Handmade sterling silver jewellery on Shopify — seaglass and abalone rings, wayfinder pendants, the flower collection. Marketing is Instagram plus paid social, scaled hard from November 2025. Your baseline snapshot: where you're at, and the line everything from here gets measured against.
Real growth: four times the visits of a year ago (1,652 a month in July 2025, 6,769 in June), and June was your best month ever for orders, 35, on 38% less traffic than March. Revenue is up 21% on the quarter and the average order is a healthy £83. What has not kept up is the share of visitors who buy: 1.15% in July 2025, 0.52% in June, because half of today’s visits come from ads that bring no recorded sales, and phone shoppers give up at checkout. Your own visitors still buy at 0.80%, close to a healthy jewellery site. Fix who arrives and where they drop out, and the traffic you already have turns into far more orders. That is what this audit is about.
The green line is visits; the dashed line is the share of visitors who buy. Conversion first dipped in August 2025 . The two move in opposite directions from November 2025 , which is the month paid social scaled. Traffic bought is not traffic earned, and the conversion line is the price of that.
March 2026 is the month to study rather than celebrate: a record 11,009 sessions, and a 0.29% conversion rate — the second worst on record. June did 38% less traffic and 9% more orders than March.
Everything in this audit is ranked by how much it moves these. Figures are the last 90 days
More of the people already visiting go on to buy
Side goal Keep visits level or growing while the rate climbs · now 19,962 a quarter
The site held above 1% for six months in 2025, so this is getting back to where it was, not a stretch
Grow a list of people you can sell to again without paying for the visit
People who arrive from your emails buy at 5.00%, more than ten times the site average
People staying longer, looking at more pages and coming back
Could also be a goal Bigger baskets · average order £83.62 Time on site · 1:12 a visit Page interactions · 2.6 pages a visit
Everything an owner needs: what happened, why, and what to do. About five minutes.
The long view: every month since launch, so this quarter sits in context rather than on its own.
In short: traffic is up 11× and revenue is real — £25,401 across 390 orders in two years. The single biggest challenge is conversion rate: the site hit 1.97% in Dec 2024 and has since dropped to 0.3–0.5% as paid social scaled. More visitors are arriving, fewer are buying in-session. The good news is this is recoverable — and this audit is your starting line. Everything we do from here gets measured against it.
The five things that matter, easiest first and the big one last — if you read nothing else, read this.
So: your Instagram visitors already buy. Send the bio and story links to the collections, not the homepage.
70% of Instagram visits land on the homepage. Those landing on a collection bought at 1.2%. Instagram is your biggest free source: 4,991 visits and 22 orders this quarter.
So: rework the Abelone Ring page first. It gets the most eyes.
Your best piece, the Silver Ocean Ring, does 8.8%.
So: ask for an email earlier, so more left baskets can be chased.
Of the 10 email orders: 5 came from basket reminders, 2 from the welcome emails, 1 from the thank-you email after a purchase. The newsletters brought 2, from 1,230 sent.
So: fix the Meta tracking before spending more, then set the ads to chase purchases.
Ads bring half your visits, but almost none of their sales are recorded, so nobody can tell which ads earn their money.
The upside: if those 10,182 ad visits bought at just 0.3%, that’s about 30 more orders and £2,550 a quarter.
https://ruletheworld.co.uk/collections/rings?utm_source=meta&utm_medium=paid_social&utm_campaign=rings_prospecting&utm_content=carousel_seaglassSo: the biggest prize of the five, so it comes last: do the four above first, then this, starting with the product page and checkout.
Only 1 in 3 phone shoppers who start checkout finish (36 of 108), against nearly half on computers.
The upside: if phone checkouts finished as often as computer ones, that’s about 17 more orders, roughly £1,400 a quarter.
The questions owners ask first, each answered in one line. Open the detail for the numbers behind it.
Analyst comment: Over the last 90 days, half of all visits came from paid social ads, and not one of those visits bought anything. Every order came from people who found you another way: Instagram, Google, typing your address or an email. The ad spend is buying visits; the sales come from your own audience.
So: move the ad money to what sells: Instagram, Google and email.
| Channel | Visits | Orders | Conv rate |
|---|---|---|---|
| Paid social | 10,182 | 0 | 0.00% |
| Organic social | 4,991 | 22 | 0.44% |
| Direct | 1,997 | 23 | 1.15% |
| Paid search & display | 1,994 | 13 | 0.65% |
| Organic search | 598 | 10 | 1.67% |
| 200 | 10 | 5.00% |
What it means: every order this quarter came from people who already follow you, search for you or type your address.
Analyst comment: Of the 5,611 people who looked at a piece, only 412 put it in the basket. That is the biggest drop on the site, bigger than every step after it put together. Once a piece is in the basket, about half go on to start paying, so the product page is where most sales are lost.
