Google Ads & analytics review

Account Review & Findings

coinsonline.com

A review of how the Google Ads account is set up, where the budget is going, and what the tracking is and isn't recording.

Google Ads reviewed
Full account history, plus 30 May – 27 Aug 2026
Analytics reviewed
1 Feb 2025 – 27 Aug 2026
Prepared
27 August 2026

Revenue per $1 of ad spend

Reported
full history
$40.88
Reported
last 90 days
$2.21
Your
target
$90–100

Verification against WooCommerce in progress

Overview

In short

One finding sits above everything else in this review, so we'll start there.

Over its full history the account records about $41 of revenue for every $1 of ad spend. Over the last 90 days it records $2.21. Ad delivery was normal throughout — no gaps, no budget exhaustion — which points to how results are being measured rather than to a change in how the ads performed.

We've identified a specific configuration error in the way purchases are counted, and we're confirming how much of the gap it accounts for. Section 02 sets out what we found and the one piece of information we need from you to close it out.

The review also turned up several things worth fixing straight away, independent of that: budget going to search terms that have never produced a sale, broken links on two of your best-selling products, and no tracking at all for phone orders or sell-to-us enquiries.

Revenue per $1 spent
last 90 days
$2.21
Search spend, 90 days
with no recorded sales
$5,169
Phone or enquiry
conversions tracked
0
Shopping return
on 7% of budget
$15.43
Revenue per $1 spent
Google calls this “conversion value / cost.” If you spend $1 and the tracked sale is worth $90, that's $90. Your target is $90–100.
Conversion
An action Google counts as a result. Should mean a completed purchase. Part of the problem here is that it currently doesn't.
Cost per click
What you pay each time someone clicks an ad.
Shopping
The campaign type that shows product images and prices directly in Google results, fed by your product catalogue.
Performance Max
Google's automated campaign type. It places ads across Search, Shopping, YouTube and Gmail, and decides for itself where to spend.
Branded search
Ads shown when someone searches for your name specifically, rather than for coins in general.

01

Where the account stands today

Your target is $90–100 of revenue for every $1 of ad spend. Here is what the account currently reports against that.

Revenue per $1 spent, shown against the $100 target

Last 90 days$2.21

$12,126 spent, $26,782 in tracked sales.

Full account history$40.88

$40,478 spent, $1,654,892 in tracked sales.

Your target$90–100

The last 90 days account for 30% of everything ever spent on this account, but only 1.6% of all sales value ever recorded. That is the discrepancy this review set out to explain.

Ad delivery itself was normal throughout. There were no dark days, no budget exhaustion, no gaps. The ads ran; the results stopped being recorded.

Worth saying plainly

To reach $90 per $1 spent on current budget, the account would need $1,091,329 of tracked sales in that 90-day window. It recorded $26,782.

That is a 41-fold difference. Gaps that size don't come from bidding or ad copy. Something structural is behind it, and identifying exactly what is the first job.

02

What we found in the tracking setup

A configuration error in how sales are counted, plus two figures that should match each other and don't.

A checkout step is being counted as a sale

Two things are set up to count as a “Purchase” in Google Ads. One of them is not a purchase.

Actions currently counting as a purchase
What it recordsStatusSales countedValue recorded
A completed purchase on the websiteCounted41$1,628,048
Someone reaching the payment stepCounted9$26,781

The second one fires when a visitor reaches checkout and enters payment details. Reaching checkout is not buying. Because it is set to count as a purchase, Google's automated bidding is currently being told to find more people who reach checkout and leave.

We're correcting this in the coming week.

Two systems reporting different order values

Google Ads and Google Analytics should agree on what a typical order is worth. Currently they don't.

Average order value — Google Analytics$22,380
Average order value — Google Ads$39,708
Difference on the same orders77%

Google Ads pulls its figures from Google Analytics, so the two should match. A 77% gap means at least one of them is calculating order value wrongly. Analytics also shows only 169 buyers from 233,545 visitors, and 5,802 people reaching the cart for every 169 who complete a purchase. Both of those ratios are far outside what a working store looks like.

What we're confirming

Three things could explain the 90-day drop. Each leads somewhere different, so we're establishing which applies before acting on it.

