is part of a label whose contents are specified by the state. Which means the answer to “where does the price live?” is not “in the point of sale.” It lives in the point of sale and on a physical tag on the plant and, if you sell online, on a product page. Three surfaces, one number, and only one of them updates itself.

The lot sign, and the merchandising rule hiding inside it

The regulation offers a way out of tagging every plant individually, and the way out has a condition that quietly governs how you are allowed to arrange your benches.

“Nursery stock that is balled and burlapped, packaged, or in a container shall be grouped by identical size, variety, and cost into a bed or lot separated from similar beds or lots. Each bed or lot shall be marked with a permanent legible sign containing the information required by §A(1) of this regulation.”COMAR 15.06.02.07A(2)(a)

Identical size, and variety, and cost. All three. So the moment you mark down half a bench — the ordinary, sensible thing to do with stock that is getting leggy in July — the lot is no longer of identical cost, and the exemption that let you use one sign no longer covers it. You must either physically separate the marked‑down plants into their own lot with their own sign, or go back to labeling individually. A promotion is not just a price change. It is a lot split, and it has a physical component.

This is the kind of thing that makes garden center retail genuinely different from apparel or hardware, and it is why the “just put it on Square” advice quietly fails. Square will happily apply a 20% discount to a category. It has no concept of the bench, no concept of a lot, and no way to know that the discount it just applied has invalidated a sign.

Two further variants of the same record appear elsewhere in the chapter, and together they make the point that this is one product with several legally distinct renderings.

A garden center with a retail floor, a landscaper account book and a website is producing all five of those renderings from what ought to be one record. The state has been explicit that they are different. Almost every off‑the‑shelf catalog treats them as one product page with a print stylesheet.

What “location” means when the shop is a website

Two definitions in the same chapter are worth putting side by side, because between them they decide whether an online seller needs a license and how many.

“‘Dealer’ means a person, except a nurseryman or broker, who: (i) Advertises nursery stock for sale or installation; or (ii) Buys, collects, or otherwise acquires wild plants or nursery stock for the purpose of selling, planting, or distributing them.” … “‘Location’ means each place from which nursery stock is sold, offered for sale, or distributed.”COMAR 15.06.02.01B(2), B(6)

Advertising alone makes you a dealer. Not selling — advertising. And COMAR 15.06.02.04A then requires a Plant Dealer License “for each sales location,” with “location” defined as each place from which stock is sold, offered for sale, or distributed.

Those definitions were plainly written for physical yards, and they were written before a garden center could offer stock for sale from a phone. We are not going to tell you what the Department’s current position is on whether a website is a location, or whether a pop‑up stall at a spring festival is one, or whether a third‑party marketplace listing is one; that is a question to ask them, in writing, before you build. What we will say is that it is a real question with a per‑location annual fee attached to the answer, and that the phrase “offered for sale” is doing a lot of work in a sentence written in a different decade. If you are about to add plants to an online store, this is worth five minutes on the phone before it is worth five thousand dollars of build.

What it costs to be licensed, and where the fee schedule bends

The fees themselves are modest, and worth laying out because the structure is more interesting than the total. COMAR 15.06.02.06 charges a flat annual license — $150 for a Nursery Inspection Certificate, $150 for a Plant Dealer License, both per sales location, and $150 for a Plant Broker License — and then adds a field inspection fee that is banded by acreage up to ten acres and linear after that.

Two things fall out of that. The first is a small cliff at ten acres: the banded fee tops out at $40, and the moment you cross into the “more than 10 acres” band the charge becomes $5 for each acre “or part of any acre,” which for a 10.1‑acre operation is eleven parts and $55. One‑tenth of an acre raises the inspection fee by 37.5%. The second is that the $1,500 maximum binds at exactly 300 acres, so from 300 acres upward the marginal cost of another acre of nursery is zero. Neither number will change anybody’s business, but both are the sort of thing worth knowing before you file an affidavit attesting to your acreage, which is what .03B(2) and .04B(4) require you to do.

One more line from the same schedule matters to anyone shipping out of state: a phytosanitary certificate costs $25 including travel and $5 excluding travel. A five‑fold difference decided by whether an inspector has to come to you. If you are planning interstate shipping at any volume, batching inspections is worth real money, and that is a scheduling problem software can actually solve.

