The number printed on the outside is capacity, and capacity is the specification buyers compare. It is also the specification that tells you least about what you are actually buying.
The short answer: two cells with identical capacity ratings can differ substantially in how long they hold that capacity, how much they heat under load, how consistently they behave across a batch, and how they age. Those four determine whether your product is still good in two years, and none of them appear in the headline number.
Why capacity is the easy part
Capacity is the measurement everyone quotes because it is easy to state and easy to compare. It also happens to be the number that two vendors can genuinely match while delivering very different products underneath.
What differs is everything about how that capacity is achieved and maintained:
Cycle life is how many charge and discharge cycles the cell survives before capacity falls below a threshold, conventionally eighty percent of original. Three hundred cycles is roughly one year of daily use. Five hundred is approaching two years. Eight hundred is a product that outlives its enclosure. Same mAh, three very different lifespans.
Capacity retention is the shape of that decline rather than the endpoint. Some cells hold near-flat and then fall off sharply. Others decline gently from the start. For a user experience standpoint these feel different, and only the second one is visible in a datasheet.
Internal resistance governs how much the cell heats under load and how much voltage sags when current is drawn. Higher resistance means more heat, worse efficiency, and a device that struggles exactly when the user needs maximum output.
Batch consistency is the one professionals care about most and buyers never ask about. In a multi-cell pack, mismatched cells do not average out — the weakest cell limits the pack, and then it ages faster than the others, accelerating divergence. Good packs are built from cells sorted into tight bands, and that sorting costs money.
Where the price difference actually comes from
When two quotes for the same capacity differ meaningfully, one of these is usually the reason, listed roughly in order of how often it applies:
Grade. Cell manufacturing is not uniform. Output is sorted into grades based on capacity, resistance, self-discharge and visual standard. Top-grade cells go into applications where failure is expensive. Lower grades are perfectly functional, have wider tolerances, and cost less. Both are legitimate products. Selling the second as the first is not.
Origin and traceability. A cell from a manufacturer with process control and batch records costs more than a cell of nominally identical specification from a source with neither. What the premium buys is predictability and a document trail, which matters enormously the first time something goes wrong.
Electrode materials and construction. The chemistry and physical build behind a given capacity vary, and that variance shows up in cycle life and thermal behaviour long before it shows up in a mAh measurement.
What you can genuinely verify, and how long it takes
Here is the awkward part, and it is worth being honest about it because it shapes how you should buy.
Cycle life cannot be verified quickly. A real assessment takes weeks of cycling. Nobody verifies it per purchase order. What you can do instead is much more useful than nothing:
Ask for the cell manufacturer and the exact model number, not the capacity. A specific model is checkable; a capacity figure is a claim.
Ask for the cycle-life spec expressed as cycles to a stated retention percentage. Vague answers like "long life" or "more than five hundred" without the retention figure attached are not answers.
Ask what the eighty-percent-definition is in their datasheet. If they cannot produce a datasheet, that is itself the answer.
Ask about sorting and matching policy for multi-cell packs, specifically how tight the capacity and resistance bands are. This is a question most suppliers can answer immediately if they actually sort, and cannot answer at all if they do not.
And then do the thing that actually protects you: charge one sample continuously through its full rated load and measure the case temperature. Thermal behaviour is visible in minutes, and it correlates strongly with internal quality. A cell that runs hot under load is telling you something about its internal resistance regardless of what any datasheet claims.
The risk nobody prices in
The cheapest cells on the market are often secondary-grade cells, cells withdrawn from other applications, or cells from decommissioned packs that have been reprocessed. Some are re-wrapped with new labels and new claimed capacities.
The failure mode is rarely dramatic. It is usually this: the product works fine for three months, then returns start climbing, then a review profile that took a year to build degrades in a single season. Nobody can trace it back to a cell decision made months earlier because by then the cost saving is buried in a spreadsheet somewhere.
If a quote comes in substantially below the rest of the field for identical headline specifications, the honest assumption is that the difference is in the cell, because that is where the money is.
How to write this into a purchase agreement
Three clauses do most of the work, and none of them are unusual requests:
Specify the cell model number in the purchase documentation, and require notification plus re-approval for any substitution. Silently swapping to an equivalent is the single most common failure.
Require batch traceability sufficient to identify which cells went into which production lot. Without this, a field problem is unmanageable.
State the cycle-life requirement explicitly, along with the retention definition, so there is something measurable to hold to. A vague commitment is not enforceable anywhere.
These take ten minutes to write and they convert a specification you cannot test into a commitment you can.
ODMKit builds OEM and ODM programmes for charging, audio and mobile accessories — see cell types and selection, power bank OEM programmes and MOQ and pricing, or start at ODMKit.
Top comments (0)