  • The purchase tracking stopped working. Sales are still happening but are no longer being recorded. This is the explanation the setup error points to.
  • The older figures were overstating order values. The historical data reports an average order of $39,708, which is higher than this catalogue would produce. Recent data reports around $2,678 per sale, which is in the right range. It's possible something was corrected about 90 days ago and the current figures are the more accurate ones.
  • Demand fell. Your branded search campaign is running at 9% of its historical impression volume and is flagged “limited by search volume.” That campaign produced most of the account's recorded sales. If fewer people are searching for you by name, fewer sales follow, with no tracking fault involved.

The first two are fixes on our side. The third would change what we recommend commercially. One piece of data separates them.

What we need from you to settle it

A WooCommerce order export for 30 May to 27 August 2026 — order count and gross revenue by month, ideally for the preceding year too.

If Woo shows steady orders that Google isn't recording, it's a tracking fix. If Woo shows a decline matching the drop in branded searches, it's a demand shift and we'd adjust the commercial plan instead. That export settles it either way.

03

What has worked, and what never has

Splitting the account's full history into branded and non-branded advertising is the most useful cut of the data.

Full account history by type
 SpendTracked salesPer $1 spent
Branded search
People searching for you by name
$12,334$1,614,531$130.90
Everything else
People searching for coins and bullion generally
$28,144$40,361$1.43

Branded search — showing ads to people already looking for Coins Online — has comfortably exceeded your target. That makes sense: those people already intended to buy from you.

Everything aimed at finding new customers is a different story. 69% of all money the account has ever spent has returned $1.43 per $1, against a $90 target. Not recently. Across the account's entire history.

What this means

The healthy overall figure was always being carried by one campaign. Non-branded advertising has never worked in this account.

That changes what we recommend. This isn't a case of tuning campaigns that are nearly there — it's a case of rebuilding an approach that hasn't yet been proven.

04

Where the budget is going

These are problems worth fixing whatever the tracking investigation concludes.

Last 90 days by campaign
CampaignSpendTracked salesPer $1
Shopping$897$13,841$15.43
Performance Max$6,060$12,940$2.14
Branded search$2,189$0$0.00
Non-branded search$1,854$0$0.00
Retargeting$1,125$0$0.00
Total$12,126$26,782$2.21

Shopping is the standout. It returned $15.43 per $1 on just 7% of the budget, at a cost per click of about 5 cents. Search campaigns took 43% of the budget and recorded nothing.

Search terms nobody has been filtering

Google shows you every phrase people actually typed before clicking. Over the last 90 days there were 58,169 of them. In the account's entire history, 27 have ever been added as targets and 10 blocked.

  • $1,331 went to people searching for your competitors by name — SD Bullion, APMEX, JM Bullion and others. No sales from any of it.
  • $2,381 went to 47 Performance Max search terms that each cost over $20 and produced nothing. One click on “buy jewelry with credit” cost $145.65. You don't sell jewellery. A single click on “gold silver” cost $148.05.
  • 36% of all Shopping clicks are people looking up what a coin is worth — “1921 silver dollar value,” “kennedy half dollar value,” “dimes worth money.” That's 4,276 clicks and no sales. These are people who found a coin in a drawer, not people buying a monster box.

Two of your best products have broken links

Your two highest-converting products — the American Gold Eagle 1 oz and the American Gold Buffalo 1 oz — are both disapproved in Google Merchant Center for “Product page unavailable.” Google is trying to send shoppers to pages it can't reach.

Meanwhile the biggest single consumer of Shopping budget is a $1,000 face-value junk silver bag. It has taken 390,052 impressions and 11,248 clicks without a single recorded sale — and it's one of the lowest-margin items you stock.

Bidding basics that have never been set

  • Time of day. Midnight to 6am takes 29.6% of the budget at 6× the cost per click of the evening hours. No adjustment has ever been applied.
  • Devices. Tablets took $202 with no sales. Ads are appearing on TV screens.
  • Ad extensions. 35 disapproved items are sitting unfixed. Nothing in the account has been updated since October 2025.

05

What your target requires

A $90–100 target is achievable, but it dictates where the budget can go. This is arithmetic rather than opinion.

At a $90 target, the most you can afford to pay for a sale is the order value divided by 90. Turn that into a required website conversion rate and the picture is clear:

Website conversion rate needed to hit the target
If a typical order isAt 5¢ per click
(Shopping)
At $1.50 per clickAt $6 per click
(Search)
$22,3800.03%0.64%2.55%
$5,0000.11%2.85%11.4%
$2,5000.23%5.70%22.8%

Your website currently converts at 0.07% of visitors. Read across the table and the conclusion is hard to avoid: expensive search clicks cannot pay for themselves at a $90 target, at any realistic order value. Cheap Shopping clicks against high-value products can.

