Astragalus Codonopsis Tea OEM/ODM One-Stop Customisation | Recipe, Packaging and Flavour to Your Needs

A two-root herbal blend is the hardest kind of herbal tea to keep consistent. Roots vary more than leaves do — by season, by origin, by how long they sat in storage before drying — and every one of those variations ends up in the cup rather than in a report your customer will never see.

Three things decide what a can of astragalus and codonopsis actually tastes like: which part of which root went in, where drying stopped, and how the two were proportioned. All three have to be locked during sampling. Change any one of them afterwards and the material already bought, the packaging already printed and the line slot already booked all get reworked. That cost lands on your account, not ours.

Most of the money a young herbal tea brand loses goes in places the factory never puts on an invoice. Four of them turn up again and again when we take over a project that started somewhere else.

The first run is approved on a sample made from one delivery of roots. The second run draws on a different delivery, and the difference is obvious in the cup. By then the listing is live and the reviews are public.

When a customer complains, the brand owner asks the factory for the record and gets an answer phrased as reassurance. There is no lot number to trace, no moisture figure, no retention sample in a cupboard.

A dried botanical blend is a living material. If the moisture target at packing was set loosely, what was pleasant at month two is dusty at month fourteen — and the shelf life printed on the can is still running.

Without a locked specification sheet, the second order is priced and specified from scratch. Some of that is the factory protecting itself. Some of it is a factory that never wrote the specification down.

None of these are exotic failures. They are all consequences of one decision: whether the blend was specified as a document or as an understanding. A written specification is not bureaucracy — it is the only artefact that survives a change of account manager on our side, a change of buyer on yours, and the eighteen months between the first sample and the third reorder.

We build this blend on a dry line. Selection, drying, cooling, cutting, blending, sieving, metal detection, filling and sealing — no extraction stage, no spray dryer, no reconstitution. The can holds whole or cut pieces of dried root that you steep and then strain.

Two botanicals and nothing that has to be declared as an extract or a carrier. On a shelf where the competing products list anti-caking agents and flavourings, that difference is visible before anyone opens the can.

Form, net content, cut size, blend ratio, moisture target, packaging format, shelf life — everything a buyer needs to compare you against a competitor, and everything a factory needs to reproduce you. One page is the point. A specification long enough to need an index is a specification nobody reads on the factory floor.

A blend locked at sampling behaves the same way at the tenth run, which is what protects a subscription or a reorder rate. This is the part of the specification that earns money rather than costing it.

Both roots come from a tradition of being eaten as food in China, not from a supplement aisle. That is a position a brand can hold for years without a regulator reopening the question — and it is a position that gets harder, not easier, to hold once a product has been reformulated around an extract.

For a brand owner the commercial argument is simple. A herbal tea line lives or dies on whether the third order tastes like the first. We would rather spend the money on the drying stage than on a redesign of the label after launch.

The phrase herbal tea covers at least three different products that share almost no manufacturing process. Being clear about which one you are buying is the first useful conversation we have with a new customer.

It is the third row. Extraction and spray drying let a manufacturer smooth out raw material variation — some of it disappears into the process. A dry line has nowhere to hide it. That is the honest trade: you get a cleaner ingredient list and a more distinctive cup, and in exchange the upstream controls have to be genuinely tight, not tight on paper.

That has a practical consequence at the buying stage. Two suppliers can quote the same product and mean two different things, and the question that separates them is not about certification. It is: what happens to a delivery that arrives outside specification? A manufacturer who answers with a procedure is describing a system. A manufacturer who answers with a reassurance is describing a relationship, and relationships do not survive scale.

Development starts with the two botanicals and the part of each plant you want. For this pair the working material is the root, cut rather than milled.

Proportioned to your target cup, and written into the specification sheet as a range with a nominal value rather than as a single figure nobody can hit.

Coarse cut infuses more slowly and strains more cleanly; a finer cut gives up flavour faster and can cloud the cup. Both are legitimate choices and they suit different markets.

Driven by your retail price point and shelf geometry. This is the change that costs the most to make late, because the can tooling and the artwork both follow it.

Adding one is a development project rather than an adjustment, because the newcomer changes the drying behaviour of the whole blend. Worth doing, but it resets the sampling schedule.

The dry route itself. Introducing an extract would change the product category, the ingredient list and the validation work behind it, so it is a different project rather than a variant of this one.

