Kids party entertainment and decor · Seattle and the Eastside, WA

Cheaper bookings from the ad clicks they already had

Wonder Spark provides kids party entertainment and decor in Seattle and on the Eastside, WA. Their Google Ads were already pulling clicks at a fairly high CTR, and the site was where that money died: an older build with usability problems, where very few of those clicks became a filled form. We designed the new site in Figma and rebuilt it on hand-written PHP, with a six-step booking request at the center of it.

Web developmentPaid mediaBooking flow

Open the live site See how we ran it

The Wonder Spark home page as it ships today

At a glance

The project in one screen

Client: Wonder Spark
Industry: Kids party entertainment and decor
Location: Seattle and the Eastside, WA
Services: Figma design, full rebuild, six-step booking form, AI chat, SEO/GEO/AEO
Timeline: First month live: visitors from ChatGPT completing forms (client-reported)
Platform: Custom PHP, no CMS or page builder
Headline result: Cheaper cost per conversion on the ads already running (client-reported)

Custom PHP application with its own form logicServer-rendered pages, no CMS or page builderSix-step booking request with a live estimated totalExpanded contact forms across the siteResponsive layouts designed in FigmaAI chat assistantSEO, GEO and AEO optimization

The starting point

The traffic was arriving. The site was where it stopped.

Wonder Spark came to us with Google Ads already running. Click-through rate was fairly high, so the ads were getting attention. What happened after the click was the problem: the previous version of the site had usability problems, conversions were low and very few visitors completed a form. Every click was already paid for by the time it hit that page. The brief that followed was specific. Build a modern, responsive site that a potential client can actually understand, widen the forms, and add a separate long booking request that takes the date, the choice of program and the rest of the details.

Paid clicks arriving at a fairly high CTR, with few conversions and form fills behind them
Usability problems on the previous version of the site
Forms too limited for what a party booking needs, per the client's brief
No dedicated way to request a specific date and program

What we built

The engineering behind it

A codebase that could hold the booking request

The site is a custom PHP application, written from scratch for this business instead of assembled on a CMS or a page builder. That decision followed from the brief: the client wanted a long booking request that behaves like a party inquiry, and owning the codebase meant the form logic and the page markup could be shaped around it. The brief describes the build as hand-written PHP with its own logic, an AI chat and a booking form, and that is the scope this page describes.

Six steps between a paid click and a booked party

The old setup gave a parent one short way to get in touch. The shipped booking request walks through where the party is, the date and time, the show, extras, guest count and contact details, with a running estimated total beside it. A request now arrives with enough in it to answer, and the parent sees the party take shape while filling it in. That form is the center of the new site and the reason the codebase is custom.

Expanded forms across the rest of the site

Alongside the long booking request, the existing forms were widened. Not every visitor arrives ready to configure a party, so the shorter route to a conversation stayed on the site and was extended to collect more of what the team needs before replying.

A chat sitting beside the booking journey

The AI chat assistant is open on the booking page itself, where a parent halfway through the steps can ask a question instead of closing the tab, and where someone who will never start the form still has a way to reach an answer. It shipped in the same delivery as the booking flow.

SEO, GEO and AEO groundwork

The site was optimized for search engines and for AI assistants at the same time, which means writing the facts about the business as readable text on the page rather than burying them in images and scripts. What that produced in the first month is described in the AI visibility section.

Design and devices

The screens between an ad click and a booked party

Every layout on this site sits between a paid click and a request, so the whole thing was drawn in Figma before development started, where a change costs minutes. The brief set the test each screen had to pass: is this modern, does it work on a phone, and is it understandable to a parent who has never heard of this company. The booking request was designed as its own flow, separate from the short way of getting in touch, because it asks for a great deal more.

A full design pass in Figma, so the arrival screen was settled before a single paid click landed on it
Six steps from where and when to the show, extras, guest count and contact details
A live estimated total beside the steps, so a parent watches the party take shape while answering
Responsive layouts, so a parent deciding on a phone follows the same path as one on a desktop
Clarity as the stated goal: a parent should understand the offer, then be asked for details

The booking flow

The Wonder Spark booking wizard — Six steps, a running estimate, and the assistant open in the corner

What came of it

Results

The campaigns were live before the rebuild and kept running after it, so the variable that moved was the destination. The client reports that the same paid traffic started converting and that the cost per conversion dropped significantly, which is the whole economics of this project in one line: the ad budget did not need to grow, the page it pointed at needed to work. The second result was the one nobody schedules. Inside the first month live, people began landing on the site from ChatGPT and completing forms, which is what the SEO, GEO and AEO work was aimed at. The client's own summary of the project was about trust, which they describe as having grown substantially since launch.

Cheaper

Cost per conversion on the Google Ads that were already runningClient-reported.

Conversions and form completions on the siteClient-reported.

First month

Visitors arriving from ChatGPT and completing formsClient-reported.

The month parents started arriving from ChatGPT

The site was optimized for search engines and for AI assistants at the same time, which in practice means stating the facts about the business as plain readable text an assistant can quote back: what the company does, where it works, and what booking a party involves. Within the first month live, the client saw visitors landing on the site from ChatGPT and filling in forms. That is not a result anyone can promise in advance, and it is the reason the booking request had to read as clearly to a machine as to a parent.

How we ran it

The work plan, in order

Every project runs on a written plan. This is how this one ran.

01

Frame the problem from the paid search picture

The starting point was what the client already knew: Google Ads running, fairly high CTR, weak conversion. That combination points away from targeting and toward the destination, so the work was scoped to the site the ads pointed at.
02

Design the site in Figma

A full design pass before development. Layouts, responsive behavior and the shape of the booking request were agreed here, where changes cost minutes.
03

Build the custom PHP application

A fresh codebase with its own logic, written for this business, so the booking flow and the content structure could match the approved design without fighting a theme.
04

Ship the booking request and the AI chat

The six-step booking form with its estimated total went live along with the widened forms elsewhere on the site and the AI chat assistant.
05

SEO, GEO and AEO optimization

The site was optimized for search engines and AI assistants together, so the same facts about the business could be extracted by both.
06

Watch what the first month did to the ad economics

After launch the client tracked where requests were coming from. That is where the cheaper cost per conversion on the ads was reported, and where the ChatGPT referrals appeared.

Straight answers

Questions about this build

Yours is probably close to one of these. If not, ask it directly.

Ask about your project

What changed about the Google Ads?

The campaigns were live before the rebuild and the click-through rate was already fairly high, so attention was never the missing piece. What the rebuild changed is what a click landed on, and the client reports that the cost per conversion dropped significantly afterwards. This page makes no claim about who manages the campaigns.

Does a six-step booking form put people off?

A long form does filter, and that is part of its job. It also gives back as it goes: the estimated total updates while the parent answers, so the steps buy something. The site keeps a shorter route to a conversation alongside it, plus the chat assistant, so a visitor who does not want to configure a party has somewhere to go. The client's report after launch is that form completions went up overall.

Why a custom PHP build rather than a CMS or a page builder?

Because the fix list was the site. The usability problems, the responsive layouts and the forms all needed to change, and a six-step booking request with a live estimate could not be bolted onto the old build. An empty PHP codebase put the form logic and the page structure under our control.

Is the booking request the only way to get in touch?

No. The shorter contact forms stayed on the site and were widened to collect more of what the team needs before replying, and the AI chat assistant is open on the booking page for anyone who would rather ask than type into fields.

Same team, same method

Your project, next

Tell us what you sell and where you sell it. You get a plan like the one above, with the first step priced out and dated before anything starts.

Start your project All case studies

No spam 100% confidential Answer within 24 hours