So: fix the product pages first: size guide, delivery date, a photo of it being worn.
| Step | Phones | Computers |
|---|---|---|
| Look at a piece, then add it to the basket | 6.4% | 9.3% |
| Basket, then start paying | 44% | 54% |
| Start paying, then enter card details | 56% | 68% |
| Card details, then bought | 60% | 72% |
| Baskets that end in a sale | 15 in 100 | 26 in 100 |
What it means: the product page loses the most people on both. After that, phones keep losing people right through checkout. 334 baskets were left this quarter, which is why the basket reminder email sells.
Analyst comment: Compared with January to March, the site had 5,794 fewer visits but exactly the same 78 orders. Each order was worth more, £83.62 against £68.86, so takings rose 21%. Losing those visits cost no sales.
So: keep cutting the visits that don’t buy, and grow the ones that do.
Fewer from every channel, ads included
The same sales from fewer visits
Because each order was bigger
Still below normal for jewellery (0.87% and up)
Buyers are choosing pricier pieces or more of them
| Jan–Mar | Apr–Jun | Change | |
|---|---|---|---|
| Visits | 25,756 | 19,962 | −22% |
| Ad visits | 12,622 | 10,182 | −19% |
| Orders | 78 | 78 | level |
| Revenue | £5,371 | £6,522 | +21% |
| Conv rate | 0.30% | 0.39% | +0.09pts |
| Avg order | £68.86 | £83.62 | +21% |
What it means: fewer ad visits lost no orders, and each order is worth more.
Analyst comment: Money per visit is what each visitor brings in on average. An email visit brings in £4.51 and a Google visit £1.32, while an ad visit brings in nothing, even though each one costs about 15p. Money spent on email sign-ups and Google goes much further than money spent on ad clicks.
So: spend on email sign-ups and Google, not on ad clicks.
By device: Desktop £0.69 · Tablet £0.32 · Mobile £0.19
| Channel | Visits | Revenue | £ per visit |
|---|---|---|---|
| 200 | £903 | £4.51 | |
| Organic search | 598 | £790 | £1.32 |
| Direct | 1,997 | £1,980 | £0.99 |
| Paid search & display | 1,994 | £1,117 | £0.56 |
| Organic social | 4,991 | £1,715 | £0.34 |
| Paid social | 10,182 | £0 | £0.00 |
What it means: an ad visit costs about 15p and brings nothing back. One email visit brings in what 30 ad visits cost.
Analyst comment: Where someone first lands decides whether they stay. People who land on a collection stay and browse 63% of the time and buy at about 1.5%, while most people who land on the homepage leave. Most visits land on the homepage today, so pointing links at the collections is the quickest win.
So: send ads and Instagram links to the collections, not the homepage.
| Homepage | Collections | |
|---|---|---|
| Visits that start here | 13,969 | 3,367 |
| Stay and look around | 33% | 63% |
| Buy | 0.09% | 1.49% |
| Money per visit | £0.07 | £1.29 |
What it means: the collections already work. The fix is getting people to them first.
Analyst comment: Each figure shows how many views a piece needs to make one sale. The Freeform and Silver Ocean rings sell once every 23 to 24 views, while the Abelone Ring, your most-viewed piece, needs 285. Showing the sellers first on the homepage, in ads and in posts turns the same visits into more sales.
So: lead the ads, the homepage and posts with the Freeform and Silver Ocean rings.
| Piece | Views | Sold |
|---|---|---|
| Abelone Ring | 1,140 | 4 |
| Abelone Wayfinder Pendant | 815 | 3 |
| Sterling Silver Freeform Ring | 371 | 16 |
| Silver Ocean Ring | 306 | 13 |
What it means: the two Abelone pieces get the most views and sell the least. Showing the sellers first costs nothing.
Analyst comment: Two in three visitors browse on a phone, but phone visitors buy at 0.31% against 0.94% on a computer. Phones lose more people at every step, from adding to the basket to entering card details. A quicker phone checkout is where the biggest gain is.
So: make paying on a phone one tap: Apple Pay and Shop Pay at the top of checkout.
| Step | Phones | Computers |
|---|---|---|
| Look at a piece, then add it to the basket | 6.4% | 9.3% |
| Baskets that end in a sale | 15 in 100 | 26 in 100 |
What it means: phones lose more people at every step, from the product page to the payment screen.
Every device converts below what's normal for a jewellery site, and phones are furthest off. They bring two thirds of your visitors but just over a third of your revenue, so the mobile journey is where most sales are being lost
Below acceptable Acceptable Normal for jewellery Compared with jewellery sites, which convert lower than most online shops · click a device to see it on its own in the chart above
One primary goal, two guardrails. The goal sets the headline numbers, the target below, and the order of the recommendations in section five.
page_view · view_item · add_to_cart · begin_checkout · add_payment_info · purchaseAnalyst's comments are hidden. Turn them back on with the switch at the top of the page.
Everything that matters is summarised above. The sections below are the full figures behind it, for anyone who wants to check the working. They start closed.
The six facts about this shop that shape every recommendation in this audit.