And the penalties, for completeness, because they are asymmetric in a way that is easy to misread. Violating an order of the Secretary carries a civil penalty “not to exceed $500 for each violation” under COMAR 15.06.04.05B, which sounds survivable. The part that is not survivable is Agriculture §9.5–303: on finding a prohibited invasive plant in violation, the Secretary may issue a written condemnation seizure order, mark or tag the plant conspicuously, and require disposal; if it is not disposed of, the Secretary destroys it, prepares a statement of the expense of destruction, and sends it to the Attorney General to collect. The fine is $500. The exposure is your inventory, plus the bill for destroying it.

The online store is where all of this lands

We build online stores for a living, so treat what follows as interested. But the case for a garden center owning its storefront rather than renting a generic one does not rest on our opinion. It rests on the fact that every rule in the previous three sections is evaluated at a moment that a generic storefront does not model: the moment somebody puts a plant in a cart and tells you where to send it.

A cart is a place, and Maryland has an opinion about it

The prohibition in Agriculture §9.5–302(b)(1) is not written as a rule about Maryland sellers. It is written as a rule about Maryland.

“Except as provided in paragraphs (2) and (3) of this subsection, a person may not propagate, import, transfer, sell, purchase, transport, or introduce any living part of a prohibited invasive plant in the State.”Md. Code, Agriculture §9.5–302(b)(1)

Import and introduce. So the question a checkout has to answer is not “am I a Maryland business?” but “is this box going to a Maryland address?” A Pennsylvania nursery with a nice Shopify theme selling a Callery pear to a customer in Towson is shipping a plant into a state where introducing it is prohibited, and the platform will process that order without a murmur, because the platform’s model of a destination is a tax jurisdiction and a shipping zone, not a plant‑health jurisdiction.

Read the other direction too, and it gets sharper for anyone in the ring around Baltimore. A Maryland garden center with an online store is selling into Virginia, Pennsylvania, Delaware and the District, each of which regulates the movement and sale of nursery stock on its own terms and its own timetable. There is no reason those timetables should agree with Maryland’s, and no reason to assume they do. The correct behavior at checkout is a lookup on the pair (species, ship‑to state), resolved at the moment of the order, against lists that change without warning. That is a small, well‑understood piece of software. It is also one that no e‑commerce platform ships, and one that no theme can fake, because it has to fail the order rather than decorate it.

And there is a version of the failure that costs more than a refund. Under COMAR 15.06.02.07B, nursery stock grown outside Maryland and shipped in must meet the same minimum standards as stock grown here, and where it is sold at retail it must carry the .07A(1) label — name, measure, price — at the time of retail sale. So the three‑field record is not a courtesy Maryland extends to its own growers. It follows the plant across the state line and attaches to the transaction.

The words on the product page are a label

Two short provisions do more work for an online seller than anything else in the chapter.

“Nursery stock may not be labeled or advertised with false or misleading information.” … “A person may not offer for sale, sell, or distribute, any nursery stock represented to be ‘virus‑free’, ‘essentially virus‑free’, or any other similar notation, unless that stock is produced under a program accepted by the Department. All plants offered for sale as ‘virus‑free,’ ‘essentially virus‑free’, or any other similar notation shall have state of origin and name of producer on the label.”COMAR 15.06.02.07C(1), C(3)

“Advertised” puts your product page inside the labeling rule. And C(3) is the cleanest example we have found anywhere in this series of a marketing word that is a regulated claim: type “virus‑free” into a product description and you have just committed to a Department‑accepted production program and to publishing two extra fields, state of origin and producer name, that your catalog probably does not carry. The same discipline applies to every adjacent phrase — “or any other similar notation” is deliberately broad.

Which points at something we end up building for almost every retailer in a regulated trade, and which is worth more than it costs: a small set of rules over the product copy itself. Certain phrases require certain fields to be populated before the item can be published. Certain phrases are simply blocked. The rule lives next to the catalog rather than in a staff handbook, and it runs when somebody writes the description, not when an inspector reads it.

