Technology · Online grocery · United States
Same traffic. Twice the orders.
The short version
Bliss Tree sells Indian groceries and natural health foods online across more than thirty US locations. Their site was getting traffic and not converting it, and management wanted to know why people were not adding anything to the cart. The answer was structural rather than cosmetic: the site made buying a single item take four actions, in a category where a customer buys twenty-five items at a time. We rebuilt the journey around that. Conversion went from1.8%to3.6%on the same traffic.
A catalogue and a shopping list are different things
Most e-commerce templates assume a browse-and-consider pattern: a shopper arrives, explores, examines a product, decides, buys one thing. That is how you buy a jacket. It is not remotely how anyone buys dal.
Grocery is high-frequency, low-consideration and high basket count. The customer knows what they want before they arrive, they want a lot of items, and they are working through a list. Every action you ask for gets multiplied by twenty-five.
The largest single problem was how pack sizes were handled. Toor dal in 2lb, 5lb and 10lb sat as variants inside one product page — so buying it meant finding the product, opening it, choosing a size and adding to cart. Four actions for one item. Multiply that across a full basket and you are asking a customer for a hundred deliberate steps to do their weekly shop.
So we split them. Each pack size became its own product on the listing, with its own Add to Cart button. Four actions became one. That single change, applied across the catalogue, is most of what moved the number.
Everything else followed from the same idea — count the actions, remove the ones that are not decisions.
What we changed
| 01 | Mapped the user journeys and counted the clicks from landing to checkout, then removed the steps that were not choices. |
| 02 | Split sub-products, SKUs and pack sizes into separate products, so a size could be chosen and added in one action from the listing page. |
| 03 | Added Quick View, Add to Cart on the product listing, product comparison and a wishlist — all of which exist to keep a shopper on the list rather than sending them into individual product pages. |
| 04 | Redesigned the site theme once the journeys were settled, rather than before. |
| 05 | Standardised every category and product image so the listing read as one coherent grid instead of a collection of supplier photographs. |
| 06 | Designed landing banners, videos, promotional and pop-up creative for engagement and conversion. |
| 07 | Redesigned the packaging across the full Bliss Tree USA product range. |
The order of items one and four is the part worth copying. Redesigning the look of a site that converts badly usually produces a better-looking site that converts badly. The journey work came first and the visual work was built on top of a structure that already made sense.
What changed, and how it compares
| Rate | Source | |
|---|---|---|
| Bliss Tree, before | 1.8% | Client’s own analytics |
| Bliss Tree, after | 3.6% | Client’s own analytics |
| All e-commerce sectors, average | 1.8% | Statista, Q4 2025 |
| Food and beverage | 3.1% | Statista, Q4 2025 |
| Food and beverage | 6.22% | Shopify 2026 industry table |
| Food and drink | 1.31% | IRP Commerce, June 2026 |
Read that table honestly and it says two things. Bliss Tree began at the average foralle-commerce, in a category that normally beats the average — which is what underperformance looks like when it is hiding behind respectable-sounding traffic. And it finished above the category figure from one credible source while sitting below another.
Published benchmarks for this category disagree with each other by a factor of five, because they measure different merchant populations. We are showing the spread rather than picking whichever number flatters us most.
How to read this
The conversion figures are the client’s. They came from Bliss Tree’s own analytics and we did not independently audit them. We have not been given the measurement window either, so we cannot tell you over what period the change settled.
Doubled, not tripled. Going from 1.8% to 3.6% is a 100% improvement, which is to say the rate is now twice what it was. We mention this because the same figure is easy to state as a 200% improvement and that would overstate it by half.
We cannot separate the causes. Journey restructuring, SKU splitting, the visual redesign, imagery standardisation and the promotional creative all shipped as one programme. The pack-size change is the one we would bet on carrying most of it, and that is a judgement rather than a measurement. A staged rollout would have told us; a client waiting on a broken site reasonably did not want one.
Nothing here says the traffic changed. The gain came from converting visitors who were already arriving, which is why the second tile says none. We make no claim about traffic volume in either direction.
What happened next
After the New Jersey site was finished, Bliss Tree gave us the redesign of their other US location websites.
That is the part of this we would put most weight on. A conversion figure is a number a client reports to you. Handing over the rest of the estate is a decision they make with their own money, having watched the first one work.
Constraints that shaped the work
Splitting pack sizes into separate products is the right answer for shopping speed and it makes the catalogue larger and harder to merchandise — more products to photograph, price, stock and keep consistent. For a business with a long tail of sizes it is a real operational cost, paid by the team maintaining the store, in exchange for a benefit that lands with the customer. We would make the same call again and it is not free.
We also shipped everything at once and so cannot attribute the result to any single change. On a site that is actively losing orders that is the correct commercial decision and it is a poor experimental one. The honest position is that we know the programme worked and we do not know precisely which part did the work.
If you have traffic and no orders
Count the actions between arriving and buying one item, then multiply by the number of items a real customer puts in a basket. If the answer is over fifty, the problem is not your design, your photography or your ad spend.
More traffic into a site that loses people at the same point just costs more to lose them. Fix the arithmetic of the journey first, and the traffic you already have is usually enough.
Read another
Every figure sourced from the client’s own systems, with the basis stated and the exclusions named.Bawarchi Atlanta
Smile Loft Dental
Sanpeggio’s Pizza
Measurement note. Conversion figures are from the client’s own analytics, reported to us, and were not independently audited by us; no measurement window was provided. Benchmark figures are published 2025–26 industry data, cited by source in the table above, and are drawn from different merchant populations which is why they differ. Scope and deliverables are from our own project records.