Points two and six are the same story told twice. Some share of paid social's failure is really direct's success: a visitor clicks an ad on Monday, comes back by typing the address on Thursday, and direct takes the credit. Direct's 1.15% is flattered; paid's 0.00% is impossible.
The honest read is that paid traffic converts worse than organic, but it is not literally worthless. Until the tagging is fixed, judge the ads on blended revenue against blended spend, not on the in-session figures in any dashboard.
Two channels dominate. Paid social brings half of all visits and almost no sales. Email is tiny in volume but converts at 5% — the most efficient channel on the site.
Pick any channel and device. The headline numbers and all four tables recalculate for that slice of the last 90 days.
Paid social on mobile is the biggest single slice: 7,126 visits, 75 baskets, 0 orders. On desktop it's 2,648 visits, 43 baskets, 0 orders. People pick pieces from the ads but never pay.
The other end: direct on desktop buys at 2.31% and organic search on desktop at 3.23%. These are people who already know you. Sorting by a rate hides anything under 20 visits, because a rate on four visits isn't a finding.
This finds the patterns — where visitors leave, which pages lose them, what each source of traffic is actually worth, and which change we would make first. Every recommendation below has the evidence for it printed next to it, so you can disagree with any of them on the facts.
It is not a line-by-line UX review. We are not auditing every screen, writing your copy or designing anything, and where we suggest a change that is a suggestion with a reason behind it rather than a specification. These are the things that have worked on sites with the same problem.
Which is also why Check 1 and Check 2 exist. A recommendation is a bet, and the checks are how you find out whether it paid — within a fortnight rather than a year.
70% of paid visits land on the homepage, which converts at 0.09%. /collections/rings converts at 1.63% and /collections/pendants at 1.59%. You are paying for the click and then spending it on a page that sells nothing.
Change the destination URL on every live ad to the collection the ad shows — rings ads to /collections/rings, pendant ads to /collections/pendants. Then judge each ad on orders, not clicks.
Do this one first. It takes an afternoon, needs no development work, and it is also the change that makes everything else measurable — right now paid and organic traffic both pile onto the same page, so you cannot tell a homepage problem from an ad-targeting problem.
The estimate assumes you capture a third of the gap between the two pages, not all of it. Someone arriving at a collection from an ad is still colder than someone who navigated there themselves.
The homepage asks the visitor to make a decision — which of five things am I here for? — at the exact moment they have the least patience. A collection page asks a smaller question: which of these do I like?
Paid social records £0 of in-session revenue on 10,182 sessions. That is a tracking failure, not a performance result. Direct's 1.15% is almost certainly borrowed from it.
Add consistent UTM parameters to every live ad URL, extend the GA4 session timeout to suit a considered purchase, and audit cross-domain tracking into the Shopify checkout.
No order estimate against this deliberately — it earns nothing by itself. What it does is turn every other decision on this list from a guess into a measurement. Jewellery is a considered purchase; a 30-minute session window was never going to capture a buyer who thinks about a £100 ring for three days.
5,611 product views produced 412 baskets — 7.3%. The Abelone Ring was viewed 1,140 times and basketed 22 times (1.9%). Flower Ring Silver – A was viewed 457 times and basketed 3 times. A healthy jewellery product page sits at 8–12%.
Add a size guide, a delivery date and a photo showing scale on a hand to every product page. Check that stock counts are not reading empty on the two worst performers.
Start with the Abelone Ring. It gets the most views of any product on the site and converts the worst of the ones that sell at all, so it is both the biggest prize and the clearest test.
A 1.9% basket rate on 1,140 views is not a taste problem. People arrived at that page wanting to see that ring. Something on the page stopped them.
Email converts at 5.00% with the highest average order on the site (£102) — and it reaches 200 sessions a quarter. Roughly 10,000 paid visitors arrive, don't buy, and are never asked for an address.
Put a single capture on the site — 10% off a first order — and let the welcome and basket-reminder flows do the rest. They already produce 8 of your 10 email orders from 808 sends.
The flows are already outperforming everything else per send. This recommendation is not about writing better email; it is about having anyone to send it to.
One caution: a discount capture will pull your average order down slightly. At £102 against a site average of £83.62 there is room for that, but watch it rather than assume it.
This quarter mobile converts at 0.26% against 0.75% on desktop, and mobile is 70% of your traffic. 108 mobile shoppers started checkout and 36 finished (33%). Desktop finishes at 49%.
Do a real test purchase on a phone, on mobile data. Show the delivery cost before the address step, offer guest checkout, and make sure Apple Pay and Google Pay appear above the card form.
Test on a real handset rather than a desktop browser's phone emulator. The three most common causes — wallet buttons that don't render, a keyboard covering the submit button, and an address autocomplete that fights the form — are all invisible in emulation.
The biggest prize. Start it once recommendations 1 and 3 are live: a mobile checkout fix pays off most once more people are reaching the checkout.
What people look at, what they pick up, and what they actually buy.