Your budget is currently weighted the other way round. Across the account's history, 68% has gone into Search and 26% into Performance Max, with just 7% into Shopping — the one channel returning $15.43 per $1.

The strategic shift

Move budget toward Shopping and Performance Max, and point it at your high-value stock — monster boxes, gold tubes, kilo bars. At a $90 target the numbers only work on higher-ticket items, which is also where your margin is.

A $50,116 silver bag absorbing 11,248 clicks without a sale is not an underperforming product. At this target it's a product that cannot reach it.

06

Analytics and tracking gaps

Your biggest orders are invisible

Google Analytics shows zero tracked enquiries or leads on every one of the last 573 days. There is no call tracking, no form tracking, and no call button on any ad.

“Sell coins online” is one of your top ten organic search terms, with 258 clicks. There is a “Sell Coins Online” ad group that has never run, and a “Schedule An Appointment” link that is disapproved.

If a meaningful share of your larger orders comes in by phone — and with $47,000 gold tubes in the catalogue we'd expect it does — none of that revenue is being credited to advertising. Google is optimising as though it doesn't exist.

Traffic is credited to the wrong sources

“Direct” traffic — people typing your address straight into a browser — accounts for 32% of visitors and 53% of revenue, while having the lowest engagement of any source. Genuine direct visitors are usually your most engaged. This pattern means visits are arriving with their true source stripped out.

We found four causes:

  • 8,874 visits arriving from your own older domains, americanrarecoinandcollectibles.com and amrarecoin.com, counted as if they were third-party referrals.
  • A malformed tracking link in the product feed, carrying leftover parameters from a third-party tool. This is likely why Shopping shows 30,358 visits but only one recorded sale.
  • Klaviyo email links aren't tagged. Analytics credits your email marketing with 19 visits when the real figure is around 556.
  • 14,235 visits that Analytics could not categorise at all.

Roughly a tenth of your traffic isn't real

Singapore is the single largest city in your analytics with 7,950 users; Lanzhou, China is sixth with 5,073. Add the data-centre locations and around 10% of recorded traffic is automated rather than human.

Your ads only run in the US, so this costs nothing in ad spend. It does distort every percentage in your analytics and every audience built from it.

07

What we're doing, in order

The first three establish a clean baseline. The rest follows from it.

  1. Pause the search campaigns This week

    $5,169 across three months with no recorded sales, on broad keyword settings with no filtering. This stops roughly $1,700 a month going out while we establish what's happening. Shopping and Performance Max keep running.

  2. Correct the checkout-step conversion This week

    Stop counting “reached the payment page” as a completed sale, so automated bidding stops chasing people who abandon checkout. We'll also remove three other conversion settings pointing at things that don't exist on your site.

  3. Match Google's figures against your WooCommerce orders Needs your input

    The order export described in section 02. This is the one piece we can't do without you, and it determines everything that follows.

  4. Fix the broken product pages Next

    Restore the links for the American Gold Eagle and Gold Buffalo so Google can advertise your two best sellers again.

  5. Add call and enquiry tracking Next

    Call tracking, call buttons on ads, and tracking for the sell-to-us and appointment forms, so phone and enquiry revenue is finally credited to the advertising that produced it.

  6. Filter out the wasted search terms Next

    Block competitor names, jewellery and finance queries, and the “what's my coin worth” searches. Around $3,700 of identified waste, and it stops recurring.

  7. Repair the analytics setup Next

    Exclude your own legacy domains, fix the product feed tracking link, tag the Klaviyo emails, and filter out automated traffic. This is what makes future reporting trustworthy.

  8. Rebalance the budget Once data is trusted

    Shift weight toward Shopping and Performance Max, focused on higher-value stock. We'll bring you a proposed structure once we know which numbers are real.

  9. Housekeeping Ongoing

    Time-of-day and device bidding, the 35 disapproved ad extensions, enriched product titles, and activating the 5,802-person abandoned-cart audience that currently sits unused.

One thing we'd like from you

Your gross margin by product category — bullion versus graded and collectible. A $90–100 target implies margins of roughly 1–3%, and we want to confirm that matches how you actually price before we build a plan around it.