Nobody buys the ring in a browser.

Bridal is a long, high-value, heavily researched purchase made by someone who will not buy it online at any price. It is also the category where ad spend concentrates, which makes the measurement gap expensive rather than academic.

Between the first search and the purchase there are typically several visits, at least one appointment, and a conversation with a parent. None of that is visible to a pixel, so the campaign that started it gets no credit and the campaign that happened to be running at the end sometimes gets all of it.

Most jewellers already collect a phone number when someone books a consultation or asks about a custom piece. That booking is the honest start of the sale, and it is what makes attribution possible without installing anything.

Match the appointment or inquiry against the eventual sale in your POS and you get the one number that matters: what a booked consultation is worth by channel, rather than what it costs.

A jewellery shop's repair bench brings people through the door constantly, and almost none of that traffic is thought of as marketing. A ring is resized, a clasp replaced, a watch battery fitted, a piece cleaned before an anniversary. The tickets are small and the margin on them is often thin, which is why the work gets defended as a service rather than measured as a source of customers. A shop advertising for repair work is buying footfall and booking it as an overhead.

What matters is what those customers do afterwards. A proportion return as buyers, sometimes years later, and because the repair ticket and the eventual sale carry the same phone number they connect without anybody maintaining a list. Exporting several years of transactions rather than a single month is what makes the pattern legible. Judged on the repair ticket alone the campaign looks marginal; judged on what those customers eventually bought across the following years it frequently is not marginal at all.

A transaction-type column is what allows both readings at once. With one, the report can rank channels on retail sales, on repair and service work, or on the two combined, and a retailer can see whether the sources producing bench work are the same sources producing buyers. They often are not, which is itself worth knowing before the next set of campaigns is written for a category the shop had assumed was one audience. The alternative is funding two audiences from a number describing neither.

Buying precious metal from the public is a genuine second business inside many jewellery shops, and it is advertised separately because it attracts a completely different person. The seller arrives with an inherited piece, a broken chain or a coin collection, and they leave with cash rather than a purchase. On a conventional channel report that spend sits awkwardly: it produces no sale, so the campaigns behind it are either excluded from measurement altogether or judged on cost per inquiry with no revenue attached to them.

There are two honest ways to measure it. The margin on the metal purchased is revenue and can be exported as such, carrying the seller's contact detail and the date, which lets the buying campaigns rank on what they actually earned the shop. And a proportion of sellers become buyers, on the day or later, so the same contact detail appearing in the retail sales file connects the two sides of the counter with no manual work at all. Both readings come from the same two files.

The reason to bother is that these campaigns compete for the same budget as the bridal and gift advertising, and at present they compete on an unequal footing: one has revenue attached and the other has volume. Putting both on one axis is not an argument for either of them. It is an argument for the shop knowing which of its two economies is being funded by assumption, which in most independent jewellers turns out to be whichever one the owner enjoys running more.

Nearly everything a jeweller sells is bought for somebody else. The buyer's name is on the receipt, the buyer's phone number goes into the system, and the person who ends up wearing the piece may never appear in any record at all. That is convenient for attribution, since the person the advertising reached is the person who paid. It has one consequence that catches stores out: the recipient's own next purchase, made for themselves, is a new customer with no history behind it.

The buyer, though, comes back on a calendar. Anniversaries, birthdays, a graduation, a second child, a significant birthday a decade later. Those purchases carry the same contact detail and credit the source that produced the first one, which is why loading several years rather than a single season changes the shape of the table entirely. A source producing buyers who return every eighteen months is worth several times one producing an identical first ticket and nothing afterwards, and one month cannot tell them apart.

This is also the case for putting the shop's own client book in the lead file, tagged as the client book. It is not advertising, it produces substantial revenue, and leaving it out of the comparison makes every paid channel look better than it is relative to work the shop has already done. Ranked honestly beside search and social, an established client list may well sit at the top of the table, which is a finding about where the next hour of effort should go rather than an excuse to stop advertising.