What the storefront is actually for in February

Now put the storefront back next to the seasonality figure, because this is where most garden center e‑commerce goes wrong. A garden center that treats its website as a shop is building a channel that does nothing for seven months and cannot cope for two. The Q1 payroll share says the staff are not there in February; the spring queue says they are overwhelmed in May. A storefront that only takes orders is useless in both months.

The storefronts that pay for themselves in this trade do three things instead, and none of them is “sell a shrub to a stranger.”

The first is pre‑selling the spring. Taking deposits in February for stock that arrives in April converts an idle quarter into working capital, and it does something more valuable than that: it tells your buyer what to order before the order has to be placed. That is a different product from a shopping cart. It needs a pickup window rather than a shipping date, a partial payment rather than a full one, and a catalog that can list a plant you do not yet have, which is precisely the thing a standard inventory‑backed store refuses to do.

The second is the landscaper account. A meaningful share of a Baltimore‑ring garden center’s revenue comes from trade customers buying at their own prices on their own terms, and the wholesale rendering of the plant record — name and measure, no price, on a delivery ticket, per .07A(3) — is a document that already exists in the regulation and almost never exists in the software. Contractor pricing, account balances, and a delivery ticket that is legally correct are worth far more per hour of engineering than a prettier home page.

The third is availability that is true. The complaint we hear most often about garden center websites, from owners and staff alike, is that the site says yes and the bench says no. That is not a design problem. It is the same point‑in‑time inventory problem the invasive rules already forced you to solve, viewed from the customer’s side. Build the movement ledger once and both problems go away together, which is the sort of coincidence worth designing around.

None of that argues for leaving Shopify or Square or BigCommerce. For a great many garden centers the right answer is to keep renting the storefront and build only the layer those platforms cannot host. We say that in every one of these articles and we mean it here too. What we would not do is put a plant catalog on a generic platform and assume the platform will tell you when something is wrong, because on everything in this article, it will not.

What custom actually costs

Our prices are fixed and public, which in this industry is rarer than it should be. Here they are against the two modeled stacks from earlier, with the payback columns computed by dividing the price by each annual subscription figure. Those columns are not a claim that you would stop paying the subscription — as we just said, mostly you should not. They are a way of putting a build next to a rental in the same units.

Every price includes the whole thing: deployed and live, full source code, authentication and roles, payments and integrations, web and native where it is needed, at a fixed price and a fixed date. Hosting and care is $150 a month afterward and you can leave whenever you want, taking the code with you. That last part is the actual argument. A subscription is rent on something you will never own; a build is an asset that keeps working after you stop paying us.

What we would actually build, and in what order

If a garden center in the Baltimore ring called us tomorrow, we would not start with the website. We would start with the two records that everything else in this article hangs off, because they are cheap and they are the ones with a deadline attached.

First, the plant record done properly. A botanical name as the identity, an approved common name and a variety name as attributes that cannot substitute for it, a measure type and value rather than a text field called size, and the four‑part invasive composite: list membership, effective date, container or in‑ground, and woody or not. That record then renders five ways — retail tag, lot sign, field row, wholesale ticket, outbound label — from one source, which is the whole point. A week of work, and it is the foundation of both the compliance story and the storefront.

Second, the movement ledger. Opening balances dated to 15 January 2026 for the thirteen listed species, every movement since recorded rather than netted, and a remaining‑lawful‑stock figure that anybody on the floor can read. This is the thing that makes a sale in 2027 defensible, and it is the thing that also, incidentally, makes your website tell the truth about availability.

Third, the watcher. A scheduled job that reads the Department’s published Prohibited Invasive Plant List and Watch List, compares them against what you actually stock, and tells a named person when something changes. This closes the gap between an annual renewal letter and a catalog that changes weekly. It is the smallest thing in this list and probably the highest‑value, because it converts a rule you have to remember into a message that arrives.

Fourth, the storefront — but the February version of it. Pre‑orders with deposits and pickup windows, contractor accounts with their own prices, availability driven by the ledger, and a ship‑to rule at checkout that can refuse. Not a catalog of pretty photographs. A machine for turning a dead quarter into committed revenue.

