The brief

Orlando has hundreds of chauffeur operators and they all show the same black SUV against the same skyline. Almost none of them will tell you what a ride costs.

So the site is built the other way round. The rate is stated before the promise is made, and the fleet is described in the words a traveller actually books in - airport, park, cruise, hourly - rather than in the words a fleet manager owns it in. Then the content programme goes and answers the questions those travellers ask.

The work

1. 01 One template, seventeen blocks Every vehicle page and every service page is assembled from the same ordered sequence. Writing a new page becomes a decision about content rather than a decision about layout, which is the only way a page count grows without the site coming apart.
2. 02 A content programme with a constitution One hundred and eighty-seven posts in ten topical hubs - deliberately cut down from two hundred and fifty-eight, because two posts on one intent split each other. Every draft is written to a framework that rotates ten narrative shapes and thirteen hooks, and passes an automated gate before it ships.
3. 03 Search results, measured in the open From position 75 to page one in a year: 4,464 clicks and 998,307 impressions, most of them on non-branded cruise and theme-park searches the blog was built to own. Roughly one booking in twelve now arrives through an AI assistant rather than a search results page, because every post is written to be quoted in isolation.
4. 04 A CRM of its own A white-labelled subscription was replaced with a custom-built CRM: calls, texts and email on one number, invoicing and payments, and a speed-to-lead automation that enrols a new booking the second it lands. Live at crm.orluxrides.com.
5. 05 And the rest of the yard The six-step booking form on the homepage, a trip manager with a customer portal and an operator app for ride day, press syndicated out to Associated Press and Business Insider, Google Ads, and a post-ride SMS review system. One vendor, one yard.

Built with

• WordPress
• Content system
• Technical SEO
• Custom CRM

The brief

Nobody searches for reverse osmosis for fun. They search for it an hour after reading something frightening about their own tap water, and an anxious search wants an answer rather than a catalogue.

The audit found the real problem before any writing started: twenty-five of the site’s most important pages - every city page, the conversion page itself - had never been crawled by Google at all. Content cannot rank from pages a crawler has never seen, so the plumbing was fixed first, and then the answers were written.

The work

1. 01 The audit came first A Search Console autopsy: seven pages indexed, twenty-five money pages never crawled, junk URLs outranking the pages that earn. The fix list ran from a cleaned sitemap and rebuilt canonicals to ten city pages whose schema each declared an invented business.
2. 02 Pillar and cluster, held together Thirty-four posts across nine clusters, each cluster owned by one worry - hardness, iron and sulfur, taste, the lab test a loan closing depends on - and every route through them ends at the free water test. A sebil hands the first cup over for nothing; so does this site.
3. 03 Written for the answer engines Every post answers its question in the first sentence, spells its entities out in full, carries a liftable fact table and FAQ schema - built to be quoted by AI Overviews and assistants as much as ranked in the blue links. The format is enforced by code: twelve blocking checks run before anything publishes.
4. 04 The lead never gets lost Two conversion landing pages for the paid work, a custom lead pipeline wired into Wix - form to contact record to dashboard notification, with an email failsafe so nothing is ever dropped - and a Google Ads plan split by water source, well against city, down to the ZIP code.

Built with

• Wix
• Pillar & cluster
• Technical SEO

The brief

A young store cannot outrank the giants for its own products, and a hundred and fifty posts written the same way become wallpaper by the fortieth. Both problems have the same answer: own the questions the giants ignore - the fabrics, the etiquette, the fit - and never take the same shape twice running.

So the programme rotates on two axes at once: the shape a piece takes, chosen from eight frameworks, and the reader it is written for, chosen from eight archetypes. Neither repeats twice in a row, and the pair is chosen before a word is written.

