Red Pin Geek

Mikel Grant Jewellery

Search and AI visibility, built around your gemstone collections  ·  for Mikel

All six collection pages live  ·  first real read due 17 October

Timeline

✓
5 August 2026
We started
Snapshot, collection architecture for up to six collections, and search measurement set up.
Complete
✓
27 August 2026
Visibility Snapshot
Four AI tools asked the same six buyer questions, so we know exactly where you are and are not showing up.
Complete
✓
31 August 2026
Google measurement live
Search Console is connected and reporting. Your first Google numbers are below.
Complete
✓
9 to 17 September 2026
All six collection pages live
Turquoise, pearl, fossil, moonstone, sapphire and opal. Each one written in your words, with your photographs, and readable by the AI tools that recommend makers.
Complete
✓
16 October 2026
First proper before and after
Each page measured over the same number of days either side of the day it went live, and the same eleven buyer questions asked of the AI tools again.
Complete

Where you started, and where you are now

304
Products, all with descriptions
49
Collections on your site
6/24
AI answers that pick you (was 2)
9
Gemstone collections still with no words (was 15)

What is complete

✓
Visibility Snapshot delivered. Four AI tools, six real buyer questions, twenty-four answers, all measured on the same day. It tells you where you get recommended, where you do not, and the reason for the difference.
✓
Google Search Console set up and reporting. This is the tool that shows what real people typed into Google and where your pages appeared. It is now connected to your site and collecting. One thing worth knowing: Google only starts counting from the day a site is connected, and it cannot be backdated. So this is genuinely day one of your record, which also makes it a clean before picture for the work ahead.
✓
All six collection pages written and live. Turquoise, pearl, fossil, moonstone and labradorite, sapphire and ruby, and opal. Each one is built from your own words at the bench, carries your photographs, and has the behind-the-scenes labelling that lets an AI tool understand what the page is about. Those six collections held no writing at all in August.
✓
59 broken links repaired. No charge. While linking your turquoise guide to its new page, we tested every product link across your blog. Ninety-six links, fifty-nine of them dead, including the SHOP NOW button on your best performing article. Nothing looked wrong on the page, which is why it had gone unnoticed: the photos still loaded. Every one now lands on the right collection instead of an error. This was included, not billed.
✓
Your product catalog tidied. Product types had drifted over the years, Necklace and Necklaces, Earring and Earrings, and so on. 218 products were corrected across your full catalog of 1,270, including drafts and archived pieces. It does not change anything a shopper sees. It matters because filters, feeds and AI tools all read those labels. This was the catalog prep line on your original invoice, finished and never reported back to you until now.
✓
Your site was checked for anything blocking you. Nothing is. All seven AI tools that read websites are allowed in, and your site map lists all 300 products and all 49 collections. Being found is not your problem. Being worth recommending is the part we are working on.

What the numbers say, 16 October

·
The AI tools are the real news, and the comparison group is what makes it count. Each month the same eleven buying questions go to two AI assistants, twice each. Six are about your stones. The other four are a comparison group, stones we deliberately did NOT write pages for: amethyst, garnet, lapis lazuli and agate. In August you came up in 2 answers out of 24 about your stones. Now it is 6. Over the same period the comparison stones went from 4 out of 16 down to none at all. They started ahead of you and fell to zero while yours tripled, and that is what tells us this is your pages rather than the assistants changing on their own.
·
The kind of mention changed, which matters more than the count. In August the single mention used your site as a source without ever saying your name. Now five of the six say your name in the answer itself, on turquoise, opal and ammolite. Being findable was never your problem. Being worth naming is what the pages were for, and that is the part that moved.
·
Google is mixed, and we are not rounding it up. Two of the six pages are clearly better, two did not move, and two went backwards. Fossil is the clearest win: shown 58 times in the 19 days before it went live and 135 times in the 19 days after, from the same spot on the page. Opal moved from about 11th to about 10th and earned its first click. Pearl climbed from about 11th to about 6th, on small numbers, 4 listings then 11. Sapphire did not move, 7 then 6. Moonstone went backwards, 16 then 5, its spot slipping from about 9th to about 13th.
·
Turquoise looks worse than it is, and it is worth a minute. In the matched 14 days it fell: shown 264 times then 217, and 7 clicks down to none. But it did not stay down, and 14 days stops too early to see that. Over the longer stretch from 23 September to 15 October it ran 475 listings, 5 clicks and a spot of about 14th, better than turquoise has ever done. The short comparison caught a dip and stopped before the recovery. It is also the hardest page of the six to judge, because the 59 broken-link repairs and the new link from your turquoise guide all went live the same day as the page.
·
Why this is not the full month we promised. Google only starts counting a site from the day it is connected, and it never fills in the past. Yours connected on 26 August, a couple of weeks before the pages went up, so there is no record of the earlier months and there never will be. Instead each page is measured over the same number of days either side of the day it went live. Turquoise gets 14 days each way, opal gets 22.
·
Canada is where you are actually chosen, and it is not close. In the last four weeks Canada brought you 195 clicks and the United States brought 34, even though Google showed your pages to Americans nearly twice as often. Canadians who see you click about ten times more readily. So pausing US shipping is costing you much less than it probably feels like. Your Canadian clicks held steady right through, about 7 a day before and after.
·
Your own work is in these numbers. Since the end of September you have been adding pieces to sapphire, turquoise and moonstone. Three of those six results are partly you filling the slim collections, not just the words on the page. And on orders: we see Google, not your till. With US shipping paused a flat order count would be the tariffs rather than the pages, so we keep those two apart.

