In June and September 2025, the US Consumer Product Safety Commission announced two recalls of Anker power banks, the first covering the PowerCore 10000 (model A1263) at about 1,158,000 units, the second covering five models, A1647, A1652, A1257, A1681 and A1689, at about 481,000 units, alongside 713,000 units in mainland China and roughly 520,000 in Japan, and adding the rounds together, more than 2.38 million power banks were recalled worldwide, a figure Anker Innovations itself later confirmed in its Hong Kong IPO prospectus, the cause was written just as plainly, an upstream battery cell supplier changed core materials during production without authorization, leaving some batches prone to overheating and even fire after extended use
This was the biggest product safety event the cross-border e-commerce industry saw last year, and it makes for easy news coverage, but news coverage has two problems, it only reports the event itself, and it fades fast, search for the Anker recall today and the discussion has cooled, the earnings report is out, the prospectus has been updated, it looks like the whole thing has been put to bed
What we wanted to look at is the other side, after the recall, what did the review sections of the power banks already sold look like
On October 28 we ran a new report on an Anker 20,000mAh power bank with a 22.5W output and a built-in USB-C cable, currently priced at $35.99, pulling 209 reviews, and this product is exactly model A1647, one of the units named in the second recall, on the other side of the comparison sits an INIU 20,000mAh power bank with 182 reviews
The most striking thing in the report is a pain-point group we had never seen in any other category, it is called Recall Process, standing on its own as a category, with 9 mentions accounting for 9.4% of all pain-point mentions, above it, safety issues account for 22 mentions or 22.9%, and the two groups together mean roughly one in three pain-point mentions ties back to the recall in some way
The product page shows a 4.4-star rating that looks perfectly normal, only by pulling apart the 55 one-to-two-star reviews one by one do you learn what this listing has been through
The recall fades, the reviews do not
Start with the scale, one-to-two-star reviews make up 26.3% of this product's sample, and drilling into the items, "recalled product" appears in 12 mentions or 12.5% of all pain-point mentions, "recalled due to fire risk" in 6 mentions or 6.3%, plus 9 mentions of a "poor recall process", these items alone add up to 27 mentions against a total of only 96 pain-point mentions for the product
In other words, for a listing that went live in November 2024 with reviews spanning two years, the recall event itself happened over a few months in the middle of last year, but its imprint in the reviews is still there, 84% of the reviews we sampled fall within the last six months, and recall-related complaints are still coming in
Reading those recall reviews closely, one pattern stands out, the anger is rarely aimed at the battery itself, it is aimed at finding out after the fact, one buyer wrote about only realizing afterward that the power bank he carries in his bag every day was on the recall list, another complained about discovering the replacement process was more hassle than expected right when he finally tried to deal with the thing, and a third simply used the review space as a public service announcement, telling later buyers to go check their own model numbers
That third type deserves special attention, it is barely a review anymore, it is a buyer-written recall notice, and it works against the product more effectively than ordinary anger does, because the people reading it treat it as useful information and bookmark it, a one-star post that keeps telling everyone who opens the reviews that this listing once had a safety problem
The trend data makes the point even better, this product's monthly negative-review rate spiked to 63.6% in July 2026 at the peak of the recall fallout, then fell to 30% in August, 20.3% in September and 14.3% in October, recovering month by month
But I would suggest reading that curve in two parts, the 63.6% was event-driven panic that recedes on its own, what is worth watching is the floor after the panic subsides, how much of October's 14.3% is recall residue and how much is the product itself, that question needs a few more months of data
News coverage lives for days, reviews live for years, a one-star review saying "this thing got recalled" stays on the page, and every potential buyer who opens the negative reviews will see it, this is the most underestimated cost of a recall, it does not appear in the claims ledger, it appears in the conversion rate
A poorly run recall becomes its own complaint
The part of this report I stared at longest is that standalone "Recall Process" group
