The short version: If you can trial a platform, test it and skip most of this. These questions are for the ones you cannot trial, and there are several: Barn2Door advertises no free trial, CSAware routes everything through a demo request, and GrazeCart and Local Food Marketplace publish no trial either. In those cases your entire evaluation happens in a sales conversation, which means the quality of your questions is the quality of your decision. The rule that makes them work: ask questions with numeric answers. "Do you handle sales tax?" gets a yes. "Do you calculate, file, and remit, and in which states?" gets something you can act on.
Because for those platforms, asking is all you have.
Most of this category lets you test: Bakesy gives 30 days, Bake.Shop 14, Big Cartel and Cococart 7, and Square Online and Cottage CMS have free tiers with no clock. Where you can trial, testing beats asking every time, because a demo shows you a configured system and a trial shows you yours.
But several platforms, mostly the farm and hub ones, sell through demos:
Those are also the platforms with setup fees attached: Barn2Door's $399 to $599, and Local Food Marketplace's $499 to $1,499 Launch Package. So you are being asked to pay before you know, which raises the stakes on the conversation considerably.
Six questions, and the first two are usually missing from the page.
Question one first, always. On a business doing $50,000 a year through cards, a half-point difference in processing is $250, which is more than most subscription differences and more than you will win by negotiating the plan.
Question four catches a real structure rather than a hypothetical. Local Food Marketplace includes its Launch Package free with annual billing and spreads that cost across your first six months on monthly, so monthly genuinely costs more in year one than a simple division suggests.
Five questions, phrased so a yes is not possible without detail.
Those five are drawn from what actually goes wrong rather than from any feature list. If you sell at more than one place, questions one and four decide whether the platform works at all.
Question three matters if any orders arrive by phone or text. Without it you keep a second list, which means two stock counts and a pick list that is always incomplete.
Question five sounds technical and is not: it determines how cleanly you can redirect if you ever leave.
One question with three parts, because the parts are genuinely different jobs.
Do you calculate, do you file, and do you remit, and in which states?
Almost every platform calculates. Very few file or remit. A feature name tells you nothing: Big Cartel lists "Sales tax autopilot" on its plans without defining the scope, which could mean any of the three.
Follow-ups worth having:
For reference, Shopify Tax does all three for enrolled and eligible merchants: calculation free on the first $100,000 of US sales then 0.35%, capped at 99¢ per order and $5,000 a year per region, with filing charged as a flat fee per return.
That last follow-up is the one nobody asks. Automated filing shifts work, and it does not automatically shift liability.
Four questions, all answerable in an email, none of which appear on a pricing page.
Question two determines whether the list is usable. Without a record of who opted in and who unsubscribed, you have addresses rather than a mailing list.
Question three has a live example: StandScout lists inventory export on its $9.99 tier and data export on Pro at $29.99. So a vendor who downgrades before migrating can find the export they needed sitting behind the plan they just left.
Question four matters more than it sounds. Chargebacks arrive well after delivery, and defending one needs an account you can still log into.
Three questions, and they matter in proportion to how concentrated your revenue is.
Question three is the honest version. For a vendor doing most of a month's revenue in a two-hour Saturday window, support is not a nice-to-have, it is a paid feature they are buying and should evaluate accordingly.
Question two is a good general-purpose question because the answer is revealing regardless of content. A company that answers specifically is telling you something different from one that says outages do not happen.
Two questions, asked politely, and both are legitimate.
Question two is worth asking because it is sometimes answered on the record already. Cottage CMS has said full-feature access is moving up when its Scale tier arrives at $370 a year, which is a clear signal about the durability of its free tier. BakeBug is free through 31 December 2026 for anyone signing up by 1 December, then $4.99 a month.
Neither is a warning. Both are facts worth having before building a season on a current arrangement.
Three tests, and they work regardless of the subject.
A good answer has a number or a name in it. "We handle sales tax" is not an answer. "We calculate everywhere and file in these states, at this fee per return" is.
A good answer distinguishes what the product does from what you would do. Onboarding that still requires three evenings of your time is a different proposition from onboarding that does not, and a straight answer says which.
A good answer includes a limitation. Any vendor who cannot name something their product does badly is either not listening to the question or not being straight with you, and both are useful to know before you pay a setup fee.
That third test is the most reliable one. The best sales conversations in this category involve someone saying "we would not be right for you if..." and meaning it.
That is an answer, and it is worth treating as one rather than as an obstacle.
Some responses and what they usually mean:
That last one is the honest case and it is common. Sales people do not always know processing rates by heart, and a follow-up email with the figure in it is a good sign about how the company operates.
The pattern to watch for is a specific question producing a general answer twice. Once is a misunderstanding; twice is a choice, and it tells you what support will be like when you have a real problem.
