Operator Intelligence from Evolette Locin: Agentic SEO and Technical Dominance, Singapore
1. The Operational Flaw: Optimising Pages for a Search Engine That No Longer Reads Pages
The standard founder mistake in SEO is to buy keywords. An agency produces a list of search terms with monthly volumes, writes a page for each, builds some links, and reports rankings every month. The assumption underneath is that a search engine matches words on a page to words in a query. That assumption is now only half true, and the half that is false is where the money leaks.
Modern search, including the AI answer layers that sit on top of it, resolves queries to entities: a specific business, in a specific place, offering specific things, connected to specific people. If the system cannot tell which entity your pages describe, it hedges. It merges you with a similarly named business, attributes your reviews to someone else, or skips you in favour of a source it is more certain about. A page can contain every right keyword and still lose, because the engine is not confident that the page belongs to the business the searcher wants.
Singapore makes this worse than most markets. Names repeat across sister outlets, group companies and rebrands. A single operator may have three legal entities, two trading names and a founder who appears on four websites. Every inconsistency is a reason for the engine to doubt. The cost is invisible in a rankings report because the report measures positions for keywords, not confidence in identity.
2. The Live Proof-of-Concept: Miyu, Winchester and PickleChoo
We run this system on businesses whose revenue depends on being found. Miyu Omakase operates in a category where a single listing, a single reservation page and a single consistent description decide whether a diner books or moves on. Winchester Tennis Arena and TAG International Tennis Academy compete for the query “tennis coach Singapore”, where the engine has to separate a venue, an academy and individual coaches. PickleChoo runs several outdoor pickleball locations, which is a textbook disambiguation problem: one brand, many places, each with its own address, hours and programmes.
Across these properties the work is the same. Define each entity once, in one canonical form. Publish that definition in machine-readable markup. Make every other mention on the web agree with it. We do not publish ranking or revenue figures for these businesses in a public article, and you should be cautious of any agency that does without showing the underlying data. What we can say is that the method is built and maintained by people who answer for the P&L of the businesses it serves, so it is judged on bookings, not on a vanity report. For the earlier case studies, see how Miyu operates at capacity without advertising.
3. The Execution Architecture: Six Steps to Entity Disambiguation
Step 1: Write the entity register. List every entity you need the engine to understand: the operating brand, each legal entity, each location, each named person, each product or programme. For each, record the exact name, the address in a single fixed format, the phone and WhatsApp number, the opening hours, the founding year and the relationships between them. This document is the source of truth. Everything else is derived from it.
Step 2: Compress the keyword set to intents. Run the full keyword export through an AI agent and cluster by intent, not by string. Hundreds of variants of “best omakase”, “omakase near me” and “sushi counter booking” collapse into a handful of intents. Assign each intent to exactly one page. Two pages competing for one intent confuse the engine as much as two businesses sharing one name. What used to take an agency a fortnight of spreadsheet work now takes an agent minutes, and the saving should go into the review, not into a lower standard.
Step 3: Mark up each entity in JSON-LD. Use schema.org types that match reality: Organization or a more specific business type for the brand, LocalBusiness subtypes for each location, Person for named experts, Article for editorial content. Give every entity a stable identifier, link people to the organisation they work for, and use sameAs to point to the profiles that confirm the identity, such as the official social accounts and the management team page. Markup must describe what is visibly on the page. Marking up claims the page does not make is a quick way to lose trust, and it can attract a manual action.
Step 4: Reconcile the off-site footprint. Audit your Google Business Profile, directory listings, social bios, press mentions and partner pages against the entity register. Fix every mismatch in name, address format, phone number and category. This is dull work, and it is the work that most reliably moves local results, because it removes the doubts the engine would otherwise resolve against you. For the local side of this, see our guide to Google Business Profile.
Step 5: Build the internal link graph around entities. Each location page links to its parent brand, each person page to the organisation, each article to the entity it discusses, using consistent anchor text. The site then reads as a connected description of a business, not a pile of pages. This is the on-site counterpart of what we covered in knowledge graph SEO.
Step 6: Monitor for drift, not just position. Every month, test the questions a customer would put to a search engine or an AI assistant about your business: who runs it, where it is, what it offers, what it costs to book. Record whether the answers are correct and which source they came from. When an answer is wrong, trace it to the source that misled the engine and correct that source. Rankings can hold while identity quietly degrades, so measure identity directly.
4. The Asymmetric Advantage: What Agencies and Consultants Cannot Copy
An SEO agency bills for output: pages written, links acquired, reports sent. Entity work produces very little visible output. The best result is a set of consistent facts that nobody notices, which is difficult to put on an invoice and easy to cut from a retainer. An academic consultant can explain how knowledge graphs work and still never have corrected a wrong opening time that was costing real bookings.
An operator sees the whole chain. We know which legal entity signs the lease, which trading name is on the door, which phone number the front desk answers and which of those the customer will search for. We also know the risk side. Structured data makes statements about your business that are public and machine-readable, so a wrong claim about an award, a price, a qualification or a guarantee is a legal and consumer-protection problem as well as an SEO one. Our legal background is why our markup describes only what a business can prove, and why we treat reviews, credentials and offers with care. The agent does the compression and the checking at speed. The judgement about what is true and safe to publish stays with a person who is accountable for it.
5. The WhatsApp Conversion Bridge: Request a Private Portfolio Audit
If you run more than one outlet, more than one entity, or a brand that has changed names or owners, there is a good chance search engines are already uncertain about you. The first step is an entity audit: we compare your website, markup, Google Business Profile and major listings against a single register, show you where they disagree, and rank the fixes by how much booking volume they put at risk.
Request a Private Portfolio Audit via WhatsApp.
This article is general commercial commentary, not legal or financial advice. Have any structured data claims about awards, pricing, reviews or guarantees checked against your own records and applicable Singapore consumer-protection rules before publishing.
