The retailer engine: how Éxito, Jumbo, and Olímpica decide your organic position

If your brand sells on Éxito, Jumbo, and Olímpica, it's being evaluated by three different search engines at the same time. The same SKU can rank at position #2 on one and position #14 on another, with exactly the same content. The difference isn't magic — each engine weights signals differently, and almost nobody is optimizing retailer-by-retailer.

This is the second article in the PDP series for LATAM. In the first one we argued that your PDP is judged simultaneously by two engines: the retailer's and Google's. Here we focus on the first. What do Éxito, Jumbo, and Olímpica reward? Where are the most common gaps? What does a well-done diagnosis of the retailer engine look like?

Spoiler: most brands in LATAM have the same title, the same images, and the same attributes across all three retailers. And that's why their rankings are so different — because the retailers are different.

Why the retailer engine matters more than you think

There's a common intuition that underestimates the retailer engine: "buyers already know what they want, they type the brand, they land on my PDP". Partially true, partially not. But even for buyers who arrive with a brand in mind, what they see next to it are alternatives — and those alternatives are ordered by the retailer engine.

Even more important: the vast majority of searches on Éxito, Jumbo, and Olímpica are generic. "Yogurt," "shampoo," "coffee," "detergent." The buyer sees 30 products ranked by the engine, scans the first 5-10, and picks. If your product is at position 14, it was invisible for that search.

That's where the retailer engine is the first battle. And although the three retailers in Colombia look similar from the buyer's side (similar catalogs, similar layouts), inside they reward different things. The same applies to Walmart Mexico, Falabella in Chile/Peru, Cencosud Brazil. Each with its own logic.

The 6 signals each retailer evaluates

At a general level, every retailer engine rewards variants of the same six signals:

1. Title structure — length, presence of key keywords, brand position
2. Description — length, keywords, HTML structure (bullets vs. paragraphs)
3. Image set — quantity, resolution, mix (packshot, lifestyle, infographic, scale)
4. Video — presence, duration, transcription
5. Structured attributes — completeness and accuracy of category specs
6. Reviews and rating — quantity, recency, distribution, Q&A

Additionally, each engine weights two behavioral signals: conversion rate (visits → purchases) and stock trend (a product that goes out of stock loses momentum).

What changes between retailers is the weighting. And that changes your strategy.

Éxito and Carulla (VTEX): the recent change that broke the old playbook

Éxito.com and Carulla.com run on VTEX, the platform that also powers Olímpica.com in Colombia and many other retailers in LATAM. The VTEX search engine is called Intelligent Search and it's highly configurable. Until recently, Éxito and Carulla weighted attribute completeness heavily because attributes were the direct input for filter facets — the side filters shoppers used to narrow their search.

That changed. Grupo Éxito removed filter facets from exito.com and carulla.com. Today shoppers on those two retailers discover products only via keyword search or by navigating categories. The direct mechanic of "filled attribute → qualifies for the filter → visible when the shopper filters" no longer applies on Éxito or Carulla.

What's still true:

• Attributes still feed the ranker as a quality signal and a keyword-relevance signal.
• 95% completeness still ranks systematically higher than 60%, but through a less visible path (keyword relevance + quality signal, not filter facets).
• Relatively long titles (70+ characters with key keywords included) still matter.
• High-resolution images (from 800×800 px and up) still matter.
• "Department" and "category" correctly declared are still critical — and now more so, because category navigation has replaced part of filter-based discovery.
• Reviews — quantity and recency — remain a strong signal.

What broke: the visible, diagnosable connection between attribute and visibility. The team that used to diagnose drops with "you're not in the X filter" lost that tool on Éxito and Carulla.

And the question of the quarter changed: where do the shoppers who used to filter on Éxito for "gluten-free", "sugar-free" or "size 3" go now? Some to more specific keyword searches. Others to Olímpica or Jumbo Colombia, where filters still exist. That redistribution is measurable — and moves share.

