From invisible, in sixty days.

Not few clicks. Zero. Across sixty days.

The homepage carried a sixty seven megabyte video that autoplayed on mobile, with the headline held behind an animation until it finished. On Tauranga mobile data a visitor got a black rectangle for several seconds, then left. Nineteen sections long, with the enquiry form buried past a free download offer.

Behind it, thirteen search terms and nine pages earning any attention at all. A business running since 2016, with a fleet, a crew and four well-known builders on the client list, was effectively absent from the place people look first.

The advertising was worse than nothing. One broad campaign buying four dollar clicks from people shopping for second-hand diggers. Five hundred and seventy eight dollars a lead.

Peard has a certified drainlayer on staff. Almost every competitor subcontracts that out, which is why homeowners spend three weeks chasing someone who does not work for the company they hired.

Their smallest machine is 1.8 tonne and goes through a standard gate, so the fence stays up on jobs where every other quote wants it removed and reinstated at the customer's expense.

Toby quotes the work himself and is on site while it happens.

None of it appeared anywhere a customer could find.

The problem was never the business. Everything worth knowing about it lived in Toby's head and on job sites, and nowhere a search engine could read.

The homepage went from nineteen sections to ten. Hero, then proof, then the form. A visitor convinced at section six was previously being given thirteen more reasons to leave.

The video now waits for someone to press play, and a still image stands in until then. A phone arriving at the site downloads a hundred and eighty six kilobytes instead of sixty seven megabytes.

The headline is now the real H1 rather than an invisible one, and it says what the business does and where. The credential strip sits inside the first screen. One call to action rather than three competing ones, and the phone number tappable in the header on every page.

Every service had its own page, and every page had a different shape. Some had no FAQ, some had no schema, one loaded a 372 kilobyte background image on mobile, and several put the testimonials below the final call to action where almost nobody reached them.

All eleven now run the same nine section structure. Hero with a credential strip, then the enquiry form at position two, then proof, then the process, then the questions people actually ask. The lead magnet sits near the bottom where it catches people who were not ready, rather than at the top where it intercepted people who were.

Not blog posts about the company. Long-form guides answering the questions people search before they hire an excavator, and the cost questions do most of the work, because someone searching what something costs is close to booking it.

What a retaining wall costs per lineal metre. Whether you need consent, and the rule that catches everyone out. What size digger fits through a standard gate. Why driveways sink. Building on sand at the Mount, which no competitor in the region had written about at all.

They cost nothing to run. An article published in July was shown again in August, and will be shown again next month, without another dollar spent on it.

The services page listed every service twice, once as long-form blocks and again as a card grid seven thousand pixels further down, and had no enquiry form at all. The about page ran thirteen thousand pixels with two calls to action, the first of which appeared at twelve thousand.

Both were rebuilt. Same structure, same form position, same credential treatment. The about page gained the accreditations section, because the page where someone reads the founder's story is exactly where external validation belongs.

This is the part nobody sees and the part that changes the numbers.

When someone clicks a search or social ad, the platform attaches an identifier to the URL. Almost every website throws it away. This one captures it on landing, holds it across every page the visitor moves through, and writes it onto the lead when they finally submit something.

Fifteen database tables, one per service, so a drainage enquiry and a driveway enquiry are never confused for each other downstream. Two independent sync functions, one to the CRM and one to the email platform, either of which can fail without affecting the other or the person filling in the form.

Then the outcome goes back to the ad platforms. Which means the platform stops looking for people who fill in forms and starts looking for people who become jobs. Every week it runs, it gets better at it.

There is also a guard that runs at seven every morning. It checks the ad account against a known-good baseline and repairs anything that has silently changed overnight. In the first month it caught four.

Search went from one broad campaign to four ad groups, each aimed at a demand pool the account's own search data had already proved existed. Seventy four keywords, phrase and exact only. Twelve ads written to their own keywords. Twelve named suburb targets instead of a blanket radius.

Cost per click went up, which was the point. The old campaign bought cheap clicks from people who were never going to hire anyone. Cost per lead fell seventy four percent.

Social went the other way. One campaign, one ad set, deliberately broad, and twenty two creatives doing the segmenting instead of the targeting. Twenty seven statics designed and produced in three aspect ratios each, because a missing ratio means a missing placement.

Most people who ask an excavator for a quote are not ready that week. They are planning a build, waiting on a consent, or getting three prices. Then they go quiet, and the enquiry you paid for goes cold while everyone is on site.

Fifteen lists, one per service plus a master. Fourteen automated flows. Forty two emails written, three per service, each continuing the conversation rather than broadcasting at people.

Immediately: acknowledgement, what happens next, and the fixed price promise restated. Day two: real numbers for their specific job, consent rules, cost per metre, what to ask any quote, no pitch at all. Day six: a soft close that explicitly says if the job has gone away, no hard feelings.

Toby asked whether the advertising could work better. The honest answer was that nothing could work better until there was somewhere worth sending people, and nothing could be improved until something was being measured.

A competitor can look at the ads and copy them in an afternoon. They cannot see the forty two articles quietly climbing, or the click ID travelling from an ad through a database into two systems and back out to the platforms, or the fourteen flows following up every enquiry while everyone is on site.

That is the part that compounds. The advertising stops the day the invoice stops. The rest of it does not.