55 %
orders paid
Was 40% in August 2025
Bustlers / Case studies
A new online store without stopping sales
Briefly
Larne’s website came to us after five or six contractors and was held together with patches. We rebuilt it piece by piece, without stopping sales, and paid orders rose to 55% from 40.
We arrived on December 31, 2025 to fix the store. In nine months, we replaced every screen from product cards to personal account, and took over SEO and advertising.
Orders in August dropped compared to a year ago, but paid revenue went up 3%. Now we are building Larne a custom CRM for the fashion business.
Real Larne screens. In the CRM, amounts are changed, customers and order numbers are fictional. Checkout was filmed on a site copy, no SMS was sent.
August 2026 vs. August 2025, online payment — June vs. July 2026.
55 %
Was 40% in August 2025
72 %
Was 57% in June, before retry payment
$3,400
18 orders in the first three weeks
$24–35
Was $82–120 before
Larne sells women clothing in its stores and on larne.ru. The site runs on 1C-Bitrix, accounting is in 1C, orders and bonuses are in RetailCRM. We connected on December 31, 2025 to fix the store and build a new one.
We spent two or three months untangling the legacy: removing unused code, patches, and settings that required a developer to change. The review happened while we updated the store — the site never closed for a single day.
3,703 files in the site working folder from previous contractorsWe deleted more than half
An offer to rebuild the site in six months puts the owner in a familiar problem: you pay now, but have to keep selling through the old store. Errors stay with managers, customers keep hitting them.
We rebuilt Larne piece by piece, starting with what affects buying and payment. Each finished piece started working right away.
Usually
Launch — and only then you see what you got
Larne, 2026
Every screen was first shown to the Larne owner as a mockup. After sign-off, we coded and released an update. Gradually we replaced everything — from header and menu to error page. Nothing of the old design remained, but we kept the brand style: big photos, plenty of white space, black buttons.
01Catalog
Products load as you scroll. On mobile, sizes, favorites, and sale tags appeared. The storefront reorders every night: faster sellers move up.



For a burgundy polka-dot dress, the AI suggests polka-dot items, ruffled pieces, and burgundy tones. The “Picked for you” block considers purchases, views, favorites, and cart; for new visitors it shows bestsellers.
Collections are calculated at night and do not burden the site during the day. We have not measured their impact on sales yet.
12
If collections do not exist, the block switches to simple rules: same category, close price, same color.


02Search
Hints appear in the header at two letters. Search checks both product names and card contents. Everything works by set rules with predictable results.

Before input — “Often searched” and “Now trending”. Search history helps Larne team see demand.

03Checkout
Three screens, each waiting for the site. City in a “Location” field, address at the very bottom.
City, delivery, pickup point, payment, and order details all together. Total amount always visible on the right.


Login inside the form. Phone, Yandex ID, or VK ID
SMS code on the same page. Number is fictional
759 148 px
Checkout button is sticky on screen, next to it — how much more for free delivery.
A sold-out item will not make it into a new order, and if it does not save, the team gets an alert right away.
04Payment
We fixed payment from the account and added a retry: if the customer did not pay right away, the site creates a new payment and opens the bank page.



57 72 %
After retry payment and new confirmation page launched.
$3,400
18 orders in the first three weeks after launch.
Once the bank switched certificates, card payment silently failed for about 18 hours. Now the team sees such problems right away.
05Account
Instead of 30 service statuses, customers see nine clear descriptions. If money reached the bank, they can not pay twice: the account shows “Payment under review”.




In the second month, our SEO specialist joined — preparing the site for search engines. We built about 80 landing pages for queries like “black dresses”, each with its own address, title, and text. We removed duplicates and redirected old filter pages to new ones to keep rankings.
In July, Larne handed us Yandex Direct. We got good results in just a month. Gradually the client handed us everything related to online.
If items are out of stock for more than 14 days, the page hides from search
82 in July and August vs 24 a year ago, by Yandex Metrica
Before us and after handing over Direct
vs four in July and August 2025
We came to fix the site. Now we run all of Larne online.
Store after five or six contractors: account payment did not work, data exchange broke.
Changes started rolling out in the second week.
Landing pages, old page redirects, product schema markup.
Larne handed us Yandex Direct.
New personal account and system for fashion retail alongside RetailCRM.
Once 1C sent an empty export and the site hid 2,155 items out of 2,156.
Usually managers write: customer can not log in, pay, or an item vanished from the order.
The site reports problems, not the customer calling.
Changes to cart, checkout, and payment get separate testing and a rollback option.
From Larne — in-house marketer, marketing agency, 1C specialist, operations director, and store managers. Anything money-related we handle first, including weekends. Every month the Larne owner gets a “was — is” report.
We compared August 2026 to August 2025. Revenue was supported by the growth in paid-order share — how many placed orders customers actually paid for: 40% → 55%. In July and August it was the highest in 13 months of data.
-10 %
First half of 2026 vs last year, by Check Index.
Now online is the priority: the owner decided to invest more in ads and SEO.
We are building a custom system for Larne to manage orders and customers. In the current RetailCRM and 1C setup, you can not trace money by status. Our CRM works alongside RetailCRM and reconciles data.
We count money per item: part of an order might be bought, part returned. Screens with photos show which models sell and get returned, what to resew, and where action is needed today. Red marks only things that need attention.
orders since 2019 loaded into the system
orders for the holiday season: matched the owner report exactly
off on a $460,000 total — due to rounding
in “in progress” orders even though items are sold by 1C data. Another ≈$67,000 — orders not taken to shipment


Next step
A seller in the store shows an item and instantly sees stock at all locations, similar models, discounts, and customer bonuses.

Similar by photo



New collection
Fitted knit dress
$212 black · green available
| Availability | XS | S | M | L |
|---|---|---|---|---|
| This showroom | 1 | 2 | - | 1 |
| St. Petersburg | - | 1 | 1 | - |
| Website warehouse | 3 | 4 | 2 | 1 |
CustomerLarne Club · 1,237 points3% discount
If your store also came from multiple contractors, write to us. No spec needed — a few sentences will do.
We will talk through which part to start with and estimate the timeline and cost of the first step.
This site uses cookies and Yandex Metrica to understand how it is used. Learn more