Olímpica (VTEX): the Colombian VTEX where attributes still are a direct lever

Olímpica.com keeps filter facets active. And because PDPs on Olímpica are typically less populated than on other retailers, the quick-win of "complete attributes to activate filters" delivers higher ROI per hour invested than on any other Colombian VTEX retailer.

• 30-40% of buyers in many categories use filters to refine.
• The filter is a high-intent zone — the buyer already knows what they want.
• Appearing inside the filter is almost as valuable as appearing on the first page of general results.

Jumbo (Cencosud): where organic content fights for slots 4-10

Jumbo.com.co also runs on VTEX, but with different configuration and, above all, a different commercial feature: slots 1-3 of many searches are dominated by paid placements (sponsored). That means organic content fights harder for slots 4-10, where there's significant traffic but not infinite.

Practical implications:

• The first results screen is more expensive to win organically on Jumbo than on Éxito or Olímpica, because part of it is blocked by sponsored.
• The second screen (slots 4-10) is the organic battle zone. Optimizing for slots 5-7 may be more realistic than for slot 1.
• Jumbo's engine tolerates more keyword stuffing in titles — but only up to a point. Well-structured 80-120 character titles outperform short titles.

Another feature: Jumbo has Cencosud-proprietary categories that don't map 1:1 with Éxito's. Making sure your product is in the right Jumbo category is sometimes more complicated than it looks, and a misclassification can cost you visibility for relevant searches.

Olímpica: the quick-win zone

Olímpica.com is the most interesting case for brands wanting fast performance. PDPs are typically less populated than on Éxito and Jumbo. Titles are shorter, descriptions briefer, attributes partially filled, images few.

This is an opportunity. The bar to "win" organically on Olímpica is lower than on its peers. A brand investing 3-4 hours improving a PDP — adding 3 images, expanding the description to 200+ words, filling 8-10 attributes — can see 5-10 position moves within weeks.

It isn't magic. It's that Olímpica's category has fewer competitors doing it well, and the engine rewards relative visibility, not absolute. A "good" PDP in a category populated with mediocre PDPs is easy to make stand out.

For a brand operating across all three retailers, the optimal strategy is:

• Olímpica: content quick-wins + complete attributes to activate filter facets (less-populated PDPs + live facets = highest ROI per hour)
• Jumbo Colombia: optimize long titles and target slots 4-10 (facets still live as well)
• Éxito and Carulla: the playbook changed. Attributes still feed the ranker but are no longer a shopper-facing filtering lever. Focus on keyword relevance (title, description, correct category) and on quality signals (images, recent reviews, conversion rate)

Same signals, different priorities. That's the intelligence that differentiates an informed team from one applying the same playbook across all three.

The 5 retailer-engine alerts

The "It wasn't X. It was Y." pattern from the first article materializes specifically for the retailer engine. Five examples of the kind of alerts a well-built system delivers:

Drop from competitor content (Éxito):

"Your PDP dropped from position 3 to 9 on Éxito. It wasn't price. It was that the competitor added 4 images and 12 attributes last week."

Drop from competitor reviews (Jumbo):

"You lost 6 positions on 'lactose-free milk' on Jumbo. It wasn't your content. It was that the competitor added 47 reviews in 10 days — probably with a sampling campaign. Your PDP has 12 reviews from 8 months ago."

Video gap (Olímpica):

"You've been at position 8 on Olímpica for 'liquid detergent' for three weeks. The 7 PDPs above you all have a video. Yours doesn't. It's the only variable you're losing on — same price, better rating, more images."

New filter facet activated (Olímpica):

"Your stage-3 diapers PDP dropped from page 1 to page 2 on Olímpica. You didn't change anything. The retailer activated a new 'fit type' filter last week — competitors filled that attribute. You have it empty and no longer appear when buyers filter."

Cross-retailer gap:

"Your top SKU ranks #2 on Jumbo but #14 on Éxito with the same content. The difference: your title has 45 characters. Éxito rewards titles 70+. It's a 30-second change that can move you 8 positions."

