TagPup — teardown

Gradient · teardown · tagpup · 27 August 2026

as the store sets it

A US personalised dog-collar store walked end to end at two widths, logged out, from an Indian residential address: what market it quotes, which countries it will actually ship to, and what its product form puts in the basket

Screens: 17 captured
Leaks: 9 measured
Walked on: a phone and a laptop
Store: tagpup.com

Every screen we captured9 findings across 17 screens — 9 of them carrying one

The buy box, with the required personalisation field empty and Add to cart livephone1 finding

The phone header: no categories, and no search a thumb can reachphone1 finding

The first step of checkout: rupees above, three countries below, none of them the visitor'slaptop1 finding

The category cards, including one whose button leads nowherelaptop1 finding

Shipping & Returns, openedlaptop1 finding

The basket total, in rupees, with delivery deferredlaptop1 finding

A shopper searching for the shipping policylaptop1 finding

Holiday collection cards whose titles and destinations disagreelaptop1 finding

The basket, after pressing that Add to cart with the field still emptyphoneevidence only

The same first step of checkout at phone widthphoneevidence only

THE CONTROL: the same product, one cookie later, in the market the checkout can servelaptopevidence only

The homepage, pricing its lead product in rupeesphoneevidence only

The same homepage, promising a shipping threshold in dollarsphoneevidence only

A search that cannot match anything, for comparisonlaptopevidence only

The whole catalogue, opening in alphabetical orderlaptopevidence only

01/9

The required personalisation field can be left empty, and two of the three choices never reach the basket.

high severityfix: an app setting and one renameon a phone

WhereThe collar page, at the box where the dog's name goes.WhyThe field is marked required and Add to cart works with it empty. The basket then records the size, the pattern and a font — and no name, and no buckle colour.What it doesA shopper pays for a personalised collar and has personalised nothing. Nobody is told, so the first person to find out is whoever opens the parcel.

Fix this firstFix the field's name before anything else. Until it is renamed, nothing a shopper types in that box can reach you, however much checking you put in front of it.

Your page, today — unretouchedmeasured

a phone

shot 1The buy box, with the required personalisation field empty and Add to cart live

The measured region on https://www.tagpup.com/products/personalized-money-dog-collar-with-leash-adjustable-bohemian-style — the region this finding is about

The measured region on https://www.tagpup.com/cart — what the basket kept

What we saw

The field is marked required and the buy button works with it empty.
The basket keeps the size, the pattern and the font — and no name, and no buckle colour.
A shopper can buy a personalised collar carrying no personalisation.
Nothing anywhere on the way through tells them that happened.

Why it matters

Three separate things are wrong, so checking the field alone will not fix it.
The font arrives because it is named the way Shopify expects and has a default already chosen.
The buckle colour is named right and still arrives empty, because neither swatch starts selected.
The name is discarded on the way to the basket however carefully it is typed.

Show me the evidence

In your store’s own words

Stella 123-456-7890the placeholder in the empty Customization field
Add to cartthe buy button, enabled with that field empty
Collar Size (Refer to Size Chart): Leash Onlythe basket line, which records the size and not the name

Quoted character for character from the capture — check any line against your own page.

Measured on a phone, on the buy box, with the required personalisation field empty and Add to cart live. It is written up as pdp-01 in our full audit of this store — ask us and we will send it.

what was measured — 5 readings

properties[Text Font] — radio, required, value Font 1 pre-checked. Reaches the basket.
properties[Buckle Color] — radio, not required, 0 of 2 checked when the page opens. Does not reach the basket.
Customization — text input, not required, empty. Named Customization, not properties[Customization]. Cannot reach the basket.
The basket line after a real press of Add to cart: {"Text Font":"Font 1","_has_gpo":"120248"}, variant Leash Only / 1, price 160000.
The product's real Shopify options are two: Collar Size (Refer to Size Chart) and Collar Pattern. The other three are drawn by a product-options app.

