Automated vs. Manual Bidding: My Professional Insights from my Adventures (and misadventures) 

Automated vs. Manual Bidding: My Professional Insights from my Adventures (and misadventures) 

Strolling down memory lane, I think back to the good old days of digital marketing. Days filled with Excel spreadsheets, data analysis, and meticulous manual targeting. From sunrise to sunset, we crunched massive amounts of data, determining the most appropriate CPC (Cost Per Click) bids, and unraveled the mysterious science of driving conversions. The hustle was real, but the gratification of boosting sales at a lower CPA (Cost Per Acquisition) was unparalleled. The only taste of automation we had back then was the simple, yet efficient Bulk upload editor. 

With the advent of machine learning and artificial intelligence, the digital marketing landscape is transforming at a staggering pace. The hands-on, granular control we used to have been gradually shifting towards automation and AI (Artificial Intelligence). And as we stand on this precipice of change, it’s essential to assess when to embrace automated bidding and when to stick with the old faithful – manual bidding.

When is Automated Bidding Right for you?

My experience has shown that automated bidding can be your best ally when it’s used persistently and prudently. It’s not a set-it-and-forget-it tool, but rather a dynamic aide that requires fine-tuning and constant monitoring. When we tested Google’s Smart Bidding on our client accounts, we witnessed a significant improvement in performance. It felt like unearthing a hidden treasure in the quest for conversions. AI was our knight in shining armor, consistently delivering enhanced campaign results. 

However, there’s a caveat. Just recently, I ran a Click to Messenger Ad on Facebook with the objective of attracting business inquiries. The outcome was baffling. My inbox was flooded with messages from unqualified leads, majorly the elderly from remote provinces in the Philippines. I figured that the audiences had a high propensity to click on messages (not to talk to an agency) but were not potential clients for a digital marketing agency. That experience was a stark reminder that machine learning, while efficient, is not always effective, especially if you lack data.

So, when is automated bidding should be used on Google ads and Facebook ads?

  • Ample Conversion Data: Automated bidding works best when it has a large pool of conversion data to learn from. This allows the algorithms to better understand the trends and patterns associated with successful conversions. If your Google Ads or Facebook Ads campaigns have been regularly accumulating significant conversions (recommended at least 30-50 conversions in the past 30 days for Google Ads and at least 50 conversions per week for Facebook’s Conversion Optimization), then automated bidding can be highly effective. 
 
  • Time Constraints: When you are pressed for time and can’t afford to constantly monitor and adjust your bids, automated bidding comes to the rescue. It shoulders the responsibility of real-time bid management, freeing you to focus on strategic tasks like developing compelling ad copy, refining your targeting, and strategizing your campaign. 
 
  • Driving Conversions: If your main aim is to ramp up conversions or maximize conversion value within a specified budget, automated bidding strategies can serve you well. Facebook’s “Cost Cap” and “Bid Cap” strategies and Google’s “Maximize Conversions” or “Target ROAS (return on ad spend)” strategies leverage extensive data and machine learning to predict which ad placements are likely to result in conversions or high conversion value. 
 
  • Venturing Into New Territories: Whether you’re experimenting with new keywords, breaking into new markets, or targeting new audiences, automated bidding can fast-track your data collection. It allows you to experiment with diverse bid amounts efficiently, thereby expediting the learning phase of your campaigns. 
 
  • Managing Complex Campaigns: If your campaign management involves multiple keywords, a myriad of ad groups, or complex targeting configurations, automated bidding can significantly simplify your task. Machine learning algorithms can adjust bids at the granular level of individual keywords or targeting settings—a feat that would be overwhelmingly time-consuming if done manually. 

When should you do manual bidding?

There’s no one-size-fits-all answer to this question. However, instances when automated bidding falls short provide a compelling argument for manual bidding. Take my recent tryst with Google Ads, for instance. We were receiving many unqualified leads, such as job applicants and suppliers. While these leads might have used the right keywords, they weren’t the conversions we were aiming for. But the algorithm was registering them as such and chasing these clicks, causing our budget to dwindle. In such scenarios, manual bidding offers a level of control that automated systems sometimes lack.

