Running paid acquisition across three brands in the same niche taught me what running one brand never could: which “best practices” are actually marketing folklore, and which ones survive A/B comparison across competing P&Ls.

The setup

DTK, ND, and LAVIS sell to overlapping (but not identical) customer pools. DTK is the established wholesale player. ND has stronger blog traffic. LAVIS is the newest entrant with different brand positioning.

All three run Meta Ads and Google Ads. Three separate Business Manager accounts, three Ad Manager dashboards, three different brand voices in creatives. Total monthly ad spend: roughly $70,000 across the portfolio.

Cross-brand experiments

Running three brands in the same niche is essentially a multi-arm A/B test machine. I can test the same hypothesis across three P&Ls and watch what holds. Here is what I learned the hard way.

1. Branded budget is portfolio-aware, not brand-isolated

When I scale a promo on DTK, BRANDED searches on ND spike too. Customers cross-shop within the niche, and any promo activity on one brand drives discovery search across the portfolio. I learned to never read “branded ROAS” in isolation again. Branded performance is a downstream signal of any-brand promotional activity, not a clean measure of organic demand.

That means if your branded ROAS jumps 3x in a month, before you celebrate, check whether you ran a flash promo. Branded is the easiest channel to mistake for organic brand strength when it is actually paid-promo halo.

2. The same creative angle survives in one brand and dies in another

A no-frills “12 colors, $X per bottle” inventory shot crushed it for DTK and bombed for LAVIS. Same product, same audience targeting, same time of year. The difference: brand positioning.

DTK’s audience reads inventory shots as “this seller is real and stocked”. LAVIS’s audience reads them as “discount/clearance vibe”. This is the lesson agency people cannot learn from one account: ad creative is downstream of brand positioning, and brand positioning is set by everything except the ad itself.

3. Cross-brand learning loop beats single-brand iteration

Most ecom operators iterate one brand’s creative against itself. I run 3-way comparisons: ship the same creative pattern across 3 brands in parallel, see which combinations of (creative + brand voice + audience) actually compound.

The signal-to-noise is 3x higher than single-brand A/B testing because you control for time-of-year and macro effects across the same niche.

The Meta Ads MCP I had to build

Switching between three Ad Manager accounts manually became the bottleneck. I built a custom Meta Ads MCP (using Anthropic’s MCP framework) that gives me a unified read view across all three accounts. Spend, ROAS, CTR, frequency, all in one query.

That is the kind of tool you build when you are an operator. Agencies that serve 50 clients cannot build it because each client wants slightly different requirements. The unit economics break. An operator with 3 brands builds it on a Sunday.

What this proves

The biggest paid acquisition learnings come from controlled comparison, not single-brand iteration. Most agencies have hundreds of clients but no portfolio data. They cannot run cross-account experiments because each client has a different P&L. Operators running 2+ brands can.

I advise B2B ecom founders on exactly this: how to set up your paid acquisition stack so the learnings compound. If you are stuck on one brand and do not know whether what you see is signal or noise, we should talk.