02/9

A sign-up modal covers the whole phone screen as soon as the visitor touches anything.

high severityfix: an app settingon a phone

WhereThe homepage, on the first thing a thumb touches.WhyThe sign-up modal stays away through a minute of stillness and opens within seconds of the first press. On a phone it and its dimmer cover the screen.What it doesThe visitor is asked for an email before they have seen a collar they could want, with no page left visible behind it to go back to.

Fix this firstLet them see one collar before you ask for the email. Moving the offer is a smaller change than removing it, and it is the one worth trying first.

Your page, today — unretouchedhand-placed

a phone

shot 2The sign-up modal, over the phone homepage

What we saw

It asks for a name and an email before the visitor has seen a product they can decide about.
A full minute of no interaction produced nothing; the first press anywhere brought it up.
So everyone who touches the store gets it, and nobody who only looks does.
On a phone the modal and its dimmer take the whole screen — there is nothing left to go back to.

Why it matters

The offer may well be worth making. The timing is the part that is not.
It trades the decision they came to make for the one you would rather they made.
On a phone there is no visible page left behind it to return to.

How this happens, in order2 steps

A minute of stillness on the homepage and nothing happens. Touch anything and this arrives, over the collar and its price.
The same modal on a laptop, where there is a page left around it. On a phone there is not: it and its dimmer are the whole screen.

Show me the evidence

In your store’s own words

Get a Free Leash!the modal heading, over the phone homepage
Complete the form to sign up and the code will be revealed!the line under it

Quoted character for character from the capture — check any line against your own page.

Marked by hand on a phone, on the sign-up modal, over the phone homepage. It is written up as home-01 in our full audit of this store — ask us and we will send it.

what was measured — 3 readings

Captured with overlay dismissal deliberately turned off, so the board shows the modal in position rather than a screen it had been cleared from.
Not seen through a 60s wait with no interaction, at 390 and again at 1440; opened within seconds of the first press on the page in every run that pressed anything.
Rendered inside a shadow root on SHOPIFY-FORMS-EMBED. Its close control carries aria-label "Close modal". Escape closes it.

03/9

The phone header offers Log in and Cart. The search box is built into it and never drawn.

high severityfix: a theme changeon a phone

WhereThe phone header, where a shopper looks for the way in.WhyIt holds two links: Log in, and Cart. The search control, its panel and its icon are all in that header and all measure as not visible.What it doesSomeone who arrives wanting a Christmas collar has nothing to type into and nothing to tap. They scroll 104 products, or they leave.

Fix this firstDraw the search control that is already in your header. It is a matter of showing it, not of building it.

Your page, today — unretouchedmeasured

a phone

shot 3The phone header: no categories, and no search a thumb can reach

What we saw

The phone header holds two links: Log in, and Cart.
No category menu, and no search a thumb can reach.
The only way into any of eight collections is to scroll the homepage to the category row.
The search button, its panel and its icon are all in that header, and all measure as not visible.

Why it matters

A visitor who arrives knowing what they want has nothing to type into and nothing to tap.
They scroll or they leave, and on 104 products scrolling is the slower of the two.
The feature is already built and paid for. It is simply not on screen.

Show me the evidence

In your store’s own words

Log inthe phone header, first link
Cartthe phone header, second and last link

Quoted character for character from the capture — check any line against your own page.

Measured on a phone, on the phone header: no categories, and no search a thumb can reach. It is written up as nav-01 in our full audit of this store — ask us and we will send it.

what was measured — 3 readings

Header links at 390: Log in, Cart. No category link.
search__button (aria-label Search), predictive-search--header and the header search icon are all present in the header and all measure visible: false.
104 live products; six homepage-linked categories, which is every category the row makes reachable.

04/9

The prices are in rupees. The checkout ships to three countries, and none of them is this shopper's.

high severityfix: two settingson a laptop

WhereCheckout, the moment they are asked for an address.WhyEvery screen from the homepage to the basket quotes rupees. Checkout then offers Canada, Mexico and the United States. That is the entire list.What it doesThere is no address this shopper can complete, and they find that out after choosing a collar, personalising it and reaching payment.

Fix this firstDecide which market you meant. Either add the shipping zone that completes the rupees, or turn the Indian market off — but not neither.