Each alert delivers three things: a movement felt by the team, a specific cause that rules out noise, a concrete action. Without those three together, it isn't intelligence — it's another KPI panel.

What a per-retailer Health Score looks like

Operationally, monitoring the retailer engine for a portfolio of 50-200 SKUs across 3 retailers requires automation. A useful Health Score has three levels of granularity:

• SKU/retailer level: a 0-100 score combining the 6 signals with that retailer's specific weighting.
• Category/retailer level: the score distribution in that category, to position your SKU against the top 5 (where the real competitive bar lives).
• Portfolio/brand level: aggregate roll-up — what percentage of your SKUs are above category/retailer benchmark, what percentage below, where the biggest gaps are.

ePerfectStore captures the 6 signals weekly for each SKU at each LATAM retailer, computes scores with each engine's specific weighting, and produces "It wasn't X. It was Y." alerts when something crosses a threshold. The question "why did my PDP drop?" stops being rhetorical and becomes a daily answer with specific cause.

Content maturity: the Crawl → Walk → Run → Sprint path

Working the 6 signals isn't binary — it's a maturity curve. The Digital Shelf Institute, in an analysis of L'Oréal and Hasbro cases published in October 2024, describes four stages worth keeping as a map when you build your optimization plan:

• Crawl — Basic compliance. Minimum attributes, legible images, descriptions that pass the retailer's guideline. Most LATAM brands are here today.
• Walk — Completeness and algorithmic optimization. Attributes at 90%+, titles calibrated by retailer, balanced image set, video present.
• Run — Personalization and advanced measurement. Per-shopper-segment PDP variants, systematic A/B testing, continuous content health scoring.
• Sprint — AI-driven optimization. Automated generation and ranking, high content velocity, closed feedback loops between retailer and brand.

Kathleen Harrington, VP of Digital Merchandising at Hasbro, frames it this way: "You have to look a little bit deeper at those internal and external factors when thinking about driving effectiveness." The line is a direct response to teams that measure only internal performance (sales, own conversion) and leave out competitor and retailer signals.

Shazer Baig, Global Ecommerce Director at L'Oréal, complements with the strategic direction: "Our definition is evolving from simple content that converts to content that drives shopper desires across the shopper journey." The difference matters: the 2026 PDP doesn't optimize only for the click — it optimizes for the shopper to move forward in their decision.

For a CPG brand at Crawl or Walk today, the most common trap is jumping to Sprint without going through Run. Without continuous Health Score measurement per retailer, "AI optimization" produces beautiful content that doesn't move the ranker. Sequence matters.

In summary

The retailer engine isn't just one engine. It's three different engines in LATAM (Éxito, Jumbo, Olímpica), plus Walmart Mexico, plus Falabella, plus Cencosud Brazil — each with its own weights. Treating them as one penalizes you on all of them. Treating them separately gives you specific levers on each.

In the next article in the series we cover the other engine: Google. Its three surfaces (organic, Shopping, AI Mode), its 2026 standards, and the metric emerging as the most important for the conversational shelf — Intent Coverage. Read: The Google engine and the new metric: Intent Coverage for conversational search.

Sources

• VTEX — Intelligent Search. Official documentation of VTEX's search engine, the platform powering Éxito and Jumbo Cencosud Colombia. help.vtex.com.
• Walmart Connect — Listing Quality Score. Explicit listing quality scoring used by Walmart Mexico as a ranking input. walmartconnect.com.
• Sorokina, D. & Cantú-Paz, E. — "Amazon Search: The Joy of Ranking Products" (2016). Technical paper from Amazon's search team. Useful as a reference for how a retailer ranking is built in general. semanticscholar.org.
• Digital Shelf Institute — "Lessons from L'Oréal and Hasbro on Driving Content Effectiveness" (Satta Sarmah Hightower, October 2024). Crawl → Walk → Run → Sprint maturity framework, quotes from Kathleen Harrington (VP Digital Merchandising, Hasbro) and Shazer Baig (Global Ecommerce Director, L'Oréal). digitalshelfinstitute.org.