Nine reviews complained about the recall being handled badly, registration hassle, unclear steps, long waits for replacements, these complaints have nothing to do with whether the battery is faulty anymore, the cell defect came from a supplier changing materials, which was beyond Anker's control, but the recall experience was entirely within it, and handled poorly, a product problem escalates into a brand problem
The costs Anker disclosed in its prospectus make the arithmetic concrete, global recall spending on replacements, logistics and after-sales advances reached 392 million yuan, a potential claims ceiling of 172 million yuan was booked as a contingent liability, plus a new 104 million yuan product quality guarantee liability, with direct recovery costs estimated between 432 and 557 million yuan, and 2025 operating cash flow fell 82.5% year over year to 481 million yuan
Those are the on-ledger costs, the nine "poor recall process" reviews are the off-ledger cost, and what they damage is the most valuable thing the brand owns, 22.5% of this product's positive reviews explicitly say "Anker products are consistently reliable", that trust is what supports a $9 premium over INIU's $26.95 price tag
Sellers reading this should take away one judgment, if your product ever has a safety problem, the decision to recall is not really a decision, you have to do it, but the design of the recall has a lot of room in it, how many steps to register, refund or replacement, how fast support responds, every one of these gets written into reviews, a decent recall keeps the damage at the product level, a careless one pushes it up to the brand level, and the downstream difference between the two is an order of magnitude
The INIU control group: same category, completely different pain structure
INIU on the other side makes a valuable control, also 20,000mAh, also a strong rating at 4.6, with a negative rate of 19.8%, lower than Anker's
But its pain structure is entirely different, INIU's top complaint is "too heavy", 26 mentions or 34.7% of its pain-point total, safety sits at only 6.7%, and recall mentions are zero, buyers are complaining about heft, and the reviews read like a running joke on the product's dimensions, one buyer wrote "can't speak for how it works but bought this for my wife and she said 'no way, too heavy'", before explaining he was looking for a hiking power bank and this one simply was not practical, so it went back
Another buyer left a long, measured takedown of the marketing rather than the hardware, "definitely not 'pocket-sized'", he wrote, acknowledging that the capacity and charging performance are solid, but pointing out that at over 11 ounces and more than an inch thick, "it will weigh down your jacket or pull down your pants", his verdict was fair and devastating at once, great as a backup battery in a backpack, wrong product if you believed the listing
Put the two complaint sheets side by side and you see two completely different ways of failing, Anker failed on trust, INIU failed on weight, one product's question is "do I dare use it", the other's is "do I want to carry it"
The platform ratings reveal none of this, 4.4 and 4.6 sitting next to each other suggest two products of similar quality, only by taking the negatives apart do you learn their buyers are upset about entirely different things, which is why we keep saying the platform score is a false smooth, it averages two completely different kinds of pain into one decimal
One more detail worth noting, INIU's 182 reviews contain 2 false-advertising complaints, and the anger there has nothing to do with product performance, the listing said pocket-sized and the product does not fit a pocket, this is the cheapest complaint category to fix, just change the copy, but it stings because buyers feel deceived, and reviews written out of that feeling are the harshest
The other side of the ledger deserves a mention too, otherwise the picture is distorted, Anker's positives include 38 mentions of high capacity and multiple charges, 27 of fast charging, 24 of the built-in USB-C cable, the product's foundation never collapsed, "charges our phones at least 3 times before it has to be charged again" is a real experience, on the INIU side fast charging drew 28 mentions with one buyer calling it "crazy fast" and double the speed of anything else he has owned, neither of these is a bad product, their complaints grew on products that work, which is exactly why the teardown is worth doing, complaints on a broken product teach you nothing, complaints on a working product are the improvement roadmap
The report also runs a sensitivity estimate, if Anker eliminated every battery-and-charging-related negative, the sample negative rate would drop from 26.3% to 7.2%, INIU's would fall from 19.8% to 6%, these are theoretical ceilings, not promises, but they size the opportunity, more than half of the negative weight on each product presses down on a single spot