Engagement is your sub-goal: 41% of visits engaged now, aiming for 55%+. Pick a channel and a device to see where those visitors land, and what each page does with them.
How to read the map: each bubble is a page people land on, sized by how many land there. Further right, more of them stay and look around. Higher up, more of them buy. Top right is where every landing page should be.
Visits, baskets, orders and £ per visit follow your picks (last 90 days). Bounce, engaged and time on page are for everyone who landed on that page: the data doesn't split those by channel or device.
| Where they land | Visits | Engaged | Conv rate |
|---|---|---|---|
| Homepage | 13,969 | 33% | 0.09% |
| Rings, pendants, seaglass, earrings, flower collections | 3,367 | 63% | 1.49% |
| All products | 1,981 | 59% | 0.71% |
| Abelone Ring (product) | 264 | 29% | 0.00% |
What it means: where people land decides whether they stay. The collections do the job; the homepage doesn’t, and it gets most of the visits.
Landing pages are only the first step. This follows people through the whole site, last 90 days.
Two places lose most people: the homepage, where 9,360 leave straight away, and the product pages, where 5,199 look at a piece and don’t pick it up. Once a piece is in the basket, most people carry on.
| Page | Views | Engaged | Time | What happens here | Verdict |
|---|---|---|---|---|---|
| Homepage/ | 24,180 | 33% | 0:34 | First impression | Loses people |
| All products/collections/all-products | 9,420 | 81% | 0:49 | Browsing | Holds people |
| Rings/collections/rings | 6,280 | 86% | 0:59 | Browsing | Holds people |
| Pendants/collections/pendants | 3,840 | 84% | 0:53 | Browsing | Holds people |
| Seaglass rings/collections/seaglass-rings | 2,110 | 89% | 0:53 | Browsing | Holds people |
| Earrings/collections/earrings | 1,760 | 83% | 2:15 | Browsing | Holds people |
| Flower collection/collections/flower-collection | 1,340 | 91% | 4:51 | Browsing | Holds people |
| Stacking rings/collections/stacking-rings | 980 | 85% | 3:35 | Browsing, 4 baskets | Holds, rarely sells |
| Abelone Ringproduct page | 1,140 | — | — | 1.9% add it to the basket | Loses people |
| Flower Ring – Aproduct page | 457 | — | — | 0.7% add it to the basket | Loses people |
| Silver Ocean Ringproduct page | 306 | — | — | 8.8% add it to the basket | Sells |
| Basket/cart | 820 | — | 0:51 | 48% start paying | Some leave |
| Checkout/checkout | 396 | — | 1:28 | 39% finish | Some leave |
Views count every time a page was seen, wherever it fell in the visit. Engaged here is for everyone who saw the page, so collections score higher than on the landing view. Product pages are shown by how many add the piece to the basket.
| Step | People | Carry on | Lost |
|---|---|---|---|
| Land on the homepage | 13,970 | 33% | 9,360 |
| Look at a piece | 5,611 | 7% | 5,199 |
| Basket | 412 | 48% | 214 |
| Start paying | 198 | 39% | 120 |
| Bought | 78 |
What it means: engagement isn’t low across the site. It’s low on the first page people see and on the product pages. Fix those two and both sales and the engagement goal move.
The same three questions asked of each page group: does it hold people, does it get them to a basket, and what is a session worth when it starts there.
Takes the most and does the least. Two thirds bounce, a third of a minute on page, and seven pence of revenue per session.
Wins on every measure. Worth 14× the homepage per session, and it gets 28% of landings. This is the whole argument for repointing the ads.
Reached from inside, never from outside. 5,611 views from 262 landings. The 7.3% cart rate is on warm internal traffic, which makes it worse than it looks.
Read the views against landing sessions row pair before anything else. Collections get 25,730 views from 5,590 landings — people arrive and browse. Products get 5,611 views from 262 landings — almost nobody arrives there directly, so the product pages are not an acquisition surface at all right now.
That has a practical consequence: SEO or ad work pointed at product URLs would be starting from nothing, while the same effort pointed at collections builds on a page group that already converts at 1.18%. Fix the product pages for the internal traffic they already get; send the external traffic to collections.
Your most-viewed product — 1,140 views — and a 1.9% basket rate. Four units sold. Every other ring that sells converts a basket at three to four times this rate.
Open it on a phone and try to buy one. Check the size selector, the delivery estimate and whether the photos show it being worn.
457 views, 3 baskets, zero sales all quarter. Meanwhile the main Flower Ring Silver converts at 7.1% from 562 views. Same design, same price band, wildly different result.
Most likely a duplicate or variant listing that shouldn't be indexed separately. Merge it into the parent product or unpublish it.
The pattern across the catalogue is clear: the pieces that sell convert a basket at 26–67%, and the pieces that don't sell never get into a basket at all. That is a page problem, not a pricing one — nobody is reaching the price and recoiling.