Your page, today — unretouchedmeasured

a laptop

shot 4The first step of checkout: rupees above, three countries below, none of them the visitor's

The measured region on https://www.tagpup.com/checkout — the region this finding is about

What we saw

Rupees on the homepage, the product page, the basket and the first screen of checkout.
The checkout's country list is Canada, Mexico and the United States. That is all of it.
The shopper being quoted is in none of the three, so no address can be completed.
They find this out after choosing a collar, personalising it and reaching payment.

Why it matters

Not a converter app: your store reports the Indian market itself and prices on the server.
Somebody switched a market on; the shipping zone that would complete it was never added.
Both halves are settings you own, and they disagree with each other.

How this happens, in order4 steps

Rs. 1,600.00 in the basket — and on the homepage, and on the collar page before it. Every screen this shopper has seen is priced in rupees.
Then checkout asks where to send it. Canada, Mexico, United States. That is the entire list, and it is the same list at both widths.
So there is no address this shopper can finish — and they learn it here, after choosing a collar, typing a dog's name and reaching payment.
One cookie later the same collar is $16.07, in a market you can deliver to. Neither price is an accident of who is looking: both are settings, and they disagree.

Show me the evidence

In your store’s own words

Canadathe Country/Region control on the first step of checkout, first option
Mexicothe same control, second option
United Statesthe same control, third and last option
Rs. 1,600.00the basket subtotal on the same session
🚚 FREE SHIPPING ON ORDERS OVER $50 • NEW ARRIVALS JUST DROPPED • SHOP NOW •the marquee near the foot of the homepage, on the same rupee-priced session

Quoted character for character from the capture — check any line against your own page.

Measured on a laptop, on the first step of checkout: rupees above, three countries below, none of them the visitor's. It is written up as checkout-01 in our full audit of this store — ask us and we will send it.

what was measured — 6 readings

Unpinned session: Shopify.country IN, Shopify.currency.active INR, Shopify.currency.rate 96.9301655 — the figure every board on this page carries in its own capture note.
The same session, with a localization=US and cart_currency=USD cookie pair set before the first request: Shopify.country US, Shopify.currency.active USD, rate 1.0.
Money collar, unpinned: Rs. 1,600.00, compare-at Rs. 2,400.00, Save Rs. 800.00. Pinned to US: $16.07, compare-at $24.73, Save $8.66.
/cart.js unpinned: currency INR, total_price 160000. Pinned: currency USD, total_price 1607.
The checkout's own address carries the locale en-in; the US-pinned session's carries en-us. The country control offers the same three options in both.
The product feed lists the collar at 16.07. 16.07 x 96.9301655 = 1557.67, and the store displays Rs. 1,600.00 — a rounding rule, recorded because it was measured, and not argued as a finding.

05/9

The Holiday card's button has no address, and the collection behind it has ten products in it.

high severityfix: one linkon a laptop

WhereThe homepage category row, on the Holiday card.WhySix of the seven cards carry an address. The Holiday card's is empty, and the collection it names held ten live products on the day we walked it.What it doesPressing Shop Now does nothing at all. A card that does nothing reads as a broken shop, and this one is sitting on seasonal stock.

Fix this firstPoint the card at the collection it names. If the season is over, take the card down — what should not stay is a card that draws and does nothing.

Your page, today — unretouchedmeasured

a laptop

shot 5The category cards, including one whose button leads nowhere

What we saw

Seven category cards. Six carry an address; the seventh, Holiday, carries an empty one.
Its Shop Now button does nothing at all when pressed.
The collection is not empty — it held ten live products on the day of capture.
The row is wider than the window, so Holiday is the one card a visitor has to scroll to reach.

Why it matters

A card that does nothing reads as a broken shop, not as an empty category.
It is sitting on seasonal stock a visitor was actively trying to reach.
The products are fine. The link is the whole of it.

Show me the evidence

In your store’s own words

Holidaythe seventh category card's heading
Shop Nowthat card's button, whose href attribute is empty

Quoted character for character from the capture — check any line against your own page.

