300x
Growth in monthly Amazon revenue, from about $520 to $156,000
#5
Organic rank for the primary category keyword on Amazon India
200%
Improvement in conversion performance
32
Listings corrected and accepted by Amazon, 8 Oct 2026
Challenge
Proven Water Industries makes RO, UV and UF water purifiers for Indian homes and sells them through its PROVEN store on Amazon India. Dcrayon runs the account's advertising and listing work.
A purifier brand competing on price in a category that buys on trust
Proven's range is built around 12-litre purifiers that combine RO (reverse osmosis, a membrane that removes dissolved salts), UV (an ultraviolet lamp that treats microbes) and UF (ultrafiltration, a fine membrane filter), most with copper, zinc or alkaline stages and a TDS adjuster. TDS means total dissolved solids, the measure of salts and minerals in water, and it is the number that tells a household whether it needs RO at all. The models carry names such as Pixer, Swift, Nexon, Jupiter, Magnum and Eleven, and in October 2026 the store held 37 listings on Amazon.in. Proven's advertising account is connected to Dcrayon's in-house Amazon Ads platform, and our scope covers Sponsored Products, listing content and catalogue data.
In early October the account's order rate had fallen steeply since August, while account health stayed normal. On 8 October 2026 we ran a full brand audit to find out why, using the account's own order, advertising and FBA (Fulfillment by Amazon) data, Amazon's Business Reports and Search Query Performance report, Amazon's catalogue data, and Jungle Scout keyword and share-of-voice data for Proven against five rival brands. The audit found three causes acting at the same time, plus listing faults that made each of them worse.
The best-selling models ran out of FBA stock
Six of the models that carried most of the August sales, including Nexon Copper, Jupiter Full Brown, Magnum Black and two Pixer Copper versions, ran out of stock in Amazon's warehouses between mid-August and mid-September. When a listing has no stock its page drops out of search, and it starts lower when stock returns. Most of the fall in orders traced back to these six products, and advertising was still running on some of them while their stock ran down.
Clicks became far more expensive without buying more sales
Several campaigns had been switched to Amazon's "dynamic bids - up and down" strategy, which Amazon describes as raising or lowering a bid based on performance by up to 100 percent. Spend moved into top-of-search placements, and the account's average cost per click more than doubled between late July and September. Budgets were not the limit: only 2 of the account's 215 campaigns reached their daily budget. The bids were. The same weekly spend was buying roughly half the clicks it had bought in July, and searches with no chance of converting, such as "amazon warehouse" and searches for a rival brand's filters, were still being paid for.
Shoppers saw PROVEN more often but bought it less
Amazon's Search Query Performance report for Proven's 15 biggest searches showed the brand's share of impressions roughly doubling between August and late September, paid for by advertising, while its share of purchases on the same searches fell by more than half. The wider market had dipped after the monsoon peak, but far less than Proven had. Price was not the cause: at the deal price Proven was the cheapest RO purifier with copper and alkaline stages on page 1. The pages were losing the shopper after the click.
The listings contradicted themselves and hid their own highlights
The best-selling Pixer page said "10-stage" in its first bullet and "12-stage" in its highlight line. Fourteen SKUs had titles past the 75-character limit Amazon introduced on 27 July 2026, so Amazon flagged them and hid their Item Highlights line. Every listing shared one 397-byte backend search string that repeated "water purifier" 16 times and was identical for every model. Fourteen listings described their purification method as "Alkaline" rather than reverse osmosis, duplicate SKUs gave the same product a different capacity and model name, and Amazon was hiding images on 23 listings for breaking its image standards. The second image on the Pixer page announced the installation charge before it showed the product's benefits.
Why Proven's Amazon sales fell, August to October 2026
The three causes the 8 October audit found, and who each one affected.
| Issue | Who it affected | Business impact |
|---|---|---|
| Six best-selling models ran out of FBA stock | Shoppers searching for those models; Proven's ranking on them | Most of the fall in orders; pages dropped out of search |
| Dynamic up-and-down bidding at top of search | Proven's advertising budget | The same spend bought about half the clicks |
| Listings contradicted themselves and hid highlights | Shoppers comparing purifiers after the click | Purchase share fell while impressions rose |
Solution
What the comparison with five rival purifiers showed
The second half of the audit put Proven's two lead models, the Pixer and the Swift Black, next to five rival brands on the 100 biggest purifier searches, using Jungle Scout's keyword ranking data from 7 October 2026. Proven's organic position (where a listing appears in unpaid results for a keyword) was top 10 on none of those searches and top 50 on 17 of them for the Pixer and 15 for the Swift. On the six head searches that carry most of the buying, it was not in the top 150 at all. That explained why advertising cost tracked click prices so closely: on those searches every Proven sale had to be bought with an ad.