Two listings, Flower Ring Silver – A and the Wayfinder Pendant Only variant, took 623 views between them and produced nothing. They look like variant pages competing with their own parent products. Worth checking in Shopify before anything else.
Where people fall out between arriving and paying. Switch device to see where the journeys differ.
Two separate problems live in this funnel and they need different owners.
The product view → basket step loses 5,199 people, the largest drop anywhere on the site. 7.3% is roughly half what a working jewellery shop does. This is a product-page job and it is recommendation 3.
The checkout → payment step on mobile is smaller in absolute terms but it is the sharper failure: 108 start, 36 finish. Desktop finishes the same journey at 49%. That gap is worth about 19 orders a quarter on its own.
When orders actually happen. This is what should dictate when you post, when you send, and when the ad budget runs.
43% of orders come in the evening. The reliable base: keep the ad budget here before anything else.
The two best days: 0.54% and 0.51% of visitors buy. Thursday is the weakest at 0.21%.
A quarter of orders from 12% of visits: payday. The last days of the month, the 28th to the 31st, sell least.
The weakest time of day: 0.25% of visitors buy. Early-morning ad spend and sends work hardest elsewhere.
With 78 orders this grid is thin, and single cells are one or two orders, so don't read too much into any one square. The shape is still usable: evenings, Tuesdays and Saturdays, and the start of the month.
The practical use is timing the ads and emails. Run the ads hardest in the evenings and from the 1st to the 4th, ease off before 9am and in the last days of the month, and send emails on Tuesday or Saturday evenings. It costs nothing and needs no new adverts.
A caveat worth holding: this shows when orders complete, not when demand is created. An ad seen at lunchtime can produce an evening order. Treat it as a scheduling guide for email, and a softer signal for ads.
Every pound of ad spend and every send, against what it returned.
The flows versus newsletters comparison is the sharpest contrast here. 808 automated sends produced 8 orders. 1,230 newsletter sends produced 2. Automation is outperforming broadcast by roughly six times per send, on two thirds of the volume.
On the ads: seven of eight campaigns have no readable name, which is why section four cannot tell you which ones work. The one named campaign is also the only one you can learn anything from. Name the rest before moving any budget.
Every in-session return figure in the ads table should be read as a floor, not a result — the attribution problem means real returns are higher than these numbers show. That is the argument for fixing the tagging before cutting any campaign.
Traffic is up 11× in two years — from 600 sessions a month in mid-2024 to a peak of 11,009 in March 2026. Orders have followed: the site did 1–2 orders a month in its first six months and hit 35 in June 2026 alone. Revenue has crossed £25,000 across 390 confirmed orders — proof the product converts and customers are willing to spend. The challenge now is efficiency, not existence: paid social has flooded the funnel with low-intent traffic, pulling the blended conversion rate down from 1.2% to 0.39%. The foundations are solid. The next phase is converting more of the traffic that's already arriving.
The whole audit as a to-do list. The first three take about a day between them and cover most of the value.
In priority order.
With 10,182 paid social sessions this quarter, £0 of in-session revenue is not reflecting reality. Add consistent UTM parameters to every live Meta and TikTok ad URL so next month's report shows the true paid performance.
https://ruletheworld.co.uk/collections/rings?utm_source=meta&utm_medium=paid_social&utm_campaign=rings_prospecting&utm_content=carousel_seaglass
https://ruletheworld.co.uk/collections/pendants?utm_source=meta&utm_medium=paid_social&utm_campaign=wayfinder_retargeting&utm_content=reel_abelone
Naming convention: channel_audience_creative_date. Seven of your eight campaigns currently have none.
Rings convert at 1.63% and pendants at 1.59%, against 0.09% on the homepage. 70% of paid visits currently land on the homepage, which means every ad is losing a meaningful share of its conversions at the first click.
1,140 views and a 1.9% basket rate — your most-viewed product and one of the worst at earning a basket. Check the size selector, the delivery estimate, whether a worn photo appears, and whether the stock count reads empty.
Flower Ring Silver – A and the Wayfinder Pendant Only variant took 623 views between them and produced nothing. They look like variant pages competing with their own parents. Merge or unpublish.
Email is converting at 5.00% with a £102 average order — your best channel by a distance — and it reaches 200 sessions a quarter. A 10% first-order code shown to the 10,000 quarterly paid visitors who don't buy would give the flows something to work with.
First-time buyers, repeat customers and never-purchased subscribers shouldn't get the same message. Start with a repeat-buyer segment — they're the readiest audience you have, and the flows already prove automation outperforms broadcast five to one.
This quarter mobile converts at 0.26% against 0.75% on desktop, and mobile is 70% of your traffic. 108 started checkout on a phone and 36 finished. The checkout audit — payment method options and address entry — is the place to start.
Against the same five headline numbers at the top of this page, so the comparison is like for like. Expect the conversion rate to move before the revenue does.
If only three things happen this month, make them 1, 2 and 3. Together they cost about a day of work and they address both the measurement problem and the biggest page problem — which means Check 1 will be worth reading, rather than another round of guessing at paid performance.