For me, marketers should opt-in for manual bidding during these scenarios:

  • Always-on Campaigns: Have you identified those money keywords or specific audiences that consistently drive profitable engagement to your campaigns? These are your money keywords or audiences – the ‘always-on’ and brand keywords that you need to maintain a firm grip on. In these cases, manual bidding gives you the reins to control ad positions effectively. This approach allows you to assert your dominance, ensuring that your brand remains at the forefront of relevant search results. 
 
  • Data Acquisition Phases: Machine learning, for all its brilliance, is only as effective as the data it must work with. If your data pool is lacking in quality, your conversions will reflect this inadequacy. As the saying goes, “garbage in, garbage out.” Here, manual bidding comes to the rescue. It allows you to actively steer your campaigns towards acquiring the right kind of data, giving machine learning the correct base to enhance both the efficiency and effectiveness of your campaigns. 
 
  • Data Reset: There will be moments when you realize that your automated bidding strategy hasn’t been working as expected. It has been chasing the wrong leads or inflating your costs. Contrary to widespread belief, this is not the end of the world. These moments of error provide opportunities for learning and growth. Switching to manual bidding in such instances lets you reset the data, offering a fresh start. You can recalibrate your campaigns, hone your targeting, and reestablish the path to achieving your desired outcomes. 

Making Manual and Automated Bidding Strategy Work Together

Choosing between manual and automated bidding can sometimes feel like an intricate balancing act. Each has its merits and drawbacks, and understanding when and how to use them is key. However, it’s not always a black-and-white choice. Combining manual and automated bidding can often yield the best results, provided they’re used effectively. Let’s explore the strategies, steps, and tricks to harmonize these two bidding approaches and maximize your campaign performance. 

1. Understand the Nature of Your Campaigns

It’s crucial to comprehend that not all campaigns are created equal. Some campaigns are more stable, with predictable performance, while others may be more volatile due to various factors such as seasonal trends or competitive landscapes. For stable campaigns with a wealth of conversion data, automated bidding could be a great option. On the other hand, volatile campaigns or those with sparse conversion data might benefit more from a manual bidding strategy. I’m a control freak so that’s which points to my second tip.  

2. Start with Manual Bidding

When launching a new campaign, consider starting with manual bidding. This allows you to gain a better understanding of the campaign’s performance, how certain keywords perform, and what kind of CPCs you’re dealing with. This information can be invaluable when you switch to automated bidding, as it will provide a benchmark and let you better evaluate the performance of automated bidding. I usually test ad groups and audiences differently and whenever I feel that I have a good data representation and it can be scalable, then you can start to trust that automated bidding can deliver.  

3. Gradually Transition to Automated Bidding

Once your campaign has a substantial amount of conversion data, and you’ve identified trends and insights from your manual bidding strategy, you can start transitioning to automated bidding. Similar to Facebook, Google’s automated systems work best when they have a large amount of conversion data to learn from.  

4. Regularly Monitor and Adjust

Whether you’re using manual, automated, or a combination of both bidding strategies, regular monitoring is essential. Automated strategies may require less day-to-day management, but they still require oversight. Keep an eye on your performance metrics and be ready to step in and make manual adjustments when necessary. This could mean adjusting your target CPA or ROAS, or even switching back to manual bidding if the conditions require it. 

5. Use a Hybrid Approach

For Facebook Ads, the process of ad creation and optimization requires both a systematic and creative approach. Before we fully entrust our campaign to Facebook’s Lookalike Audience (LLA) feature, powered by machine learning, we need to ensure the data is as clean and as high-quality as possible. Herein lies the importance of what I like to call the “rinse and cleanse” method. 

The “rinse and cleanse” method is like a purification process for your audience data. We start by creating ad sets for each audience expansion. Each ad set serves as a distinct playing field for a particular audience group. This approach provides a clear picture of each audience’s performance, enabling us to assess and adjust our strategies effectively. 

