We found whatbroke, when it broke,and what it costs.
Every number on this page came out of your own Shopify, Google Ads and Google Analytics accounts. None of it is a benchmark.
- Your conversion rate fell from 1.4% to 1.03%.You were on category median. Something took 26% off it.
- It is freight, and we can date it.Flat rate went $10 → $12 → $15 across April. Express went $20 → $22 → $25 by June. That is up 50% and 25%, not the 20% mentioned.
- Every 0.1 of a percentage point is worth $1.09M a year.2,340,532 sessions, 26,262 online orders, $466 average order.
- Getting back to 1.4% is worth $3.8M a year.Same traffic. Same ad spend. Same warehouse. That is more than half your $7.2M growth gap.
- Freight is not the whole story, and we will not pretend it is.Checkout completion fell from 44.8% to 39.7% before any surcharge existed. That slow decline is the site. The sharp Q2 drop is freight. Two problems, two fixes.
- The freight fix takes four weeks and costs $30,000.The rebuild is a separate decision you make afterwards, once you have seen whether the first one worked.
What we heard
The real deadline is not a website. It is the agency contract ending in August, and four vendor relationships being rebuilt at once: paid media, creative volume, search, and the site.
The target is 20% topline two years running, $17.8M toward $25M. Growth has been flat a couple of years.
Underneath it: less dependency on agencies, more capability inside the building. The right engagement leaves Edwina and Anna running things we used to run.
Where the business sits today
Turnover. $11.9M online, $5.9M across four stores.
Average online order, from 26,262 orders in the last twelve months.
Online sessions a year. Roughly 195,000 a month.
Conversion rate. Down from 1.4%, which is the Shopify home and furniture median.
Every 0.1 of a percentage point of conversion is worth $1.09M a year.
One chart explains it.
Checkouts that were abandoned versus checkouts that completed, April to July 2026. Same store, same period, your own data.
Freight quoted, abandoned versus completed
The carts people walked away from were quoted 78% more freight, and those carts were 55% bigger.
Your freight model punishes larger baskets. So larger baskets leave.
That is the whole problem in one line, and it is the opposite of what you want a shipping model to do. Every incentive currently points the wrong way: the more someone buys, the more they are penalised at the last screen.
It is fixable without changing what freight actually costs you. The costs are fine. The way they are calculated and presented is not.
And it scales with the number
Time removed entirely. Same period, same store, same weeks. Every checkout split by the freight quoted, then measured for how many walked away.
Seven bands, no exceptions. Someone quoted over $400 is more than twice as likely to leave as someone quoted under $20.
Because these customers all shopped in the same weeks, seasonality, ad mix and traffic quality cannot explain the difference. The only thing that varied was the number on the screen.
Your furniture base rate is $249 in Melbourne and Sydney. NSW is 32.3% of your online orders and Victoria is 28.0%. Sixty percent of your online business now sits in a band where more than half of quoted customers walk away.
Four other systems say the same thing
We did not want to rely on one source. These are four separate measurements that have nothing to do with each other, all pointing at the same window.
Shopify's own checkout funnel
Sessions reaching checkout stayed flat all year. Sessions completing checkout fell from 44.8% to 32.8%. Cart additions did not move. The loss is entirely at the final step, which is where freight appears.
Your order history, showing the exact price changes
Flat rate shipping was $10.00 from June 2025 through March 2026. It became $12.00 in early April, then $15.00 by late April. Express went from $20.00 to $22.00 in April, then $25.00 in June.
That is +50% on standard and +25% on express, across three separate changes in three months. Average freight per online order went from $28.50 in March to $42.73 in June.
Google Ads, paid conversion rate
Paid conversion rate by month: January 2.14%, March 2.32%, May 2.27%, June 2.09%, July 1.44%. July is the worst month in twelve. The same weeks last year ran at 1.97%, so this is not seasonal.
Cost per conversion went from $43 in May to $70 in July on flat spend. Roughly 180 conversions a week not landing.
Google Analytics, checkout starts against orders
Comparing February–March against June–July: checkout starts per day are up 10.8%. Completed orders per day are down 10.8%.
More people are starting checkout than ever. Fewer are finishing. The gap widened by about 21% between those two windows.