The comparison also showed where shoppers were being lost. The rival pages stated the TDS level their purifiers handle, the flow rate in litres per hour and the filter life, while Proven's pages stated none of them. Two rivals included installation, and Proven's image and warranty text disagreed about what the warranty covered. These were findings for the brand's own decisions, so the audit listed them as questions for Proven rather than inventing answers on its behalf: the true stage count, tank capacity, TDS range, flow rate, filter life and warranty scope.
Listing corrections applied on 8 October
The brand approved the listing and advertising changes on 8 October and we applied them the same day through Amazon's Selling Partner API, the programming interface Amazon provides for seller data. Every patch was submitted as a dry run first, and the state of every listing and ad setting was saved before any change so that each one could be rolled back. All 32 SKUs (stock keeping units, the seller's identifier for each listing) came back valid in the dry run and were accepted.
Titles. The 14 titles over 75 characters were shortened while keeping the family naming, and the two lead titles were rewritten in a brand, model, technology, product, materials and size order: "PROVEN Pixer RO+UV+UF Water Purifier for Home, Copper Zinc Alkaline, 12L" for the Pixer and the same structure for the Swift Black. With the titles under the limit, Amazon shows the Item Highlights line again.
One number per fact. Where a page claimed both 10 and 12 stages, the bullet now says "Multi-Stage" and the highlight and images keep "12-stage", so each page gives a shopper one answer. No new number was written in, because the brand had not yet confirmed the true count.
Backend search terms. Backend search terms are keywords that Amazon describes as not visible to customers. The 397-byte string on all 32 listings was replaced with a 226-byte string that contains no title words, no brand names and no repeated phrases.
Catalogue attributes. The purification method on 14 SKUs was corrected from "Alkaline" to "Reverse Osmosis", and seven duplicate SKUs had their capacity corrected from 8 litres to 12 and their model name aligned with the main SKU, so each ASIN (Amazon Standard Identification Number, Amazon's identifier for a product listing) gives Amazon one answer.
Images. On the Pixer page the installation-charge image moved from the second slot to the fifth. Two images that Amazon was already hiding both carried app store badges, so they were removed rather than re-uploaded.
Sponsored Products bidding reset
The advertising changes went into the brand's Amazon Ads profile on the same day, under the same before-and-after record. Twelve of the 13 campaigns on "dynamic bids - up and down" were moved to "down only" (the thirteenth is managed by Amazon and cannot be changed through the API), and the top-of-search placement adjustments on seven campaigns, which had been set between 40 and 100 percent, were capped at 25 percent, or 50 percent on brand campaigns. One campaign that had produced a single order on heavy spend was paused, and 29 product ads were paused on models with one to four units left, to be switched back on when stock arrives. We added 140 negative exact-match keywords for searches that had never converted, paused three keywords that kept buying irrelevant searches, and capped the highest bids and ad-group defaults.
The bidding work also moved spend towards searches Proven can win. Jungle Scout showed six searches where the Pixer or Swift already ranked between 34 and 47 organically, such as "ro water purifier filter for home" and "r o water purifier for home". Bids on five of those exact-match keywords were raised and "water ro" was added as a new exact-match keyword, because those searches cost less per click than the head terms and Proven's listings already appear on them.
Because the brand's deal price started on the same morning, we set a rule for deal days: if impressions fall by more than 30 percent while a deal is running, the capped bids are lifted back towards their earlier level until the deal ends. That rule was used on 8 October, and the down-only bidding and placement caps stayed in place.
Checks that protected price, deals and the account
The brand had asked us to leave prices and deals alone, so after the changes we compared all 148 price and offer fields on the 37 listings against the pre-change backup. All 148 were unchanged and the deal prices were still live. The live pages were then read back: the new titles, keywords, bullets, capacity and purification method were showing, and the title-length error was gone from every corrected listing. This is the working method of our Amazon account management service, and our guide to managing Amazon PPC against TACoS explains why we read advertising cost against total sales rather than ad sales alone.