The biggest thing sitting unused

·
Your gold guide came up in Google 14,932 times in the last four weeks and was clicked ten times. It sits about sixth, which is the first page, near the top of what people actually look at. The article is clearly good enough to rank there. What is failing is the title and the description Google prints underneath it, which is the bit that decides whether anyone clicks. Your own site already proves this is fixable: your French pages carry the same articles and get clicked about five times more often. Same writing, different wrapper. Quoted separately at $297 USD.

What we are tracking

You are recommended for the one question your website answers in plain words, and absent from every question it does not.

There are two different doors into your shop, and they no longer open together.

For years being findable meant one thing: showing up high in Google’s list of blue links. That still matters, and plenty of your customers shop exactly that way.

But a growing number never see that list. They ask an assistant a question in plain English and get back one answer, with a few makers named in it. Being named in that answer is a different thing from showing up high in Google’s list, and the two have come apart. Being strong in one does not carry you in the other.

So this section has two parts. The first asks: do assistants name you? The second asks: where do you sit in Google’s list? You are a striking example of the two disagreeing, and that disagreement is exactly where the opportunity is.

Four words used below, in plain terms. Being named is when an assistant mentions you anywhere in its answer. Being picked is stronger: the assistant puts you forward as the answer, not as one of a list. A listing is one time your page appeared in somebody’s Google results. Your spot is how far down that Google list you were, where 1st to 10th is the first page and almost nobody scrolls past it. A click from Google is somebody seeing your listing and choosing to open it.

1. Door one: do AI assistants name you? Six questions a real buyer would ask, put to four different AI tools on 27 August. Twenty-four answers in total.

The buyer questionWhat happened
Independent makers of one-of-a-kind gemstone jewelleryNot named, 0 of 4
Handmade jewellery from an independent Canadian designer near the Sunshine CoastPicked by 3 of 4
Designers who specialise in moonstone, turquoise or opalNot named, 0 of 4
Nature-inspired, west coast or artisan silversmith jewelleryNot named, 0 of 4
Designers who work in ammolite or fossilNot named, 0 of 4
Ethical one-of-a-kind instead of mass-producedNot named, 0 of 4
On the question you win, three of the four tools put you forward as the answer and called you things like the strongest local maker. The fourth named you first but inside a list of seven makers with no particular recommendation, which is a real difference and worth knowing: being mentioned is not the same as being chosen.

2. Door two: where do you sit in Google’s list? Your first six days of Google data, 26 to 31 August.

 First 6 days
Times your pages appeared in Google results8,526
Clicks from Google63
Different search terms you appeared for1,105
Different pages of yours that appeared358
One honest limit on that clicks number. This table can only see Google. It does not mean nobody else reached you: when an assistant names you and somebody comes straight to your shop, Google never sees that visit at all. Some people click straight through and your own site analytics catch them, but plenty read the answer and then simply type your name into their browser, which arrives looking identical to somebody who already knew you. This is genuinely hard to count, and anybody who says they can measure it precisely is selling something. In practice it means the effect of being named is almost certainly larger than any number here can show, not smaller.

The finding that explains both tables. Your home page says Sunshine Coast, in plain words, on the page. That is the one question the assistants get right about you. Fifteen of your seventeen gemstone collection pages have no words on them at all, so when somebody asks about moonstone or turquoise or opal, there is nothing on your site for a machine to read and repeat. The pattern is that exact, and it is the whole case for the work ahead.

You are not short of writing, and this is the encouraging part. All 304 of your products carry a description, and eight of your story collections carry your own writing, some of it 800 characters and genuinely lovely. The gap is not ability. It is that the writing never reached the gemstone pages, which are exactly the ones buyers search for.

Turquoise is the clearest opening you have. Google already puts your turquoise guide in the top ten, and around forty different turquoise questions bring up your pages between the 5th and 12th spot. Yet not one AI tool names you when somebody asks for a turquoise specialist. The reputation is already built. It simply is not reaching the assistants, and that is a much easier problem than starting from nothing.

Your biggest near-miss on Google. Your guide on the difference between solid gold, gold filled, gold vermeil and gold plated appeared 14,932 times in the last four weeks, sitting around the 6th spot, and was opened ten times. People are being shown it and choosing not to click, which almost always means the title and description in the results are not pulling their weight. That is a small fix with a large number behind it.

Two honest limits. One reading is never a trend, and the next one is in mid November. And these AI answers are one measurement on one day; assistants vary their wording every time they are asked, so what we watch over time is whether your name keeps coming back, not whether any single answer repeats.

Your next moves

Two small things, and neither is urgent 0 of 2 done

Check them off as you go; your progress saves on this device.

Red Pin Geek
Snapshot 27 August 2026  ·  Google measurement live 31 August 2026  ·  first before and after 16 October 2026  ·  next update mid November