Stated plainly so you can weigh it: Google Ads conversions are not the same as Shopify orders, July is a partial month, and Performance Max attribution is opaque. We have deliberately used four independent sources rather than leaning on any one of them.
How we know this
Between Tuesday's conversation and this document, we read the following. All of it read only. Nothing on your store was changed, and no customer data left your systems.
We have shown you the working because the conclusion is uncomfortable. It is easier to accept a $3.8M number when you can see where it came from, and easier to disagree with us if you think we have it wrong.
All of it is yours. The analysis, the scripts we wrote to do it, and the raw findings. Whether you work with us or not.
Ranked by what each one is costing you.
Six findings from your Shopify admin, your order and product exports, and the shipping app itself. Biggest first.
Freight rose 50%, in three steps, over three months
Standard $10 → $12 → $15. Express $20 → $22 → $25. Conversion tracked it down. This is the single cause of the revenue gap, and reversing the presentation costs nothing in real freight spend.
Your furniture rate lands exactly where customers leave
3,054 variants carry no weight in Shopify. That is deliberate, not an error: those products route to the LBD Unified Shipping profile and take a flat furniture base rate instead of a weight calculation. We initially read this as a margin leak. It is not, and Tim was right to challenge it.
What it does show is the shape of your furniture pricing. Median freight on those orders is $249, against $15 on the weight-based homewares side. Nothing sits between. And $249 falls in the band where 51.7% of quoted customers abandon.
The two halves of your catalogue are priced by two different systems, and there is no middle ground for a customer buying one chair.
1,022 orders were charged freight twice
You have two fulfilment locations at the same address. When an order draws from both, Shopify calculates freight separately for each and adds them together. Confirmed by Living By Design: the freight micro service cannot combine rates across two origins.
Those orders averaged $173.18 against a normal $47.44. It shows in your order data as "Shipping 1 and Flat Rate Shipping".
This is solvable. It is a limitation of how the current rules are written, not of Shopify. Consolidating the origin logic is part of Rescue.
Mobile is 71% of your traffic and 56% of your checkouts
Normalised, a desktop visitor is 1.95 times more likely to reach checkout than a mobile visitor. Not convert. Just reach checkout. Nearly half of mobile intent is lost before anyone sees a freight number.
Google Analytics has never recorded a single sale
Your GA4 checkout journey returns zero for add shipping, add payment and purchase. Only the first step fires. There is no revenue in GA4 at all.
Which means every GA4 figure anyone has shown you is unreliable, and there has been no second source to check Google Ads against. That is part of why the ROAS numbers went unchallenged for two years. Two hours to fix.
Quiet accumulation across the storefront
Twelve lookbooks live in the navigation going back to Autumn Winter 2020. Menu items pointing at nothing. Ottomans and Benches appearing in five places across two different URLs. Your About page says five stores, your Locations page says four. Both live.
None of these are disasters alone. That is exactly the point. On a catalogue your size they accumulate until the store is harder to shop than it should be, because nobody is watching.
One thing that is working
1,817 orders were created manually last year. $2.1M. That is 9.3% of your revenue at an average order value of $1,166, nearly three times your online average.
That is your corporate, hospitality and trade business, and it is running entirely on staff typing orders in by hand. Around 140 a month. You are also processing trade discounts through generic discount codes, 436 orders and $180k of them.
It works, which is the impressive part. But it is a $2.1M revenue line with no system behind it, and it does not scale past the people currently doing it.
Whatever the full audit finds, the list is yours. Take it to any agency you like, including the incumbent. If someone else fixes it faster and cheaper, that is a good outcome for you and a useful signal for us.
A technology partner, not a fifth agency.
You asked what we are actually selling. Here it is, including the parts we are not selling.
Five workstreams
Freight, product data and checkout
The rate engine rebuilt from your own order history. 3,054 product weights remediated at source. Multi-origin duplication removed. Cost shown before the cart, not at the payment step.
Site rebuild and integrations
Migration off the current Impulse theme onto Shopify's Horizon framework. Mobile first, because that is where the loss is. Cin7 sync fixed so product data stops arriving broken.
Continuous improvement and monitoring
Daily checks on the things that broke this year. Six weeks of falling conversion should never again run unnoticed by anyone.