Each round produces a sample, a specification sheet revision and a written note of what changed and why. Nothing is carried in someone's memory between rounds, which matters more than it sounds: the person who tastes sample three is rarely the person who will sign off production run thirty, and the sheet is what connects them. The two roots in this blend sit inside a tradition of food-medicine homology — ingredients recognised in Chinese culture as both food and medicine — and a customer who wants that framing on the label gets it as a statement about the recipe, not about the health effect.

Customisation on this line divides into three tracks, and knowing where each one ends prevents most of the arguments that happen later.

Blend ratio, cut size, net content, and whether a third botanical joins the blend. Bounded by the dry route and by what the two base materials will tolerate.

Can format, lid, label artwork, carton configuration and case markings for the destination market. All of it open, and all of it subject to a lead time that is longer than most first-time buyers expect.

Label wording, claim boundaries, the document set that travels with the shipment, and market-specific declarations. This is the track where a change made late is most expensive, because artwork is approved against it. A claim removed after the printer has run is not a correction; it is a reprint.

The common worry behind all of this is the belief that a contract manufacturer keeps two factories — one for its own brand and one, lesser, for yours. On this site both routes run on the same line, to the same parameters, with the same batch documentation. The difference between ODM, where we develop the recipe, and OEM, where we build to yours, is who owns the formula and who carries the development cost. It is not a difference in equipment or in quality control.

Development is where most new products fail, and they usually fail at the front rather than at the back. A study of new product development practice published in Heliyon in 2024 by Falahat and colleagues went through the factors that separate successful projects from unsuccessful ones and put the emphasis on how well the brief and the front-end definition are handled before any development resource is spent. That matches what we see: the projects that stall are the ones where the brief arrived as an idea rather than as a specification.

Three kinds of enquiry get a no, and it is worth stating them plainly because the third one is the least obvious.

Anything positioned as a medicine. No disease names, no treatment claims.

Concepts that only sell if a disease claim carries them. If removing the claim removes the reason to buy, there is no product underneath.

Ingredients whose identity cannot be documented. If we cannot state what a material is and show the paperwork, it does not enter the blend, however good the price is.

Product format is not on that list. If the format you want is not one we keep in stock, that is a project question — scope it as a standalone development — not a refusal.

Compliance is not a certificate you display. It is a chain of documents that has to hold together at the point of import, and it is decided by the strictest market you intend to sell into rather than the easiest one.

A 2026 review in Foods by Guo and colleagues examined how regulatory mandates have become the main driver behind advanced chemical analysis in food safety work — in other words, the analytical capability a supplier can demonstrate is increasingly the thing that determines market access, not a supporting detail. For a botanical blend that means contaminant testing, moisture and microbiological control, and the analytical records that sit behind them.

A locked specification sheet for the finished product, signed by both sides before the first run.

Batch documentation tied to the lot number on the can — what went in, what was measured, what was released.

Label content review against the destination market's rules, done before artwork is finalised rather than after.

A written statement of where responsibility sits for each element, so that a complaint has an owner instead of a discussion.

Certificates exist for this facility, and which certificate covers which product is a question with a specific answer that we provide on request rather than a list we paste into a web page. A buyer who cannot get that answer out of a supplier should treat it as information about the supplier.

Three questions sort a certificate register quickly. Who issued it? What scope does it cover — which processes, which product categories? And does the certificate name the product you are actually buying, or a neighbouring category in the same plant? A supplier who answers all three without hesitating has a system behind the paperwork. One who answers the first and improvises the rest has a PDF.

The line is short and unglamorous, and that is the point: every stage exists because removing it changes the product. What follows is the sequence and, for each stage, the control that is recorded against the batch.

Nobody looking at that sequence would call it advanced. Its value is that each row can be pointed at when a question arrives six months later — by us, by you, or by an auditor who has never seen the plant before.

The order matters as much as the steps. Cutting before cooling produces an uneven cut, and an uneven cut is what makes a blend stratify in the can. Sieving after blending rather than before leaves fines distributed unevenly through the batch. Metal detection placed after filling would inspect a sealed can instead of the material, which is the wrong question asked at the wrong moment. None of these are sophisticated decisions; they are simply decisions that have to be made in a particular order and then written down.

Control points matter only if the numbers behind them are written down, and stage-specific verification is what makes a food safety system real rather than nominal. A 2021 case study in the Journal of Food Science by Chen and colleagues described implementing a food safety management system meeting ISO 22000:2018 and HACCP requirements in a botanical product operation, and the work that carried the project was the verification step at each stage rather than the manual at the front.