As the campaign progresses, we begin finding outliers – those users within our audience sets that don’t align with our ideal customer profile or those whose interactions are not contributing to our campaign objectives. The “rinse” part involves systematically removing these outliers from our audience sets. Think of it as pruning a tree – cutting away the parts that don’t contribute to its growth to allow for healthier development. 

Once we’ve done a thorough “rinse,” we move to the “cleanse” phase. This stage involves creating a data association of the performing audiences. We look at the characteristics of the users that are positively impacting on our campaign performance. By recognizing these trends, we can further refine our audience sets, focusing on those user groups that are most likely to contribute to our campaign objectives. 

Only after we’ve gone through this “rinse and cleanse” process do we let machine learning take the reins. By this point, our LLA feature has been provided with high-quality, streamlined audience data. This ensures that the machine learning algorithms work in our favor, using reliable data to find similar high-performing users within Facebook’s vast user network. 

6. Don’t Forget About Manual Overrides

While it might seem counterintuitive in an automated setup, manual overrides can be highly beneficial. For instance, if you notice that certain keywords or audiences are underperforming or costing too much per engagement, you can make manual bid adjustments to these keywords. This allows you to retain some control while still enjoying the benefits of automation. This is effective specially if you have clients who don’t care about data-driven bids but would want to always show their ads up as I mentioned in the scenarios above. 

7. Refine and Retest

Finally, always keep refining your bidding strategies. What worked once might not work indefinitely. Regularly retesting and tweaking your approaches is the key to staying ahead. You may find that different strategies work better at different times, so be open to switching between manual and automated bidding as needed. There will certainly be a point when ad fatigue or audience saturation happens. When this time comes, it’s not the end of the world and certainly you can start again.  

 

In Conclusion

To wrap it up, AI and machine learning are powerful tools. They adapt, learn, and can make your campaigns more effective. But like all tools, they’re not perfect and sometimes may not make the best decisions for your specific business needs. When this happens, don’t hesitate to step in and adjust. 

Digital marketing is a thrilling journey, full of constant learning and adapting. While AI and machine learning represent the future, don’t forget about manual bidding. There’s a time and place for both, and it’s important to stay flexible. Keep learning from your successes and mistakes, and adjust your strategies as needed. 

Remember the phrase, “Garbage in, garbage out”? This rings especially true in automated bidding. If you’re going to use automated bidding, make sure your data is accurate. Otherwise, you might spend a lot of time cleaning up and filtering data. Regardless of the advancements in technology, the results you get are only as good as the data you put in. 

But you don’t have to go at it alone. At LeapOut, we specialize in both Google Ads and Facebook Ads. Our team can help you make the most out of both manual and automated bidding. So, don’t hesitate to reach out if you need help. We’re ready and eager to assist in boosting your digital marketing campaigns. 

Picture of Marvin Ortiz
Marvin Ortiz

Marvin Ortiz is a digital entrepreneur with a wealth of experience working with local and international brands as an e-commerce and digital growth expert. He is the Founder and Co-Owner of LeapOut, the Philippines leading E-commerce SEO and Digital Marketing Agency. He loves to read and hit the beach during his free time. You can get in touch with him on Twitter at @marvinortizph.

RECENT BLOGS

AEO and GEO for Local Business: The New Rules of Being Found When AI Answers First

