Why this is happening
In a word (because, well, there is a word that's causing it—or rather three together): jargon.
It may not seem like it… both words involved are common language words. But that is actually exacerbating the issue, not helping it.
You are using the word "re-order" to mean something that is a common way to use it on the supply side of your industry. However, clearly the people using the service do not understand it to mean the same thing you do.
Now, you don't perceive it as jargon, but that's because you are familiar with your own terminology within this context. The issue, however, is not what you think "re-order" means, but rather what your customers do, and particularly what they think it means in contextual relation to "new order."
"New order" in this context is equally problematic, because it stands as a binary, blocking/exclusive choice when its own wording seems to be obviously something the customer is engaging in. They are making a new order, from their perspective, even if it is one that shares similarities/common basis to an older order. The presence of this wording likely means it will always have a higher selection rate than whatever it is being placed in opposition to, even when the other option does match the customer's underlying intent as understood in your perspective.
How to fix it
Stop trying to explain the words. Stop referring to things based on your perspective or the supply/service side context. Stop trying to mold the customer into your perspective.
Instead, change your own language entirely. You need to shift your perspective from how the ordering fulfillment process looks to you, to that of someone placing an order. How do they describe their intent for different "types" of orders? To them, are they even different "types" the way you see/categorize them? How can you, in turn, capture and address that?
While you can "train" repeat users at least (to varying extents), people tend to skip explanations. The longer the explanation, the more likely it is to be skipped. Particularly, people take shortcuts. That is to say, if you give them a word that they believe they know the meaning of, they are far less likely to read any text that seems to be explaining it.
If naming the option "re-order" is causing problems, stop calling it that entirely. You may also need to stop putting "new order" as a binary option in this context, because, from the customer perspective, of course it's a "new order." Don't expect to be able to explain around this. Re-ordering the explanation to be in front of the option may help, but ultimately it's masking the underlying issue and will likely have a lower success rate than addressing it more directly.
It's sometimes hard to remember from the context of designing a service that something which is every day and central to you or the operators of the service is often merely a minute or two out of a month for the person using it. In other words, it's given no weight as something needing to be remembered/integrated into that person's schema: trying to effectively train (because that's what this becomes) someone who is not actively using your service continually is an exercise in futility for everyone involved and merely serves to create friction between you and the people using the service.
You're going to need to both ask questions and test
I can't give you a specific answer on what is going to be best suited to the people using your service, just ideas on how to go about finding some. I carried forward a print ordering service with minimal form changes (as a requirement) and most of the people using it have no issue with the wording you are using. But they are also mostly heavy repeat users of the service, who probably spend at least an hour a month if not an hour a week, in some cases, placing orders. The few who are not heavy users are often only placing new orders (usually one offs), so it works well.
Talk to your customers. How would they refer to the process of placing an order based on an old order?
Change the flow!
One way I would approach this, personally, would be to try out something where you always are placing "a new order" and instead at the start of the ordering process you have the option to (with no directly opposing options):
- "import from a previous order" (I consider the wording here to be important in distinguishing what is happening)
- "Use order templating for faster orders!" (sometimes, using unfamiliar phrasing that seems to clearly connotate a contextually specific meaning and makes someone new ask "what's that?" is actually helpful, because it changes their context from one where they presume an understanding of what you are saying to one where they are open to your explanation… but remember, gaining an understanding of this is of far less significance to them than you, and they may still instead just skip it)
- "Reload with changes" (However they conceive of what they are doing when going through this process, that is what you need to match)
Examples would be a page in the order flow that offers the central screen choice to:
- "Yes! Import from an older order so I can quickly modify a new copy of it"
With a "normal flow" continuation option (e.g. bottom right hand corner)
- "No, this order isn't like any of my previous ones"
Not as directly opposed binary elements in a drop down or radial options, but rather as effectively forking flow buttons, with the only exclusion aspect being a natural one following flow choices rather than more direct diametric opposition of an exclusive control (separating them in terms of space will help here, as will having one be an action item in the main selection area and the other being in the normal flow continuation area, due to how it changes their related contexts to each other).
Or alternatively, as an "add-on" to the current order, early in the order process, a checkbox/similar control not placed in opposition to any other options. Or a button that takes you to the selection view for the previous order to select for "importing"/"re-ordering." (you'll presumably need a new view for this, so that there is enough information to help someone select: personally I have no idea what I ordered on a given date/order number from a given company, I need to see the order details at least to some quick summary extent).
Given that you will need to somehow display an order selection from previous orders, I'm personally in favor of an action button of some kind, as it's a strong cue for the flow response or spawning a related modal (however you approach this).
Part of the problem with your current flow is that it creates an element of confusion by how it places an exclusionary binary choice between something that they obviously want to do (place a new order, in their understanding thereof) and something that they also want to do but don't necessarily realize (re-use an older order, with modifications… while placing a new order). Even changing the language won't entirely fix this problem: you need to remove the binary exclusion aspect of this in terms of it being opposed to starting a "new order." Make your "re-ordering" something that instead can be layered on top of a current "new order."
Even if the difference is essentially non-existent from a software logic side, it's crucial in how it guides an understanding and matches the flow to the intent and perspective of the customer based on their context.
Can you programmatically spot when an order looks a lot like a previous order? Even if it's "too late" by the time you can spot this to make the current order faster, that doesn't mean you can't still offer input the next time they're ordering that highlights the "import/re-order/whatever" option. Or sends them an email about a "new feature to streamline your orders!" (which they're likely to miss, ignore, have end up in spam, or ultimately forget by the time they order again, but you can certainly try—I recommend focusing on things you can do within the service's flow, though)
Note that it doesn't matter if the feature isn't "new" in an absolute sense or from your context. The point is that it is new to them. Context and perspective.
See if you can recruit a group of your customers for targeted testing and related feedback, if you can set up a separate instance or otherwise create an A/B style setup with your service by giving your test group a specific, different link to use. Do not discuss what you are changing before hand. Once they have placed some orders with the new format, you should be able to see if it made a difference.
Ideally, if you can shadow some of these customers while they place an order, it would be even more beneficial to getting an actual idea of their process and related thinking.
You'll never get everyone. People are different, they approach processes differently, and trying to design for everyone often creates more problems than it solves. Don't worry about edge cases, in terms of this design aspect. Capture them through support or direct outreach, instead. But you should be able to make a better experience for most of the people using your service, so long as you can approach it from their perspective instead of yours, and then design something that intuitively works with how they are approaching the task they are trying to complete via your service.