We would leave the till alone. Whatever you are running at the counter almost certainly works, and replacing a working point of sale in a trade whose busiest eight weeks are non‑negotiable is a bad trade. Build above it, read from it, and leave it be.

Build, buy, or both

Here is the honest sorting, and it is the only list in this article, because these really are alternatives rather than a paragraph in disguise.

• Keep renting the point of sale, the card processing, the accounting integration and the email marketing. These are commodity problems solved well by people whose whole company is that problem, and Lightspeed, Square, Rain and Celerant are all real products built by people who understand retail.
• Build the plant record, the movement ledger, the list watcher, the pre‑order and contractor flows, and anything that has to answer to a Maryland regulation. Nobody is going to build these for a market of 145 stores, and you cannot configure your way to them.
• Do both, in that order, which is what almost every client of ours actually ends up with: a rented counter, a rented card reader, and an owned layer above them that holds the parts of the business that are specific to being a licensed plant dealer in Maryland in 2026.

If you take one thing from this article and never speak to us, take this: go and find out whether your current system can tell you how many units of a given plant you held on a particular morning eighteen months ago. If it can, you are in better shape than most. If it cannot, you have until 15 January 2027 to fix it for container stock and 15 January 2028 for the rest, and the fix is much cheaper before the records are gone than after.

Why a Baltimore studio is writing this

We are founderandai, a small studio of ex‑startup founders. We build custom software at a fixed price, we work directly with the people who will use it, and the client owns the code at the end. We are based in Baltimore, which is why this series exists: we kept meeting owners in trades where the national software genuinely does not fit, being told the fit problem was their fault.

It usually is not. The pattern in every one of these articles is the same, and garden centers are the clearest case yet. A platform built for a business that sells the same widget in January and July meets a business that earns a fifth of its year in a quarter, sells a product whose legal identity is a botanical name, has to print four different unit systems on a tag, and holds inventory whose lawfulness depends on a date eighteen months in the past. There is no configuration screen for that. There is only somebody sitting down with your actual catalog and building the missing part.

If you want that conversation, it is free and it is thirty minutes, and you will get a straight answer including “keep what you have” when that is the right answer, which it often is.

Questions we get from garden centers

How much does garden center and nursery software cost in 2026?

Most vendors will not say. On 6 September 2026 we opened thirty‑five pricing pages from one machine and eight of them published a subscription price. Square publishes $0, $49 and $149 a month per location. Lightspeed Retail publishes $89, $149 and $289 a month with one register included. Celerant’s Cumulus Retail publishes $125 for point of sale, $350 with e‑commerce and $499 with its Range module, each including one POS and five back‑office users. Rain POS publishes $99, POS Nation $49 against a listed $69, and IT Retail $49. BigCommerce publishes $29, $79 and $299 a month billed annually. Nine products returned a 404 on their pricing URL, three blocked an ordinary request, two never answered, and one resolved to a different company’s homepage. Two honest floors built only from published prices: a stitched stack of Lightspeed Retail Basic plus BigCommerce Growth is $2,016 a year, and a single‑vendor stack of Cumulus Retail with e‑commerce is $4,200. Hardware, card processing, extra registers and add‑on modules sit outside both.

Which plants can Maryland garden centers no longer sell?

Thirteen species took effect on the Maryland Department of Agriculture’s Consolidated List of Maryland Invasive Plant Species on 15 January 2026: Japanese angelica tree, Japanese barberry, Scotch broom, burning bush, border privet, nandina, golden bamboo, yellow groove bamboo, Callery or “Bradford” pear, bee bee tree, Japanese wisteria, Chinese wisteria and their hybrid. Several of them were previously classified as tier 2, which meant they could be sold with required signage; the Biodiversity and Agriculture Protection Act of 2024 abolished the tier system and required every tier 2 plant to be reassessed by 31 December 2025 into either the prohibited list or a Watch List, under Agriculture Article §9.5–301.1. Check the Department’s own page for the current list before acting, because it can be amended at any time and it is the authoritative version.

Can I still sell a Callery pear I already have in stock?

