Should Mercado Libre show more Mercado Ads Product Ads in search results or protect organic search relevance?

Introduction

Mercado Ads is one of MercadoLibre's fastest-growing businesses: advertising grew 67% FX-neutral year over year in Q4 2025, helped by AI bidding and the AI Advisor that sends sellers budget and ROAS recommendations over WhatsApp. Product Ads appear inside Mercado Libre and Mercado Livre search results, marked "Patrocinado". Each extra ad slot is almost pure margin, but each slot also pushes an organic listing down. If buyers in Mexico, Brazil or Argentina start seeing less relevant, more expensive or slower-to-deliver products at the top of search, conversion and trust fall, and sellers end up paying for traffic they used to get for free. I'll analyze whether to increase ad load in search or protect organic relevance.

Step 1

Clarifying Questions (3 minutes)

• Context: What is the current ad load on a typical search results page, for example how many of the first 20 results are sponsored, and does it differ by country and app vs. web?

Why it matters: The marginal cost of an extra slot depends on where we start. Expected answer: Around 3–4 of the first 20 results on mobile, slightly more on desktop. Impact on approach: Test increments such as +1 and +2 slots rather than a big jump.

• Business Context: Is the goal to hit an ads revenue target this year, or to maximize total marketplace value (GMV, take rate and ads together)?

Why it matters: A revenue target favors more slots; a total-value goal makes us count lost GMV. Expected answer: Total value, with ads as a growing share of revenue. Impact on approach: The decision metric combines ads revenue with GMV and commission impact.

• User Impact: Are ads ranked only by bid, or by bid combined with relevance and predicted conversion?

Why it matters: Relevance-weighted ads hurt the buyer much less than bid-only ads. Expected answer: A blended quality score already exists. Impact on approach: The real question becomes how strict the quality threshold should be, not only how many slots.

• Seller Impact: What share of advertisers are small sellers vs. large brands and official stores?

Why it matters: More slots can help small sellers gain visibility, or price them out if large brands win most auctions. Expected answer: Most advertisers are small sellers, but large ones spend most of the budget. Impact on approach: Track seller-level outcomes, not only totals.

• Resource: Can we run search experiments by user, or only by query or geography?

Why it matters: Determines experiment design and cannibalization measurement. Expected answer: User-level randomization is available. Impact on approach: Run a user-level A/B test with holdouts.