How to split D2C performance budget between marketplaces and own site

In most D2C brands, marketplaces and the own-site are treated like competing channels. Separate budget, separate team, separate dashboard. Neither tends to be profitable in isolation. The fix isn't to pick one. It's to wire them into a single funnel with a budget split that comes from a SKU-level model.

The two failure modes

D2C brands tend to fall into one of two camps. Camp one believes Amazon (or Flipkart, Nykaa, etc.) is the discovery channel and the website is for high-intent repeat. Camp two believes the website is the brand-builder and marketplace is a discount-eating necessary evil. Both are wrong, and both lose money quietly.

Camp one ends up subsidising marketplace search ranking with paid spend until the COGS-to-fees math collapses. Camp two ends up paying marketplace prices to Meta and Google to acquire customers their competitors are quietly serving on Amazon at half the CAC.

How to model the split

Start with a channel-level unit-economics model, not a budget guess. For each SKU, compute the four things that matter: marketplace fee structure, ad take rate, return rate, and contribution margin per channel. Then score each SKU on which channel pays it back faster — not which channel sells more.

What usually drops out: a small subset of SKUs (typically 15–25% of the catalogue) drives the majority of own-site revenue at acceptable CM3. The rest pays back only on marketplaces, where the customer is already shopping the category and intent is higher.

Once the model is in place, the split writes itself. The number can be 60/40, 70/30, 50/50 — what matters is that it came from SKU-level economics, not a CEO preference.

What "marketplace majority" should mean

It is not all the spend poured into Sponsored Products. The marketplace ad architecture is itself three motions:

• Sponsored Products on bestsellers and high-velocity SKUs — defending top-of-search.
• Sponsored Brands as the category-shaping motion — a banner that owns the head term and drives traffic into the storefront, not the PDP.
• Sponsored Display as the audience-driven motion — retargeting marketplace viewers and similar-category shoppers, where ROAS doesn't include the same fee drag.

Re-introducing SB and SD on top of an SP-only account is one of the cheapest wins available in beauty & personal care, fashion and food categories. It shifts revenue mix toward higher-margin storefront sessions and lets you defend share without bleeding margin into bid wars on SP alone.

What "own-site minority" should mean

The own-site share gets concentrated on the SKUs that earned it. Full-funnel media mix — prospecting + brand defence + creator-led — pushed against retention loops on the website (RFM-segmented winbacks, refill journeys, bundle prompts). The website becomes a retention asset for the brand's best customers, not an acquisition substitute for marketplaces.

The supply-side change that makes it work

Performance is half the rebuild. The other half is sourcing. COGS reductions on the hero portfolio lift CM1 enough that the marketplace majority actually has margin to play with. Without COGS work, the marketplace share usually stays in the red — performance is asked to do something the supply chain made impossible.

The wrong question is "marketplace or D2C". The right question is "which customer am I winning, on which channel, at what margin?".

The split won't be 60/40 for everyone — maybe yours is 70/30 in fashion, 50/50 in food, 80/20 in electronics. The number isn't the lesson. The lesson is that the split has to come from a SKU-level model, not a budget hunch.