A single batch number follows the material from the incoming lot through drying, cutting, blending and filling to the can on a shelf. That is the thread a complaint gets pulled along.

A sample is held for every batch, for longer than the declared shelf life, so that a question six months later can be tested against the physical product rather than argued about.

When a batch misses a target, the response is recorded rather than discussed: what failed, what was done with the material, and what changed afterwards. A deviation that leaves no trace is a deviation that will happen again.

Roots arrive with variability attached — different growing seasons, different origins, different lengths of time in a warehouse somewhere upstream. Catching that at receiving is cheaper than catching it in the cup, and it is the only point at which the material can still be rejected without cost to a finished batch.

This is also the control point most often skipped, because it is the least visible. A factory that inspects hard at the end of the line and lightly at the start is inspecting its own work rather than its suppliers' material, and by then the only options left are rework and write-off.

What happened. A consignment of codonopsis arrived wetter than the lots we were used to. It was the same supplier, a different growing season, and the difference sat inside the incoming tolerance we were working to at the time — a range, not a window. The roots cut soft. The blades dragged, the slices came off uneven, and the cutting stage threw off far more fines than a normal run. Blending was done by weight against the locked ratio, so the numbers on the record were correct, and the batch went through sieving, metal detection and filling without anything triggering.

How it surfaced. A buyer in Southeast Asia raised it: the first servings out of a can were noticeably stronger than the last ones. We opened the retained sample from that batch and could see it without any instrument. The fine material had settled to the bottom of the can and the larger slices had stayed on top. The blend was right. The distribution inside the can was not.

What we did with it. We did not ship the balance. The cans still in our warehouse were held and reworked, and the cans already in the customer's hands were replaced at our cost. The customer carried neither the loss nor the argument, and the conversation stopped being about fault once the replacement was moving.

What changed afterwards. Four things, all of them recorded. Incoming moisture became a release condition rather than a preference, so an out-of-window lot is rejected at receiving instead of being absorbed quietly into production. Cutting is preceded by a re-firing step that brings the whole lot to one moisture window before the blades touch it, so cut size no longer depends on which field the roots came out of. Fines removed at sieving are weighed and recorded separately instead of being fed back into the blend. And every batch now gets a stratification check before release: one can is opened, the contents divided into top, middle and bottom, and each portion brewed and compared against the others.

That last check takes about ten minutes and no equipment beyond a scale and a cup. It is the cheapest control point on this line, and it is the one we were not running.

If you already know the blend you want, the fastest path is to send the specification and let us confirm what on this line can hit it. If you know the shelf position but not the recipe, we develop it and you own the formula.

Either way the first step is the same: a written specification, agreed by both sides, before anything is printed or bought. A herbal tea project that skips that step is not saving time — it is deferring the cost to a point where the packaging is already paid for.

To move forward, ask for the specification sheet template and the batch documentation we hand over for a typical run. Both are documents, not presentations, and a supplier who cannot produce them quickly is unlikely to produce them at all once your order is in the queue.

The reason to start with the specification rather than the price is that price without a specification is not a number anybody can act on. Two quotes for the same product can differ by a wide margin and both be honest, because they are pricing different cut sizes, different packaging and different drying cycles. Fix the specification first and the quotes become comparable — and the conversation turns into a decision instead of a negotiation.

Two dried roots — astragalus root and codonopsis root — cut rather than milled, blended to a locked ratio and filled into a 90 g can, 12 cans to a carton. You steep the pieces and strain them. There is no extract, no carrier and no instant powder in the format.

By locking the specification before the first run and controlling incoming material afterwards. Cut size, blend ratio and the moisture target at packing are all measured per batch and recorded against the batch number, so a drift is visible in the records rather than only in the cup.

Yes, and that is usually the most useful thing a serious buyer asks for. For any batch we can show the incoming lot records, the drying parameters, the blend ratio, the fill weight checks, the metal detection log and the retained sample register entry.

On this line, anything that requires extraction or spray drying — the route is dry processing from start to finish. That is a statement about the line, not about the company. A format this line does not run can be scoped as a separate development project, and we will tell you honestly whether it is worth doing.

Five things: the recipe or the direction you want us to develop, the target markets in order of priority, the net content and packaging format, the label category you intend to use, and your launch window. With those we can give you a written scope, a sampling plan and a delivery range instead of a single number that means nothing.