AEO and GEO for Local Business: The New Rules of Being Found When AI Answers First I was looking at our agency’s Google Business Profile the other day. Six months of data. 11,000 views. 2,100 searches. 811 interactions. On the surface, healthy numbers. The kind of dashboard that would have made me nod approvingly two years ago.  Then a question landed that I couldn’t shake: how many potential customers searched for an agency like ours in that same window and never showed up in my dashboard at all — because an AI tool answered for them?  That number is unknowable. And that’s exactly the point.  A year ago, a customer searching “best steak near me” got a familiar result: a map with pins, a list of nearby businesses, a stack of reviews. The job of a local business was simple on paper — climb the list, get the click, win the customer.  Today, more of those same customers are asking that question inside ChatGPT, Gemini, Perplexity, or Google’s own AI Overview. They don’t get a list back. They get a paragraph. Three businesses named. Maybe five. A line or two on each. And a decision made before a single map pin has loaded.  If your business isn’t in that paragraph, you don’t exist for that search. And the search never appears in your analytics.  That’s the whole shift. Everything else flows from it.  What Are AEO and GEO, Exactly? Two acronyms are doing the rounds in marketing circles: AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization). Agencies love debating the difference. For most business owners, it’s a distinction without much of a difference.  Answer Engine Optimization (AEO) is the practice of structuring content so that AI assistants like ChatGPT, Perplexity, and voice search cite your business directly inside their answers. Generative Engine Optimization (GEO) is the broader discipline of shaping how generative AI systems — including Google’s AI Overviews and Gemini — perceive, trust, and surface your brand when customers ask questions in natural language.  Different surfaces. Same game. You’re optimizing to be the named answer, not the clicked link.  The reason it matters now is that the underlying numbers have moved fast. A Pew Research Center study of 68,000 real search queries found that when an AI Overview appeared, users clicked on results only 8% of the time, compared with 15% without one — a relative drop of around 47%. Seer Interactive’s analysis of more than 25 million organic impressions found that organic click-through rates on AI-Overview queries fell from 1.76% to 0.61% between mid-2024 and late 2025, a 61% decline. Gartner is now projecting that 25% of organic search traffic will shift to AI chatbots and voice assistants by the end of 2026. Put differently: zero-click searches now account for roughly 58 to 69% of all queries, with the rise directly correlated to AI Overview rollout.  The link economy that powered local SEO for fifteen years is being replaced by an answer economy. The currency has changed.  Is Google Maps Dying? No — But Its Role Is Changing I get asked often whether Google Maps is on the way out. The answer is no. For near-me, “open now,” and “directions to” intent, Maps is probably more durable than most parts of the search experience. Billions of people use it every month.  What’s changing is the role Google Maps — and your Google Business Profile inside it — plays in the broader search ecosystem.  For the last decade, your GBP was a destination. A customer found it, read it, and called. You optimized it so that final page view converted.  In 2026, your GBP is increasingly a data feed. It’s one of the most heavily weighted inputs AI systems use when composing local answers. Your categories, service descriptions, hours, attributes, photos, reviews, and Q&A are no longer just things humans read — they’re machine-readable signals teaching AI what to say about you when someone somewhere asks.  Three implications most local business owners miss:  Staleness is penalized harder than ever. Industry reporting now suggests that GBP profiles that haven’t been updated with fresh photos or posts in over 30 days can see dramatic drops in impressions. AI systems prefer fresh, frequently verified sources. Your profile isn’t a brochure you set up once. It’s a living feed.  A perfect 5.0 isn’t a trophy anymore. AI systems summarize reviews rather than count stars. They look for recency, volume, diversity of voice, and how owners engage with criticism. A profile with a perfect 5.0 rating and zero negative feedback can actually be flagged as suspicious by AI filters. A 4.6 with 200 recent reviews and thoughtful owner replies often outperforms it. The trust signal is authenticity, not spotlessness.  What isn’t structured doesn’t get counted. AI systems can only cite what they can confidently understand. LocalBusiness schema, service pages with clear question-and-answer structure, and consistent name-address-phone details across directories used to be nice-to-haves. They’re now the difference between being legible to AI systems and being invisible to them.  Look at our own profile again. 80% strength. Google itself is telling us there’s 20% of signal we haven’t given it yet. Multiply that across every local business I know — most are sitting somewhere between 60 and 80% — and you start to see the collective blind spot. We’ve been leaving machine-readable signal on the table for years, because the cost of leaving it there was minimal. In the answer economy, that cost compounds.  Separately, a bigger wave is approaching. Agentic AI — where AI assistants don’t just recommend a business but book the appointment, check availability, and complete the transaction on the user’s behalf — is moving from roadmap to reality. That future compresses the customer journey even further. Whoever the AI picks doesn’t just win the recommendation. They win the booking.  How Can Local Businesses Optimize for AEO and GEO? You don’t need to become technical overnight. But you do need to change what you’re playing for.  Stop chasing rank. Start earning citations.  Five moves matter more than the rest.  Treat your GBP like a product, not a profile. Publish