Paid media
Referred to Matt at Scale Digital. We take no commission, and we do not sell media buying.
Search
Not a vendor problem. A documented workflow handed to Edwina, run in house, reviewed with us quarterly.
You will own everything
You paid $75,000 for a shipping calculator you cannot open, change or take with you. We are not going to do that to you.
Every line of code, every configuration, every credential and all documentation is yours. Assigned to you in writing on final payment, not licensed. It lives in a repository you control. If you replace us, you keep working without asking us for anything, and we will hand over cleanly.
That clause goes in the contract, not just in this document.
Who is doing the work
Sonder is small on purpose. Tim Sullivan and Ying Chen, plus specialists where a build needs them. You will talk to the people writing the code, not an account manager relaying messages.
The relevant precedent is RMIT. Ying and Tim built and now run Shopify platforms across three separate business areas there, together turning over more than $100M a year. Different sector, same problem: large catalogues, multiple customer types, complex fulfilment, and internal teams who need to run it themselves afterwards.
The honest version: a large agency will put more bodies on this than we will. What you get instead is the two people who found everything in this document doing the work, and a much shorter line between a problem and the person who can fix it.
How it fits your August deadline
Decision and access
Confirm the pathway. Read-only access is already in place, so nothing else is needed to start.
Freight engine and product data
Rate logic rebuilt from your order history. Weights remediated across 3,054 variants. Multi-origin duplication removed. GA4 purchase tracking installed.
Presentation and measurement
Freight shown on product pages and in the cart. Delivery expectations made explicit. Baseline measurement in place so the effect is visible, not assumed.
Your contract ends with nothing broken
The freight problem is fixed and measurable. Whatever you decide about the rebuild, you are not carrying an unsolved conversion problem into a vendor transition.
The rebuild, if you want it
Ten to twenty weeks depending on pathway. A separate decision, made with four weeks of evidence behind it rather than a pitch.
Start with the four week fix.
Rescue stands on its own. Everything after it is a decision you make once Rescue has either worked or it hasn't.
- Freight engine rebuiltFrom your own twelve months of order data. Owned by you.
- 3,054 weights remediatedFixed at source so they stop coming back wrong.
- Multi-origin bug removedNo more doubled rates.
- Freight moved earlierProduct page and cart, not the payment step.
- GA4 tracking installedSo you can see it working.
Target: 1.03% back to 1.4%.
- Everything in Rescue
- Horizon migrationHomepage, product, collection and cart templates.
- TrainingEdwina and Anna on the block system.
Does not touch Cin7 or B2B.
- Everything above
- Full rebuild, mobile firstEvery template. Mobile is 71% of traffic and 56% of checkouts.
- Cin7 sync fixed at sourceProduct data stops arriving broken.
- Conversion program1.4% recovery, then toward 2.0%.
- Navigation and catalogue rebuilt3,673 products, structured properly.
- Twelve months monitoringIncluded.
Rescue runs first, inside this.
- Everything in Growth
- Trade and hospitality systemReplaces 1,817 manual orders and $2.1M of typing.
- Retail and online unifiedOne view of stock and customer across four stores.
- Quarterly strategic review
The version that scales past your people.
Rescue is fixed price. If conversion has not moved by week six, we both walk away and you are under no obligation to proceed to the rebuild. You keep the freight engine, the remediated product data and the tracking regardless.
We can offer that because we already know what is wrong.
Payment
50% to commence, 25% on go-live, 25% on your sign-off. Not before you have seen it work.
Ongoing monitoring
After the included twelve months. Month to month, no lock-in. Cancel the moment it stops making you more than it costs.
Ownership
Code, data, configuration, credentials and documentation. Assigned in writing on final payment.
For context. $165,000 is fourteen percent of your annual Google spend, or about six weeks of ads. Each 0.1 of a conversion point is worth $1.09M a year, and the recovery on the table is $3.8M.
Fire us the moment we stop making you more than we cost.Start the conversation →
Five things left to confirm.
None of them block Rescue starting. They do affect how accurately we can scope the rebuild. Tap any item to tick it off.
Ticks save on this device only. Nothing is sent anywhere.
Send these through →