Measured on a laptop, on the category cards, including one whose button leads nowhere. It is written up as quick-win-3 in our full audit of this store — ask us and we will send it.

what was measured — 3 readings

Seven cards in the row. Six hrefs resolve; the seventh is href="".
holiday-collection: 10 products. featured-products (17) has no card in the row at all.
The row measures 2,238px wide inside a 1,440px window, so the Holiday card sits off the right edge until the carousel's own dots are used.

06/9

Everything the product page says about shipping and returns is one line, and it does not mention returns.

high severityfix: some writing, and one linkon a laptop

WhereThe collar page, in the panel titled Shipping & Returns.WhyOpened, the whole of what it holds is one line about dispatch. No returns window, no destination, no carrier, no link to a policy.What it doesA shopper checking the terms before paying learns when it leaves you and nothing about what happens if it is wrong — on a collar made to order with their dog's name on it.

Fix this firstAnswer the question the heading asks. State the returns window, and say plainly what applies to a collar made with somebody's dog's name on it.

Your page, today — unretouchedmeasured

a laptop

shot 6Shipping & Returns, opened

What we saw

The accordion is titled Shipping & Returns.
Opened, the whole of what it holds is one line about dispatch time.
No returns window, no destination, no carrier, no delivery estimate, no link to a policy.
This is on a made-to-order personalised product — the category where that question matters most.

Why it matters

The heading makes a promise the panel does not keep.
A shopper opening it for the returns terms learns a dispatch time instead.
You do have policy pages. They are not reachable from the place the question gets asked.

Show me the evidence

In your store’s own words

Shipping & Returnsthe accordion heading on the product page
Product will ship with in 1-2 daysthe entire contents of that accordion, opened

Quoted character for character from the capture — check any line against your own page.

Measured on a laptop, on shipping & Returns, opened. It is written up as trust-01 in our full audit of this store — ask us and we will send it.

what was measured — 2 readings

The opened accordion's full text: "Shipping & Returns Product will ship with in 1-2 days".
Opened by pressing the store's own heading, which is an H2 inside a summary with role="button".

07/9

The basket totals in rupees and defers the delivery cost to a checkout that cannot deliver.

high severityfix: a setting, with the first findingon a laptop

WhereThe basket, on the line under the subtotal.WhyIt totals in rupees and defers taxes and shipping to checkout — the screen that will then refuse their country.What it doesThe last cheap moment to discover the problem is the one that tells them to wait for it.

Fix this firstName the destination in the basket, before checkout is reached. One line there removes the whole class of surprise, whichever way the first finding is settled.

Your page, today — unretouchedmeasured

a laptop

shot 7The basket total, in rupees, with delivery deferred

What we saw

The basket shows a rupee subtotal and defers taxes and shipping to checkout.
So the delivery amount is unknown until the screen that will refuse their country.
The last cheap moment to find out is the one that tells them to wait.

Why it matters

Deferring shipping to checkout is ordinary and usually harmless.
It stops being harmless when checkout is where the refusal lives.
The deferral is what carries a shopper past every earlier chance to find out.

Show me the evidence

In your store’s own words

Taxes and shipping calculated at checkoutunder the basket subtotal
Rs. 1,600.00the basket subtotal

Quoted character for character from the capture — check any line against your own page.

Measured on a laptop, on the basket total, in rupees, with delivery deferred. It is written up as cart-01 in our full audit of this store — ask us and we will send it.

what was measured — 2 readings

Basket subtotal: Rs. 1,600.00 INR. /cart.js: currency INR, total_price 160000.
The basket names no destination and no carrier.

08/9

A shopper searching for your shipping policy is told there are no results, and offered nothing else.

high severityfix: a theme changeon a laptop

WhereThe search results page, after typing shipping.WhyIt returns nothing and advises checking the spelling. You have a shipping policy page; the search does not reach it, and the page carries no onward link of any kind.What it doesSomeone with a question about delivery is told they cannot spell, on a page with nowhere to go next, at the moment they were settling the last thing before buying.

Fix this firstSend policy words to your policy pages. Then give the empty result somewhere else to go — the recovery links cost nothing and catch everything the routing misses.

