The bigger the appliance, the more likely someone calls.

Anything that has to be measured, delivered, installed or hauled away produces a phone call, because the customer has a question the product page cannot answer. That call is where the largest orders are written.

It is also where attribution stops. The session that produced the call is recorded as a visit with no conversion, so the campaign that generated your best revenue looks like your worst.

You cannot deliver an appliance without a phone number, so the data needed to attribute the order is already in the system for operational reasons. There is nothing new to collect.

Match that against ad platform lead exports or a call tracker and the phone orders resolve to channels. What is left in Direct / Unknown is the genuine walk-in trade, reported as exactly that rather than distributed across paid campaigns.

Two customers walk into an appliance retailer and they have nothing in common. One has a freezer full of food and a compressor that died overnight; they will buy today, take whatever is in stock, and pay for delivery tomorrow morning. The other is nine months into a kitchen project, arrives with a designer's specification, and will not take delivery until the cabinets are fitted. The same campaign can produce both of them, and on a monthly lead report they look like one inquiry each.

They are worth different amounts and they arrive on completely different schedules. Emergency replacement converts within days at a moderate ticket. Project business converts over months at several times the value and frequently includes the built-in appliances nobody buys in a hurry. Any measurement that closes its books inside a month counts the first kind properly and loses most of the second, which systematically undervalues whichever campaigns are aimed at people planning rather than panicking — usually the more expensive half of the media plan.

Because there is no window here, the project customer matches whenever the order is finally written. What makes the distinction usable is a column: order type, lead time, or simply a flag the salesperson already sets on the ticket. Mapped as Product / service, it shows emergency and project revenue as separate lines, each with how much of it traces to a channel. It does not rank channels within each, but a retailer who has only ever seen the blended version has been funding emergency demand and project demand as though they were a single market with one cost per sale.

In this category the retailer frequently cannot compete on price. The manufacturer sets what may be advertised, the same model is listed at the same figure by several sellers, and the differences that remain are delivery, installation, haul-away, connection, taking a measurement seriously and answering the phone when it rings. Those differences are real and they are the reason the larger orders close in conversation rather than in a cart. None of them is visible to a click-based report. A report that stops at the click is measuring the only part of the transaction that was never the differentiator.

The practical consequence is that a campaign's value is decided by what happens after the click, which means the only useful ranking is one built from completed orders. Export written or delivered orders carrying the customer's contact detail and the campaign that produced a caller who became a full kitchen package separates instantly from the campaign that produced a price check on one model. Both looked identical inside the ad account, and only one of them paid for the month.

Attachment is where that difference concentrates. Installation, connection, removal of the old unit, extended cover and the accessories that go with a built-in are usually the most profitable lines on the invoice, and they are sold entirely by the conversation. Including them in the export rather than exporting the appliance alone is what lets the report show which sources produce customers who buy the whole job, and which produce customers who buy the box and arrange the rest themselves. That distinction is worth more than the appliance margin on most invoices the shop writes.

Builders, property managers and landlords buy on a replacement cycle rather than in response to advertising. An apartment building needs twenty of the same oven every few years; a developer specifies a package per unit and reorders as sites complete. That revenue is steady, large, and produced almost entirely by a relationship somebody in the business maintains, which is why it distorts a channel report so badly when it is left in unlabeled. One account can outweigh a quarter of retail advertising on its own.

An account-type column mapped as Product / service makes it visible in both directions. Retail and trade show as separate lines, each with its own revenue and how much of it traces to a channel, so a trade order no longer hides inside a retail figure. The channel ranking itself still covers both; to rank retail advertising on retail revenue alone, leave trade accounts out of the sales export. The trade line also shows the part of it that did arrive through a campaign, which can be larger than expected: a builder's first order may begin with somebody searching for a supplier who will deliver to a site on the day it is needed.

Once those accounts are visible as their own line, the trade contact list is worth adding to the lead file too, tagged as trade relationships. It is not advertising, it produces predictable revenue, and ranked beside paid channels it shows the true proportion of the business that relationships carry. For a retailer deciding whether to fund another trade salesperson or another month of search, that proportion is what the decision turns on. Until both sit in one report, the argument is made from impressions of how busy each side feels.