Omnichannel and pure-play run on different playbooks.

Two retailers can hand over the same budget and get back two different plans. That isn't inconsistency. The margin sits in different places, and media follows the margin or it wastes it. A store with a website and a website with no store are not one business in two outfits. They pay the rent by different means. So they buy differently and read differently, and they fail for different reasons.

Start with the floor. A retailer with doors is judged on a fuller room. The lease demands that outcome week after week, whether or not any dashboard reports it. Online revenue still counts. But it sits beside a second number. How many people walked in.

Pure-play answers to a different landlord. There is no room to fill. There's a warehouse, and a contribution margin per order that either survives the shipping label or doesn't. The whole plan gets judged on blended return and on what a genuinely new customer costs to acquire.

The scoreboard decides what you should buy.

Store visits are modelled. Platforms lean on shoppers who agreed to share location, then scale that sample into a period total. It's an estimate, reported without a single name in it. This desk says so out loud rather than dressing it up. The line still earns its place, because little else in an ad account points at the floor. Performance Max can carry store goals, and Local Inventory Ads put what's actually on the shelf into the ad. A Business Profile feeds the local surfaces where a nearby shopper decides whether the drive is worth it.

Pure-play measures none of that, because none of it exists. It reads blended return first. All revenue over all spend, on one line, before any platform gets to tell its own story. Then it separates acquisition from repeat. A handsome return is often just your existing customers buying again, and those people were already yours.

Watch one channel change jobs and the point lands. Performance Max can chase store goals for a retailer with doors, or run purely on online conversions for a pure-play. Same product. Different assignment. The desk sets the goal before it sets the budget, because the goal decides what the money is even trying to buy.

Geography stops meaning the same thing.

For a store, a radius is a real object. It's the drive people will actually make, and it's shorter than it looks on a map. A furniture retailer in Barrie is not competing for Ottawa. Spend that leaves the catchment leaves the building with it.

For pure-play, geography is a shipping cost and a delivery promise. A parcel to Thunder Bay costs more and lands later than one to Mississauga. That difference belongs in the bidding, not in an apology email on Thursday. So the same map produces two opposite instincts. One plan tightens until the radius matches the drive. The other widens until the margin per parcel stops working.

Furniture makes the gap obvious. The delivery van is a real cost with a real limit. A sofa sold well outside the delivery zone can land with less margin than the same sofa sold across town. A pure-play brand prices that in and keeps selling. A showroom simply declines the trip. Two honest answers to the same ad click.

Fashion makes the other half obvious. Returns are the tax on shipping clothes to people guessing their size, and a fitting room is a return that never happens. Treat that as a real advantage and buy store visits for it, instead of writing a warm line in a brand deck.

Week four tells you which playbook you're actually running.

Week four is where the two separate hardest. By then an omnichannel plan gets read on the floor first. The store-visit trend goes into the blended read, beside online revenue and the calls that come in. Cost per visit sits in that read as an estimate you can still act on. No single line settles the season on its own. Online revenue may barely move without the plan having failed. The money went where the lease points.

By week four a pure-play plan gets read the other way. Blended return comes first, and acquisition cost gets checked against the ceiling you set. Order volume matters less than what the orders cost. A week of cheap revenue from returning buyers reads like a win on the surface. You already paid for those customers, some time ago, in another quarter.

Owners get hurt importing one playbook into the other, and the same two mistakes keep repeating. The first is a store judging itself on return alone. The site takes credit for everything it touched. The floor gets no line in the report. Budget drifts steadily toward the channel that can prove itself. A season later the online number looks respectable and the room is thinner. Nobody wrote down why.

The second is pure-play buying as though it has a store. Broad local awareness and geofenced display, inside a radius drawn for a showroom. It fills a floor that doesn't exist. The parcel still has to ship, and the acquisition cost still has to clear the contribution margin.

Most retailers sit somewhere between the two. That's the normal case, not a confession. You have three stores and a site that quietly does a slice of the business. Or a site doing most of it, plus one showroom that closes the expensive orders nobody buys unseen. Both are workable. Neither survives a plan that pretends the split isn't there.

The honest question is which half pays the rent. Answer it and the plan writes most of itself. The paying half takes the bigger share of the money and the primary number. The other half gets what it earns, and gets read on its own terms instead of borrowed ones.

One thing both models share is the plumbing. Each lives on a product feed with required attributes, and each suffers the same way when that feed breaks. The store loses its Local Inventory Ads and the shelf goes invisible. The pure-play loses Shopping coverage and the parcel never gets ordered. One fault, two different losses.

So the split gets written down instead of being implied by whichever dashboard opens first. A plan on this desk names the paying half and sets the ceiling on acquisition cost. Then it runs in weeks, because BFCM is not November.

You're allowed to disagree with the split. That's the whole reason for putting it on paper. An argument you can read is an argument you can win, and reading it in August beats discovering it in February.