Continue Reading

Shopify B2B Is Now Available on Every Plan: What It Means for Merchants (and the Playbook to Launch It)

On April 2, 2026, Shopify extended its native B2B features to merchants on Basic, Grow, and Advanced plans — ending nearly four years of Plus-only access. Here’s why the announcement matters, what it unlocks for Southeast Asian merchants, and a step-by-step playbook for getting your first wholesale buyer live. The news, in one line Shopify B2B — company profiles, custom catalogs, volume discounts, quantity rules, vaulted credit cards, and payment terms — is now available at no extra cost on Basic, Grow, and Advanced plans. Previously, these features were exclusive to Shopify Plus.  For nearly four years, native B2B lived behind the Plus paywall. That paywall was the single biggest structural reason DTC-first brands didn’t touch wholesale. It wasn’t that the demand wasn’t there — it was that doing it properly meant either replatforming or stitching together third-party apps. Both were expensive. Both killed momentum.  That reason is now gone. What replaces it is a harder problem most merchants aren’t ready to face: designing a B2B offer worth buying.  Why Shopify opening B2B to every plan matters The global B2B ecommerce market is worth roughly $36 trillion — an order of magnitude larger than DTC. Most brand founders don’t feel the gap because their entire operating stack (ads, funnels, attribution, CRM) is built for the consumer. Procurement lives in a different universe.  But the signals are almost always there. A retailer DMs asking for wholesale pricing. A clinic chain places five identical orders in a month. A corporate gifting buyer asks for an invoice with payment terms. Most merchants treat these as edge cases. They’re not edge cases. They’re the opening of a second business inside the first one.  Shopify’s own data on merchants already running B2B is hard to ignore:  Up to 4.1x reorder frequency versus DTC  Up to 33% increase in self-serve orders within six months  40% higher average customer spend (Snyder Performance Engineering case)  25% reduction in back-office time  Those numbers don’t come from a new acquisition channel. They come from unlocking revenue that was already trying to happen.  What’s now included on Basic, Grow, and Advanced plans Shopify merchants on non-Plus plans now have access to:  Company profiles for wholesale buyers (separate identity from DTC customers)  Up to three custom catalogs with tailored pricing per buyer group  Volume discounts and quantity rules (tiered pricing, minimum order quantities)  Vaulted credit cards for repeat-order convenience  Payment terms — Net 15, Net 30, Net 60, and custom arrangements  Native integration with Shopify Payments, Shopify Flow, and Shopify Markets  Everything runs from one admin. One source of truth for both DTC and B2B. No plugins required.  What’s still exclusive to Shopify Plus For brands with complex wholesale operations, Plus retains meaningful advantages:  Unlimited custom catalogs (vs. the three-catalog cap on lower tiers)  Direct catalog assignment to specific companies and company locations  Partial payments and deposits  Advanced B2B checkout customization  The full suite of enterprise B2B workflows  The takeaway: Plus remains the right home for brands with dozens of wholesale accounts across multi-location buyers. For everyone else — the 90% whose B2B ambition starts with “a handful of clinics,” “twenty boutique resellers,” or “a growing list of cafes” — three catalogs is more than enough to test, prove, and scale before replatforming becomes a real question.  