Proven listing fields corrected on 8 October 2026
The catalogue changes applied through Amazon's Selling Partner API, all accepted after a dry run.
| Listing element | Before | After | Why |
|---|---|---|---|
| Titles | 14 titles over 75 characters, highlights hidden | Shortened, family naming kept | Amazon's 75-character title limit from 27 Jul 2026 |
| Stage count | 10-stage in bullet, 12-stage in highlight | One number per page | A shopper gets one answer |
| Backend search terms | One 397-byte string on all 32 listings | One 226-byte string, no repeats or brand names | Distinct terms instead of one phrase 16 times |
| Purification method | Alkaline on 14 SKUs | Reverse Osmosis | Matches what the product is |
| Pixer gallery | Installation charge in slot 2 | Moved to slot 5 | Benefits shown before costs |
Sponsored Products changes on Proven's ads profile
The bidding and targeting changes made on 8 October, with the count of each.
| Change | Count | Status |
|---|---|---|
| Campaigns moved from up-and-down to down-only bidding | 12 of 13 campaigns | Delivered |
| Top-of-search placement adjustments capped at 25 percent (brand 50 percent) | 7 campaigns | Delivered |
| Product ads paused on models with 1 to 4 units left | 29 ads | Delivered |
| Negative exact-match keywords added | 140 keywords | Delivered |
| Exact-match bids raised on rankable searches | 5 keywords plus 1 new | Delivered |
Results
Proven's growth on Amazon over the engagement
Dcrayon's work on Proven's Amazon account has run as a long-term growth programme rather than a short-term advertising campaign: listing optimisation, conversion-focused product content and A+ Content, a keyword strategy built on buying intent, and advertising structured by product, keyword intent and match type. Over the engagement, Dcrayon's account performance data shows:
- Monthly Amazon revenue grew 300x, from approximately $520 to $156,000.
- Proven's primary category keyword reached #5 in organic results on Amazon India, over approximately 14 months.
- Conversion performance improved by 200%.
The October 2026 audit described above dealt with a fall in orders that began in August 2026.
Where Proven's account stands after the audit
Proven now has titles that fit Amazon's limit and show their highlight lines, pages that state one stage count, a shorter backend search string with no repeated phrases, catalogue attributes that describe each product correctly, and a gallery that leads with the product rather than the installation charge. Its advertising runs on down-only bidding with capped placement adjustments, it no longer pays for searches that never convert, it no longer advertises models that are nearly out of stock, and more of its spend sits on the searches where its listings already have some organic standing.
What remains is on the brand's side, and the audit set it out plainly: restock the six models that sold out, confirm the product facts the listings still lack (TDS range, flow rate, filter life and what the warranty covers), bring back coupons, and supply new artwork for the empty image slots. The next review compares cost per click, advertising cost of sales by campaign and impressions on the rankable keywords against the pre-change record.
Proven organic, sponsored and BSR observations
Real observations for Proven's lead models, one measure per column.
| Date | Keyword and ASIN | Organic | Sponsored | BSR |
|---|---|---|---|---|
| 7 Oct 2026 | ro water purifier filter for home, Pixer (B0DYPGSDX1) | 37 | Not tracked | Not read that day |
| 7 Oct 2026 | ro water purifier filter for home, Swift Black (B0DT16GM4V) | 35 | Not tracked | Not read that day |
| 7 Oct 2026 | r o water purifier for home, Pixer (B0DYPGSDX1) | 44 | Not tracked | Not read that day |
| 8 Oct 2026 | No keyword, Pixer family (B0DYPGSDX1) | Not applicable | Not applicable | 67 in water purifiers |
| 13 Oct 2026 | No keyword, Swift Black (B0DT16GM4V) | Not applicable | Not applicable | 101 in Water Coolers, Filters and Cartridges |
Key takeaways
What Proven's audit teaches about Amazon sales falls
A sales fall on Amazon rarely has one cause. Proven's had three at once, and each made the others look worse: stock-outs pushed pages out of search, so more sales had to be bought with ads, just as automatic bidding was raising the price of every click. Reading orders, advertising, FBA stock and search share together is what separated the causes, and it kept the fixes in the right order: stock and bids first, then the listing faults that were losing shoppers after the click.
The second lesson is to keep three measurements apart. Organic position for a keyword, sponsored placement and Best Sellers Rank (Amazon's sales-based ranking within a category) each move for different reasons, and Proven's ranking table below reports them in separate columns. Our article on Amazon listing optimisation covers how we structure titles and attributes, and the GetSetNova case study shows category-by-category account work for a brand with several product types.
If your Amazon sales have dropped and the cause is not obvious, ask us for an account audit.
CategoryEcommerce