The work

1. 01 Heritage first, catalogue second Dupatta etiquette, jamawar, what to wear to a dholki: the guides answer what the customer is actually typing, and they rank - the dupatta guide alone sits around position five. Every piece still has a garment behind it and a route to it, because editorial that ends in admiration is a magazine somebody else is paying for.
2. 02 Archetypes, not personas The reader is chosen as deliberately as the subject - a Karachi professional, a bride in the UK diaspora, a mother of the groom - a wardrobe and an occasion rather than an average customer, so the writing has somebody specific to be addressed to.
3. 03 The audience the store was built for The search traffic splits the way the label sells: more clicks from the United States than from Pakistan, with the UK, India and Canada behind them. Diaspora buyers were the plan, and the first quarter of data shows them arriving.
4. 04 The store, the pixels, the channel Six bespoke theme sections built into the Shopify storefront, the tracking stack verified end to end - Meta pixel and server-side events with zero errors, GA4, Merchant Center - and a YouTube strategy that treats the channel as an editorial film platform rather than a catalogue.

Built with

• Shopify
• Editorial programme
• Brand voice

The brief

The store had been dark for eight months. The first job was not to reopen it - it was to find out what had gone wrong while it was open.

Two things had, and both of them were structural. Half the catalogue was not salt. And the pixel every paid campaign would have to report back to had never fired, which meant there was nothing to optimise toward and no way to know it.

The work

1. 01 The audit came before the relaunch Reopening a store that failed once, unchanged, is a way of failing twice, more slowly. Nothing was scheduled until the catalogue and the measurement had been read cover to cover.
2. 02 Half the catalogue was off-niche A salt store that also sells other things is not a salt store to a shopper or to a search engine. The split is the first thing the relaunch has to resolve.
3. 03 A missing pixel blocked the whole paid plan No pixel means no conversion signal, no audience, and no optimisation - the spend would all have been buying a number nobody could read. It is fixed first, and it is why the paid work starts after the measurement rather than beside it.
4. 04 A relaunch sequenced around both Catalogue, then measurement, then spend. In that order, because the other orders all pay for the same lesson twice.
5. 05 Tooling built for the job The audit was run with instruments made for it - authenticated store access held across sessions, the catalogue read in full rather than sampled - and the defect list came back ranked: placeholder policy text, phantom compare-at prices on nearly every variant, tax flags off on all but two.

Built with

• Shopify
• Store audit
• Paid media

The brief

The decision to move a parent into care arrives at 2am, and a six-bed home in Florida is asleep - so the family rings the next number on the list, and the home never knows the call happened.

The homes that do get enquiries mostly buy them from a referral service, at something close to a month’s rent per placement - renting back the top of their own funnel. Lantern is the argument that one vendor should own the loop instead.

The work

1. 01 A front desk that is awake at 2am Calls and texts answered in under a minute, around the clock, with the tour already on the calendar by the time the home sits down to breakfast. Nothing about qualifying a family requires a person to be awake.
2. 02 A CRM built for six beds, not six hundred Small homes are run by the person who also cooks. Anything with an implementation phase does not get implemented, so the product has to be useful on the first afternoon.
3. 03 One vendor, no joins Enquiry, tour, follow-up, placement and the record afterwards. Split across three tools it leaks at every join, and the referral service exists because of the leaks.

Built with

• Next.js
• Product
• Go-to-market

The brief

Writing about technology divides into two piles: press releases with adjectives on them, and papers nobody outside the field can read. Between the piles is a large readership that is being served by neither.

Mind Stream Tribune is written for that readership, across three pillars, each of which has to be able to hold an audience on its own rather than borrowing one from the other two.

The work

1. 01 Three pillars, each self-supporting A pillar that only works as part of a set is a category, not a pillar. Each of these is commissioned as though it were the whole publication.
2. 02 Schema and search, set up properly Article markup, an author who exists, a structure a crawler can walk, and no invented properties - one wrong claim in structured data is enough for the rest to be discounted.
3. 03 Forty-odd posts, and a shape for the next four hundred The archive is small. The framework it is being written into is not, which is the difference between a blog and a publication.

Built with

• WordPress
• Publishing
• Schema

The brief

Three shelves: what happened, what is happening, and what is about to. The argument is that they are one subject and get written about as three.

It is being built. There is nothing to link to yet, and a portfolio that links to an empty site is worse than one that says so.

The work

1. 01 Not open yet The shutter comes the rest of the way up when there is enough on the shelves to be worth the walk. If you want to know when that is, the card on the last bay of the street takes messages.

Built with

• In development