What this unlocks for Southeast Asian merchants Most SEA-based Shopify brands are on Basic, Grow, or Advanced. Plus adoption in the region remains concentrated among enterprise merchants. Which means native B2B, until this rollout, was effectively out of reach for the majority of brands who would benefit from it most — DTC-first operators with growing trade demand they didn’t know how to serve.  Here’s what that looks like on the ground.  The skincare brand getting DMs from clinics. A Manila-based skincare label notices aesthetic clinics and spas ordering in bulk through regular checkout, then asking for invoices and wholesale pricing after the fact. Instead of building a messy workaround, they spin up a B2B catalog with per-unit pricing tiers and Net 30 terms. Each clinic gets a company profile. Orders now self-serve, invoices go out automatically, and the founder stops being the accounts receivable department.  The coffee roaster selling to cafes. A specialty roaster outside Metro Manila has fifteen cafes on a Viber order list, each messaging their weekly orders to one sales coordinator. They move that list onto a B2B catalog with per-kilo pricing, a 5kg minimum, and vaulted card payment. Cafes log in, reorder their usual, and get dispatched the same day. The sales coordinator stops managing spreadsheets and starts calling prospects.  The apparel brand selling to boutiques. A streetwear label building a reseller network creates a tiered catalog — Tier 1 at 40% off RRP with a 50-unit quarterly commitment, Tier 2 at 30%. Each boutique logs in, sees only their pricing, and places orders without renegotiating every season. Sell-through data starts flowing in, and the brand finally learns which retailers are actually moving product versus sitting on inventory.  The wellness brand doing corporate gifting. A supplements brand gets a Q4 inquiry from a corporate wellness program for 500 curated bundles. Instead of handling it over email with a spreadsheet, they create a company profile for the client, a private catalog with the negotiated bundle price, and invoice-based payment terms. Next year, the same client reorders themselves. A new revenue line exists inside the same store.  None of these require replatforming. None require an agency to build a custom portal. All of them require the brand to decide what its B2B offer actually is. The trap: the tech is easy. The commercial design isn’t. This is the part most merchants will miss.  Turning on native B2B takes an afternoon. Designing a B2B offer that’s actually worth buying takes real thinking. What’s your MOQ? What’s your wholesale margin structure? Who qualifies for Net 30 and who pays upfront? What does pricing look like for a boutique committing to a quarterly order versus one reordering ad hoc? Do you ship to multi-location companies, and how do you handle split invoicing and taxes?  These are commercial questions, not technical ones. Shopify just removed the technical excuse. The brands