Item 7 is the biggest prize, so start it once the quick wins are live. A mobile checkout rebuild feels decisive, but it only pays off once more people are reaching the checkout, and right now the loss is happening two steps earlier.
The site has gone through three distinct phases. Apr–Dec 2024 was quiet organic growth — low volume, learning what works, peaking at 1.97% in December. Mar–Jul 2025 was the conversion peak — above 1% for five straight months while traffic doubled. Aug 2025 onwards is the paid social scale-up — sessions grew 5× but conversion fell to 0.25–0.52%.
The quick wins are in and holding. Ads and Instagram land on the collections, the Abelone Ring page is reworked, and the Meta tracking is fixed, so ad sales are finally counted. Visitors stay longer and orders are well ahead of the audit’s pace. Next comes the phone checkout, the biggest gap left on the site.
Against what the audit quarter averaged over the same 14 days (its pace), with June’s underneath. June was the best month on record, so it’s the tougher test.
Two weeks after the quick wins went live, the line sits above the audit’s average: 20 orders against 12 at the audit’s pace. The lift came from where people land. Ads and Instagram now send them to the collections, and more of them stay and look. It’s two weeks against three months, so read it as a direction, not a result. Our advice: leave the pages alone so the next check reads cleanly, move a fifth of the ad budget to the campaign that’s working, and build the first part of the phone fix, still the biggest gain left.
June, then the 14 days since the changes, on one chart. The dashed line is where the changes went live.
Two weeks in: each of the audit’s five points, what was done and what it did so far
Done: bio and story links now go to the collections.
7 orders from 810 Instagram visits, about 3 at the audit’s pace. Best post: the 4 July Silver Ocean Ring reel, 380 visits and 2 orders in a day.
Next: nothing: keep posting, the links do the rest.
The cheapest win in the plan. Same posts, same followers, just a better first page, and twice as many buy.
Done: page reworked: size guide, delivery date, a photo of it being worn.
147 views and 1 sale: at the audit’s rate of 1 per 285 that would have been half a sale. The Silver Ocean Ring does 8.8%.
Next: leave it alone so Check 2 can read it cleanly.
Nearly double, but on 147 views. It’s heading the right way. Keep leading with the Freeform and Silver Ocean rings until it catches up.
Done: sign-up at the basket went live on 9 July: 10% off a first order.
A slight rise after five days. Email visitors still buy at about 4.7%, your best channel.
Next: send the 1 August payday email, see Best times below.
A slight rise after five days is about what we’d expect. The real test is the 1 August email, sent to a bigger list.
Done: Meta’s tracking connected in week one, so ad sales are counted; every ad link tagged; the campaigns told to go for purchases, not clicks.
Ad visitors who buy: 0.00% → 0.19%. And one campaign is working better: Wayfinder retargeting (ads to people who already visited) 2 orders from 380 visits (0.53%), Rings prospecting (ads to new people) 1 from 1,180 (0.08%).
Next: move a fifth of the budget to retargeting, same total spend.
The ads haven’t suddenly got better: their sales are finally being counted. Meta can now see who buys and find more people like them. Prospecting stays on, because it brings the new visitors that retargeting then wins back.
So far: the plan is written: add-to-basket and delivery date on the first phone screen, Apple Pay at the top of checkout.
A small rise from better landing pages, not the checkout: still only 1 in 3 phone checkouts finish.
Next: the first part of the phone fix goes live after this check.
The small rise comes from better landing pages, not the checkout. Phones are still where most sales are lost, which is why this is the next big job.
Two weeks in: the four measures against their goals, then what to do next.
The audit and Check 1, and where each measure should get to
Check 2 picks the story up from 15 July: the ads measured properly, email at the basket, and how payday went.
The change that works first, because it touches half the traffic and costs nothing.
Before the changes against the 14 days since
Ad visitors now see a collection instead of the homepage, and far more of them stay. With the tracking fixed, their sales now show: 3 orders from 1,560 ad visits. The campaigns now aim for purchases and Meta is still learning who buys, so expect this to climb slowly.
Your most-viewed piece, and the weakest at being picked up.
14 days, 147 views
The Silver Ocean Ring gets picked up at 8.8%. 147 views is a lead, not a fact: Check 2 will say whether it holds.
The audit found when your customers buy. The first two weeks matched it, so the next payday is the one to plan for.
The new pages went live on payday, and it showed
6pm to midnight, and still the busiest
Weakest is still Thu
People browse here, they don’t buy
Coming up to the end of July, hold back and save the ad budget. On Saturday 1 August, payday and your best day, send the email at 6pm and turn the ads up until Tuesday 4th · post around 5pm, before the evening rush · pause the ads before 9am
When: evenings, and the 1st to the 4th. Ease off before 9am and from the 28th.
How: each ad goes to the collection it shows, and any ad with no orders after 14 days is switched off.