During the phase‑out period, and only from existing stock. COMAR 15.06.04.06B(4) limits sales to “the existing stock of a prohibited invasive plant that person has in inventory as of the date the plant was placed on the prohibited invasive plant list” — 15 January 2026 for these thirteen. B(1) gives one year from that date to a person in possession of such a plant, except a person growing a woody plant in ground, and B(2) gives two years to a person growing a prohibited invasive plant in ground. So container stock runs to 15 January 2027 and established in‑ground woody stock to 15 January 2028. B(5) lets a licensed nursery, broker or dealer ask the Secretary to extend a species’ phase‑out by no more than one further year. The practical consequence is that the lawfulness of a sale depends on a stock position on a date now well in the past, which is a question your inventory system has to be able to answer.

Can I still buy prohibited invasive plants for resale during the phase‑out?

No. This is the part people miss, because it runs the opposite way to the sales rule. COMAR 15.06.04.06B(3) says a person “may not acquire a new prohibited invasive plant after that plant’s effective date on the prohibited invasive plant list,” with no phase‑out attached. Acquisition stopped on 15 January 2026 while selling continued. For a year or two, depending on how the stock is held, the same plant is lawful to sell and unlawful to reorder — which is a state most point‑of‑sale systems cannot represent, because they use one flag for both.

Does a Maryland garden center need a Plant Dealer License, and what does it cost?

If you sell or distribute nursery stock, yes, and you need one for each sales location under COMAR 15.06.02.04A. Note that the definition of “dealer” at 15.06.02.01B(2) catches a person who merely advertises nursery stock for sale or installation, and “location” at B(6) means each place from which stock is sold, offered for sale or distributed. The license is $150 a year per sales location, and a field inspection fee is added on top: $20 for an acre or less, $30 for more than one to five acres, $40 for more than five to ten, and $5 for each acre or part above ten acres up to a maximum of $1,500. So a typical retail yard of an acre or less pays $170 a year, and the cap binds at 300 acres for a total of $1,650. A grower also needs a Nursery Inspection Certificate at $150 per sales location, and a broker needs a Plant Broker License at $150.

What has to be on a plant label in Maryland?

Three things, under COMAR 15.06.02.07A(1): the plant name, which may be the scientific name or an approved common name or both, with the variety name permitted in addition but not in place of them; the height, caliper, grade or diameter if the plant is sold by that measure; and the price of each plant. Individual labeling can be avoided under A(2)(a) if balled‑and‑burlapped, packaged or container stock is grouped into a bed or lot of identical size, variety and cost and marked with a permanent legible sign carrying the same three items — which means a markdown on part of a bench breaks the lot and the exemption with it. Wholesale stock carries the same information except price under A(3), field rows carry the plant name only under A(2)(b), and stock leaving Maryland for another state must stay conspicuously labeled with the plant name and state of origin under B(4).

Can an out‑of‑state nursery ship a prohibited plant to a Maryland customer?

The prohibition in Agriculture §9.5–302(b)(1) covers importing, introducing and transporting a prohibited invasive plant in the State, not merely selling one from within it, so the destination matters rather than the seller’s address. And COMAR 15.06.02.07B requires nursery stock grown outside Maryland and shipped in to meet the same minimum standards as stock grown here, and to carry the A(1) label at the time of retail sale. In software terms that means a checkout serving Maryland addresses needs a rule evaluated on the pair of species and ship‑to state, capable of refusing the order. No general e‑commerce platform ships that behavior, because its model of a destination is a tax jurisdiction and a shipping zone.

How long do I have to keep records of what I bought?

COMAR 15.06.02.04B(3) requires a plant dealer to maintain records of plant acquisitions for one year, including a certificate of inspection or another document showing consignor and state of origin. That is a floor rather than a ceiling, and for the thirteen listed species we would treat it as far too short. The lawfulness of a sale during the phase‑out depends on the stock you held on 15 January 2026, and the in‑ground window runs to 15 January 2028 — or 15 January 2029 if the Secretary grants an extension under B(5)(d). A dealer keeping exactly one year could lawfully destroy the January 2026 records in January 2027 and then spend a year making sales that depend on them. Keeping those records to at least 15 January 2029 costs nothing and closes the gap.