Continue Reading

GEO in the Philippines:
Why Most Filipino Businesses—Especially E-Commerce—Are Already Behind

Marv  │  Managing Partner, LeapOut Digital  │  Former Head of Search, Major US Retail E-commerce  │  April 2026 I lead a team of search specialists—SEO and SEM—for one of the largest US retail e-commerce operations before moving back to build LeapOut Digital. I’ve managed search strategy across millions of SKUs, watched consumer intent data at scale, and seen firsthand how a single infrastructure decision can either surface or bury an entire product catalog. When I say most Philippine businesses are not ready for Generative Engine Optimization—I’m not guessing. I’m pattern-matching against what I watched happen in US retail five years ago. We had the same debates. The same hesitations. The same tendency to wait until the problem was undeniable. GEO is the practice of optimizing your content and brand presence so that AI platforms—ChatGPT, Google AI Overviews, Perplexity, Gemini, Claude—cite, mention, or recommend you when users ask questions. Not ranking at position #1. Being part of the answer itself. This article covers GEO for all businesses. But I’m going to spend significant time on e-commerce specifically—because the e-commerce challenge is more structural, more urgent, and more misunderstood than most GEO content acknowledges. 🇵🇭  The Philippine context in one sentence: Filipinos are high-volume, high-trust searchers—and AI search is now inheriting that trust. When ChatGPT or Gemini gives a confident answer in the Philippines, users act on it. Being cited is no longer just a visibility play. It’s a trust play.   1. AI Is Already Deciding What Gets Bought Before we talk strategy, look at what’s already happening. These two screenshots are from real AI conversations in the Philippines on April 2, 2026. SCENARIO 1: “I WANT A DESSERT THAT CAN DELIVER TODAY IN SAN JUAN CITY” AI recommends a specific store, explains why it fits, and suggests an exact order. Beard Papa’s Greenhills won—not because they ran ads, but because their data was accessible. SCENARIO 2: “I AM A BJJ DAD LOOKING FOR INNER SPORTSWEAR THAT CAN DELIVER IN 5 DAYS” AI reads the buyer’s context, filters by delivery reliability, and surfaces specific SKUs with prices and ratings. Decathlon, ZALORA, adidas.com.ph, Nike Philippines won the citation. No ad was served. What these screenshots are telling you: AI is not just answering questions. It is making purchasing recommendations with specific products, specific prices, specific stores, and specific delivery windows. If your brand, product, or store didn’t appear in those answers—it’s not because the AI couldn’t find you. It’s because your data wasn’t structured well enough for the AI to trust you with a recommendation. 2. GEO vs. SEO: The Key Differences Understanding GEO starts with knowing how it differs from—and builds on—traditional SEO services in the Philippines. The table below captures the key distinctions.   3. The E-Commerce Problem Nobody’s Talking About Here’s the conversation I keep having with e-commerce clients: “We have 10,000 SKUs. Our site is on Shopify. We’re running Google Shopping. We’re doing SEO. Why aren’t we showing up in AI answers?” The answer is structural—and it has nothing to do with how much content you have. The Deep Catalog Problem A traditional search engine indexes your pages and ranks them. A generative AI does something fundamentally different: it reads your product data, evaluates whether it can confidently recommend a specific product for a specific user need, and makes a judgment call. For a business with 10,000 SKUs, that judgment call fails for most of your catalog because: Product descriptions are written for humans, not machines. “Premium quality, stylish design, perfect for any occasion.” This tells an AI nothing. It cannot answer “is this good for sweat management?” from that description. Attributes are incomplete or inconsistent. Size, color, material, use case, compatibility—these need to be machine-readable structured fields, not prose buried in a paragraph. Inventory data is stale or siloed. AI agents need real-time stock levels per location. If your inventory system doesn’t sync with your product pages, the AI cannot confidently recommend a product with a specific delivery window. Schema markup is missing or shallow. Most PH e-commerce stores implement basic product schema at best. The full picture—availability by variant, shipping estimates, return policy, aggregate ratings—is rarely structured correctly.   What AI needs vs. what most PH e-commerce stores provide Source: LeapOut assessment framework, industry benchmarks (Mirakl, Creatuity 2026). PH estimates based on client audits.   The Merchandising Disconnect Here’s what makes this worse for Philippine e-commerce specifically: most local brands separate their merchandising team from their SEO team. The people who decide how products are described are not the same people optimizing for search. With traditional SEO, that gap was manageable. With GEO, it’s a structural failure. AI systems make recommendations by synthesizing product attributes, reviews, delivery capabilities, and brand credibility. If your merchandising data doesn’t feed correctly into a machine-readable format, the AI simply skips you—not out of preference, but out of insufficient confidence. The merchandising fix: GEO forces a conversation that should have happened at the start of every e-commerce build: “How will a machine understand this product?” Every SKU needs structured, attribute-level data that answers the questions a customer would ask an expert: What is it made of? What is it best used for? What size/color/variant is in stock? How fast can it deliver to this location? What do verified buyers say about it? If your product page can’t answer those questions in a machine-readable format, you are invisible to AI agents regardless of your SEO rankings. 4. Can AI Actually Recommend a Specific Product From 10,000 SKUs Based on Color, Stock, and Delivery? This is the question I get most from e-commerce operators—and it’s the right question to ask. The honest answer: yes, but only if your infrastructure supports it. And most stores’ infrastructure does not. Let me break down what has to be true for an AI agent to answer: “I want a navy blue compression top in large, in stock, that can deliver to Quezon City within 5 days.”   What Actually Happens When You Ask AI to Shop for

Continue Reading