When: around 5pm on Tues and Sats, a sneak peek on the 30th, and payday itself.
How: bio and story links go to the collections, leading with the Freeform and Silver Ocean rings.
When: Saturday 1 August at 6pm, then a reminder on Tuesday 4th to people who opened but didn’t click.
How: one piece per email with a link straight to it, and basket reminders within the hour.
Bars: visits a day · line: share who buy · 0.74% buy on the 1st to the 4th, 2.2× the rest of the month
| Day | Share of orders | Visitors who buy |
|---|---|---|
| Mon | 16% | 0.42% |
| Tue Best | 20% | 0.54% |
| Wed | 9% | 0.27% |
| Thu | 8% | 0.21% |
| Fri | 13% | 0.37% |
| Sat | 19% | 0.51% |
| Sun | 15% | 0.34% |
Darker = more visitors · bigger dot = more likely to buy · faded = too few visits to judge · UK time
Bars are orders; the orange dots are the share of visitors who bought that day of the week. In these two weeks Sunday had both the most orders (4) and the highest share who buy (0.87%). Two weeks is too small to overturn the audit’s 90-day pattern, Tuesday and Saturday, which is why the payday plan below still leans on them. Hours are not split here; the audit’s hour-by-day grid is below.
The audit’s 78 orders give the pattern; 20 orders in two weeks only confirm it. Check 2 covers 1–4 August and will show whether the timing paid.
With the tracking fixed and a fifth of the budget moved to the ads that were working, the ads brought 6 orders, where the audit quarter had none. The first part of the phone fix went live and more phone checkouts finish. The payday plan brought 13 orders in four days.
Against what the audit quarter averaged over the same four weeks (its pace), with June’s underneath. Visitors who buy: 0.65% in Check 1, 0.73% now. Ad figures now come from fixed tracking, so they’re more complete than before.
The line has climbed in two steps. Check 1’s quick wins lifted it first, and these four weeks lifted it again: 44 orders against 24 at the audit’s pace, with the share of visitors who buy rising alongside, so the extra orders come from a site that sells better, not just from more visitors. The ads and the first part of the phone fix did most of that, and payday did the rest. Our advice: finish the phone fix on the product page next, hold the ad budget where it is, and run the payday plan again in September.
June, Check 1’s two weeks, then these four weeks.
Six weeks in: each of the audit’s five points, what was done since Check 1 and what it did
Since Check 1: nothing new: the links stay on the collections.
16 orders from 1,690 Instagram visits, more than twice the audit’s 0.44%. Best post: the 24 July Freeform carousel, 3 orders in two days.
Next: keep posting, around 5pm on Tuesdays and Saturdays.
Still more than twice the audit’s rate, with no extra work. Keep posting; the links do the rest.
Since Check 1: left alone, so the change could be read cleanly.
From 410 views: 1 sale per 150 views → 1 per 120. The Silver Ocean Ring does 8.8%.
Next: take the Abelone Ring out of the ads that show it.
Still climbing, but well short of the best sellers. Taking it out of the ads stops paying to send people to the weakest page.
Since Check 1: the payday emails on 1 and 4 August; the sign-up stays at the basket.
Orders from email 3 (June’s pace) → 6: 3 from the payday emails, 3 from basket reminders.
Next: repeat the payday emails on Tuesday 1 September.
Sign-ups rose only slightly. The payday emails earned 3 orders, so timing mattered more than list size.
Since Check 1: a fifth of the budget moved to retargeting (ads to people who already visited), same total spend.
Retargeting (ads to people who already visited) 4 orders from 620 visits (0.65%), prospecting (ads to new people) 2 from 1,860 (0.11%). Both up.
Next: hold the total budget until the ads earn twice what they cost in profit (profit is about half of each sale).
Moving a fifth of the budget to retargeting paid off: it brought 4 of the 6 ad orders. The ads now cover their cost in sales but not yet in profit, so the total budget holds.
Since Check 1: first part of the phone fix live on 22 July: Apple Pay and Shop Pay at the top of checkout, add-to-basket on the first screen.
Phone checkouts that finish: 33 in 100 → 42 in 100. On computers, 1.15% of visitors buy.
Next: finish the phone product page: the delivery date on the first screen and fewer fields.
The first part of the phone fix is working: more phone checkouts finish. Phones still buy at half the rate of computers, so the product page is next.
Six weeks in: the four measures against their goals, then what to do next.
The audit, Check 1 and Check 2, and where each measure should get to
The Full Summary puts the audit, Check 1 and Check 2 together, with the next steps in full.
The change the plan was built around.
June’s pace against the four weeks since
Meta also counts people who saw an ad and bought later. The truth sits between the two, and both say the same thing: the ads now sell, slowly.
The ads now bring in about what they cost in sales: £510 from about £520 of spend. In profit they’re still below cost, so the budget stays where it is for now.
Catching addresses at the basket, so left baskets can be chased.