Your page, today — unretouchedmeasured

a laptop

shot 8A shopper searching for the shipping policy

The measured region on https://www.tagpup.com/search?q=shipping — the region this finding is about

The measured region on https://www.tagpup.com/search?q=xyzzyqwerty — a nonsense query, returning the same page

What we saw

Searching "shipping" returns nothing, and advises checking the spelling.
You have a shipping policy page. The search does not reach it.
The result page carries no onward link at all — no collections, no policy, no support.
A nonsense query returns the identical page, word for word.

Why it matters

Someone typing "shipping" into a shop's search has a question, not a spelling problem.
They are told to check the spelling of a word they spelled correctly.
It ends the visit at the moment they were resolving the last thing between them and buying.

Show me the evidence

In your store’s own words

No results found for “shipping”. Check the spelling or use a different word or phrase.the search results page for a policy query
No results found for “xyzzyqwerty”. Check the spelling or use a different word or phrase.the same page for a nonsense query

Quoted character for character from the capture — check any line against your own page.

Measured on a laptop, on a shopper searching for the shipping policy. It is written up as search-01 in our full audit of this store — ask us and we will send it.

what was measured — 2 readings

/search?q=shipping — 0 results, 0 onward links in the results region.
/search?q=xyzzyqwerty — the same template and the same wording, with only the quoted term differing.

09/9

Ten product cards carry a title that names a different product from the one they open.

high severityfix: a data job, with redirectson a laptop

WhereThe Holiday collection, on a card a shopper is about to press.WhyTen of your 104 products carry a title that names one collar and an address that opens another — one duplication chain nobody finished.What it doesA shopper who checks the address before trusting the page finds it disagrees with the title, on the seasonal stock where being in the right place is the whole purchase.

Fix this firstFix the titles that publicly say "(Copy)" first — those are the ones a shopper can see today. The addresses are the longer job, and they need a redirect list.

Your page, today — unretouchedmeasured

a laptop

shot 9Holiday collection cards whose titles and destinations disagree

The measured region on https://www.tagpup.com/collections/holiday-collection — the region this finding is about

What we saw

A card titled for Christmas opens a product whose address is named for Halloween.
It is not one stray card: ten of your 104 products have a title and an address that disagree.
They form one chain — each duplicated from the last and retitled, the address never updated.
At least one of the ten still says "(Copy)" in the title a shopper reads, on the screen above.

Why it matters

The address is what search engines index and what a shared link shows.
A Christmas card landing on a Halloween address undermines the one check a shopper can make.
It happens on seasonal stock, where being in the right place is the whole purchase.

Show me the evidence

In your store’s own words

Christmas Candy Cane Dog Collar, Personalized Holiday Collarthe card title on the Holiday collection
halloween-candy-dog-collar-personalized-trick-or-treat-usa-handmadethe address that card opens
Ginger Bread Dog Collar, Personalized Holiday Collar (Copy)a second card on the same collection, with (Copy) in the title a shopper reads

Quoted character for character from the capture — check any line against your own page.

Measured on a laptop, on holiday collection cards whose titles and destinations disagree. It is written up as collection-01 in our full audit of this store — ask us and we will send it.

what was measured — 3 readings

10 of 104 products have a handle whose leading tokens disagree with their title.
Three of them are on the Holiday collection, which is the board above. Which collections the other seven sit on was not counted.
Titles still carrying a public "(Copy)": the audit reports several and names "Ginger Bread Dog Collar, Personalized Holiday Collar (Copy)", which is on the board above. We did not count the rest.

Checking this audit

Everything above, and what it rests on

Four sections, and they are here so the 9 findings can be argued with: what we did not look at, how to reproduce any figure yourself, why no percentage appears anywhere, and exactly what we did on your site. Each is coloured the way the findings above are: red where this audit is exposed, green where there is something you can do, amber where we stop short of a claim, and ink where the evidence itself sits.