Our guide to tracking income and expenses covers why the processing rate in particular is worth being stubborn about: it is the largest recurring cost in this category and the one most often left off the page. If you want a benchmark to hold any quote against, building your real catalog in a trial elsewhere gives you a total you can put next to theirs.
Four questions, before any of the above, because they set the weights.
Question one settles more platform decisions than anything a vendor will tell you. Four combinations is ordinary for a market vendor, and it already exceeds GrazeCart Starter's three delivery zones.
Question four is the useful one to write down. A requirement you can name in advance is one you can ask about; a requirement you discover in month eight is a migration.
Our guides to selling at multiple farmers markets and calculating the real cost per item cover the two numbers most people arrive at a sales call without.
Three things, within the day.
Point three is the discipline that keeps demo-led evaluation honest. A polished demo of a farm platform is persuasive, and the useful counterweight is spending an evening actually building your catalog somewhere you can test for free. If the trialled option does what you need, the demo has to justify a considerable difference.
The FTC's privacy and data security guidance for businesses covers obligations that follow you to whichever platform you choose, and the SBA's guidance on managing your finances is a reasonable framework for tracking what you eventually agree to.
If you want that control, Homegrown is $10 a month billed annually with 0% commission, no setup fee, and 2.9% plus $0.30 processing stated on the pricing page, with a 7-day free trial. It handles pickup at each place you sell with its own schedule and cutoff, local delivery with a radius and a route, and sales tax calculated, filed, and remitted in all 50 states. The honest bounds, which are the answer to several questions above: no point-of-sale, no national shipping, no sell-by-weight, no CSA subscriptions, no bulk product import, and no app ecosystem. If a demo-led farm platform does something on that list that you genuinely need, it is worth its setup fee. You can build your real catalog in a trial and use it as the benchmark you hold every quote against.
Every platform below shows the same four commercial facts, because a table that lists one platform's transaction fee and not another's is not a comparison. "Not published" means exactly that: the company does not state it publicly.
| Platform | Answers you can get without asking | Subscription (annual) | Free trial | Platform fee | Card processing |
|---|---|---|---|---|---|
| Homegrown | Subscription, trial, commission and card rate all on the page | $10/mo billed annually | 7-day free trial | $0 platform fee (0% commission) | 2.9% + $0.30 processing |
| Bake.Shop | All four published | $149/yr (= $12.42/mo) | 14-day free trial | $0 platform fee (0% commission) | 2.9% + $0.30 processing |
| Cococart | Fees not published, so you must ask | $19/mo store, $59 Pro | 7-day free trial | Platform fee not published | Processing not published |
| GrazeCart | One price published, the rest behind a call | $89/mo Starter, rest unpublished | No trial published | Platform fee not published | Processing not published |
| Barn2Door | Published, including the setup fee | $119/mo + $399 one-time setup | No free trial, demo only | $0 commission | 2.9% + $0.30 processing |
| Farmigo | Commission published, processing not | No subscription | No trial published | 2% commission on delivery revenue, $150/mo min | Processing not published |
Barn2Door advertises no free trial and uses demo-based signup, CSAware routes everything through a demo request, GrazeCart publishes one price with the rest behind a conversation, and Local Food Marketplace publishes tiers but no trial.
The payment processing rate, and whether it varies by card type or plan. On $50,000 a year through cards, half a point is $250, more than most subscription differences and more than you will negotiate off a plan.
As three separate things: do you calculate, do you file, and do you remit, and in which states. A feature name like "Sales tax autopilot" could describe any of the three and tells you nothing on its own.
What exports and in what format, whether the customer export includes email consent status, whether export is gated by tier, and how long after cancellation you can still retrieve it.
It contains a number or a name, it distinguishes what the product does from what you would still do yourself, and it includes a limitation. A vendor who cannot name a weakness is not answering the question.
How many distinct combinations of place, day, and cutoff you run, your order count and average order value, what you are doing manually that you want to stop, and what would make you leave in two years.
Write down the numeric answers immediately, ask for anything vague in writing, and compare against a platform you could simply have trialled. A demo is persuasive; a catalog you built yourself is evidence.
If you can trial it, trial it. These questions exist for the platforms you cannot, and there are several: Barn2Door, CSAware, GrazeCart, and Local Food Marketplace all evaluate through a conversation, and most of them attach a setup fee to the decision.
The rule that makes a sales call useful is to ask questions with numeric answers. "Do you handle sales tax?" gets a yes. "Do you calculate, file, and remit, and in which states, at what cost per return?" gets something you can compare. Same for processing rates, exports, and caps.
And judge the answers by a simple test: a good one contains a number or a name, separates what the product does from what you would still do, and admits a limitation. A vendor who cannot name something their product does badly has told you more than they intended.