June’s pace against the four weeks since
Check 1 planned the 1st to the 4th of August: hold back, then email and turn the ads up. Here’s what happened.
1.31% of visitors bought: the best four days since launch
Nothing was lost by holding back
Half of all email orders this check
Same budget, spent on the days that sell
Do it again. Tuesday 1 September is payday and your best day: ease the ads back from the 28th, send the email at 6pm on the 1st, a reminder on Friday 4th, and turn the ads up for the four days
One payday is one result: 13 orders in four days is a strong sign, not proof. The push kept giving after it ended: two quiet days, then 11 orders in the week on normal visits, as the retargeting ads and the reminder email did their work with the ad spend eased back. September will confirm it.
Six weeks in, the site is selling better and the ads have started to earn. The biggest gain left is phones: two in three visitors, and paying is where most of them give up.
Everything below is in order. The phone checkout comes first because it’s the biggest gain left.
Five measures: the audit, each check, and the goal for the next one.
| Measure | Audit | Check 1 | Check 2 | Goal | Status |
|---|---|---|---|---|---|
| Orders a weeksix-week average | 6.0 | 10.0 | 10.7 | 12 | On track |
| Orders from adsin the period | 0 | 0 | 6 (9 over six weeks) | twice their cost in profit | Earning |
| Visitors who stay and look | 41% | 48% | 50% | 55% | On track |
| Phone checkouts that finishin every 100 | 33 | – | 42 | 50 | Next fix |
| Own visitors who buynot from ads | 0.80% | – | 1.10% | 1.2% | Healthy |
Green means the number is moving the right way and on pace for the goal. “Next fix” is the one the actions below are for.
Five things, in this order, each with a date.
Apple Pay and Shop Pay are at the top now. Next: the product page, fewer fields, and the delivery cost before the address step. Phones are two thirds of visitors and still lose the most people at checkout.
The ads now bring in about what they cost in sales, and about half of it in profit. Raise the budget only when they earn twice their cost in profit for four weeks running.
Freeform and Silver Ocean rings first in ads and on the homepage; the posts already lead with them. Keep the Abelone Ring on sale, out of the ads.
Linked straight to it. The newsletters sold 2 orders from 1,230 sends at the Baseline, while the payday emails sold 3 in four days.
Tuesday 1 September is payday and your best day. Ease the ads back from the 28th, send the email at 6pm on the 1st, a reminder on Friday 4th, and turn the ads up for the four days. One payday gave 13 orders in four days; a second makes it a pattern.
That is the whole list. How to judge the ad budget is in the Full Summary. Below: the two ways to run it from here, and the five reports on one line.
The check is small and repeatable, and it doesn’t have to be ours.
The method is written out in full on the left, because a checklist you can’t see isn’t a service. What the monthly fee buys is that the job gets done in a busy month, by someone who has read the whole story.
Each one built on the last. The next check starts a new line on the same chart.
Six weeks from now we check again: the same five measures, against the goals above. Your analyst will email you to set the date.
Across Check 1 and Check 2, orders went from 36 to 64 and money in from £3,043 to £5,460, against what the audit quarter was doing over the same 42 days, on about the same number of visits. The ads now sell, slowly. The biggest gain left is finishing the phone checkout, the first of the next steps.
Compared with what the audit quarter was doing each day, over the same 42 days, so both sides are measured the same way.
Check 1, two weeks: the quick wins are holding
Check 2, four weeks: every measure is up
The pale part is the audit quarter, and the dashed line is its average. C1’s quick wins lift the line straight away; C2’s ad and email changes lift it again. The first part of the phone checkout fix landed inside C2, on 22 July; the rest comes next.
Six weeks in, across both checks: what’s working, what isn’t yet, and what happens next.
The six weeks since the audit against the audit quarter’s pace over the same 42 days.
1 July–11 August 2026 against the audit quarter’s pace
Each of the audit’s five points after both checks: what was done, and what it did to conversion.
All five points acted on in six weeks, and all five moving the right way. The share of visitors who buy went from 0.39% to 0.70% without a redesign. Phones are still the biggest gain left, which is why finishing the phone checkout is the first of the next steps.
A simple rule, so the budget never moves on a hunch.
The four weeks to 11 August 2026 · 52% of each sale is kept as profit (your figure)
Ads earn about half their cost in profit; the bar is twice their cost. So the budget holds. Re-check in six weeks, after the phone checkout.
For a year the honest answer to “should we spend more on ads” was no, because the ads brought people who didn’t buy. That’s no longer the reason. The ads now sell; they just don’t pay for themselves yet. That’s fixed with targeting and a better phone checkout, not with more money.
The check is small and repeatable, and it doesn’t have to be ours.
The method is written out in full on the left, because a checklist you can’t see isn’t a service. What the monthly fee buys is that the job gets done in a busy month, by someone who has read the whole story.
Each one built on the last.
Finish the phone checkout first, hold the ad budget until the ads earn twice their cost, and repeat the payday plan on 1 September.