What we could not see10 named gaps

An audit that hides its blind spots is guessing. These are ours, named.

not measured

How fast your homepage really is

A measurement of our own. The audit ran a lab test on 27 August and graded the phone homepage very slow; we did not re-run it, and a figure we cannot check is a figure this page will not print — so its numbers are in the audit and not here. One thing the walk did establish and can hand you: your product page draws its personalisation fields from a product-options app, and the audit's own script inventory found more than one such app installed. Resolving that duplication is worth doing on its own merits, before anybody argues about images.

not measured

What order your catalogue should open in

A view about what should lead, and then a test. /collections/all opens alphabetically, which is a filing order rather than a selling one — but alphabetical is a coherent choice and we cannot show from outside that it costs you anything. Best selling is the usual answer and a hand-set Featured order is better if you have an opinion. That is something to try together, not a defect to charge you for.

not measured

Anything after the payment button

No order was placed and no payment method was entered, so nothing about confirmation, dispatch, the order email or the packing slip is known. Whether the personalisation a shopper types reaches you by some route the basket does not show can only be settled from an order export.

not measured

Account pages

Nothing was logged in. Registration, the account area and any returning-customer path are unexamined; the audit's own login attempt timed out.

not measured

The mockup generator

Reachable from the announcement bar and priced separately. It was inventoried only, never exercised.

not measured

Whether markets other than India are quoted but unservable

One unservable market was measured, and one servable one as a control. Whether the same contradiction holds for every other country the store quotes needs the same two-column test from a second address on a different network.

not measured

Why the search controls do not render

They are measurably present and measurably invisible. Whether a theme setting or a stylesheet rule hides them was not diagnosed, and that decides how long the fix takes.

not measured

The exact rule that fires the sign-up modal

It stayed away for a minute of no interaction and arrived within seconds of the first press. Separating first-interaction from scroll depth from a press-restarted timer needs a purpose-built run.

not measured

Whether the rupee rounding is a setting

The displayed Rs. 1,600.00 is above the converted 1,557.88. That is consistent with a price-rounding rule, but the rule itself is merchant-side and was not read.

not measured

What any of this costs you

Your own numbers — what people click, what they abandon, what comes back. We had none of it, which is why there is not a single money, lift or percentage figure on this page.

How to check any line on this page

Open the page named under any figure, at the width in its caption — 390 px is a phone, 1440 px is a laptop. In Chrome: DevTools, then the device toolbar, then set the width.
Select any quoted text on this page and search your own page for it. It should match character for character.
The red ring on a picture is the region we measured, drawn where and at the size we measured it. Any pixel figure under "what was measured" is in CSS pixels, at the width named in that figure's caption.
If the store has changed since we captured it, a claim may no longer reproduce. That is worth knowing — and it does not un-measure the capture: every figure here comes from the dated screenshots on this page.

Why there is no percentage anywhere here

Not a percentage, not a range, not an "up to".
We can tell you the leaks and their direction with confidence; we cannot size them without your analytics, and anyone who says otherwise is selling you a spreadsheet.
Every figure on this page is either measured on the capture or quoted from the audit with its source named.

How this was captured

Walked at both widths. The money path is argued where the shopper meets it: the product form, the basket after a real press of Add to cart, and the homepage price are phone screens, because that is where a personalised collar is bought. The checkout is shown at both widths, because the country list is the finding and it must be seen to be the same list on both. The category row, the sort order, the drifted cards and the opened Shipping & Returns accordion are laid out for a laptop and are shown there. One board, the US-pinned control, is desktop only — it exists to be compared against the rupee boards, not to be read on its own.
Captured 27 August 2026 from a residential connection in India, in Chromium at a phone width and a laptop width, logged out. Sixteen boards were captured in the market the store serves that address — Indian, in rupees — and one, the control, with a US market cookie set before the first request. No script was blocked on any board. One test basket was filled by pressing the store's own Add to cart button and was cleared afterwards; no order was placed and no field was entered at checkout. Every rectangle on this page is a getBoundingClientRect reading taken in the same instant as the picture beside it, with one exception that says so on its own card: the sign-up modal renders inside a shadow root the measurer cannot see into, so its rectangle is placed by hand.

Want the rest of it?

Every figure above is checkable today. If you want the written audit behind it, or the next twenty screens we did not walk, book half an hour.

Book a walkthrough

The Wall · Worksheet · Every string and every number on this page is read from the capture and the written audit — nothing here is typed in by hand when the page is drawn.