The category floor: battery complaints nobody escapes
Where the two products do agree is the top group, both point at battery, charging and endurance, 44% for Anker and 42% for INIU, some complaints in this category belong to everyone
Anker's reviews include 9 mentions of fast battery drain, INIU's include 8 of intermittent charging and 6 of very slow recharging, different directions, same group
The three-star reviews say even more, 28 of Anker's 48 three-star reviews also mention battery endurance, the product works, buyers are just not satisfied, we call these complaint-level issues, they get written into reviews and weigh on word of mouth without directly driving one-to-two stars, and their fix priority sits below return-level issues, fix what drives returns first, then what drives grumbling
The three-star count carries its own signal, Anker has 48 three-star reviews against 55 one-to-two-star reviews, nearly equal, INIU has 41 against 36, also nearly equal, we verified across 60 product records on our site that buyer ratings concentrate heavily in three cells, five stars, three stars and one-to-two stars, with four stars a dead cell whose median share is under 10%, both of these power banks fit that shape exactly, satisfied buyers are generous, unhappy buyers are blunt, and the ones who settle all land on three stars, take the three-stars apart on their own and you see one layer more than the negatives alone can show you, why the people who stayed are not excited
Across our site this category now has three power bank reports, Anker vs INIU, Anker vs UGREEN and ZZI vs Anker, together 1,179 reviews, and the category's weighted negative rate is 30.7%, ranking 7th among the 20 categories we track, a complaint-dense category sitting behind automatic litter boxes at 47.5%, translation earbuds at 39.3% and smart feeders at 37.3%
So do not let the commodity status fool you, mature technology and huge volumes do not insulate power banks, the battery carries three risks by nature, capacity shortfall is an experience problem, heat is a safety problem, and degradation is a trust problem, all three end up in reviews
What a 2.38-million-unit recall teaches smaller sellers
Coming back to Anker, the event itself cannot be copied by smaller sellers, but the remediation checklist in the prospectus can, it was paid for with 2.38 million power banks
The updated prospectus lists the fixes, terminating cooperation with the cell supplier involved, building a dual-supplier cell system, cutting the top five contract manufacturers' capacity share from 82% to 73%, adding 22 backup vendors, signing supplementary agreements with manufacturers that require advance reporting of any raw-material changes and make unauthorized substitution fully compensable, and on the product side culling 70% of low-margin SKUs, trimming the lineup from 126 models to 39
None of these moves is sophisticated, what was hard is that all of them came after the fact, a single reporting clause upfront could have stopped what cost 2.38 million units and hundreds of millions of yuan, most Amazon sellers also outsource production, and a mechanism like this costs almost nothing to adopt, so copy it
The second thing to copy is the monitoring habit, this A1647's negative rate fell from 63.6% back to 14.3% month by month, if the operator only watched the platform score, 4.4 stars looks serene, only by pulling the one-to-two-stars out monthly do you learn when the event's impact bottoms out, and even for products that never approach a recall, monthly shifts in complaint structure are the earliest warning you get, a group's mentions suddenly rising tells you where the problem is long before the total negative rate moves
The third is the comparison habit, knowing where your own product should improve takes more than tearing down your own reviews, tear down your strongest competitor's too, and see where the pain structures differ, the Anker-INIU report is a ready example, one product's illness is safety and trust, the other's is weight and advertising, and only a matched remedy counts as improvement
A note at the end
As always, the data boundary, this report's sample prioritizes negative and recent reviews, the 26.3% negative rate is based on the sample and does not represent the product's true platform-wide negative rate, all shares in the report describe the structure within the sample
News coverage wraps up, earnings calls wrap up, prospectus filings wrap up, but those 27 recall-related mentions still sit on the review page, and everyone who researches this product will read them, for sellers the review section is the archive that never moves on, it remembers everything, including the part you would rather forget
All of this is laid out in the full analysis report, the Anker vs INIU report is on our site, and the other two power bank reports, Anker vs UGREEN and ZZI vs Anker, are in the library as well