Two connected phases, both independently verified: a rigorous technical audit and re-audit that caught a developer's rebuild falling short — then a full relaunch on Shopify that put every recommendation into practice. Every number below comes from a live technical re-check, Google Search Console, or GA4. Nothing here is projected.
Before launch (Google Search Console data) vs. first 23 days after
For "bakery near me," "custom cakes," "birthday cakes"
Plus a measurable 3% from AI assistants
Google reviews — never the problem, carried through intact
These results come from two different builds of the same business's website. The audit-and-rebuild phase below is where we found — and proved — that a technical rebuild alone isn't enough. The relaunch phase is where those lessons actually shipped.
La Crème is a single-location, scratch-made bakery café at 6549 Coit Road, Frisco, TX, serving custom celebration cakes, everyday bakery items, drinks and dietary-restriction cakes — eggless, gluten-free and nut-free among them. Its core audience is Frisco and North Dallas families and event planners buying occasion cakes (birthdays, weddings, graduations), with a genuine secondary niche in South Asian-influenced flavors like Rasmalai cake. Before this engagement, the business had a strong in-person reputation but a website and search presence that didn't reflect it.
Our initial Brand Discovery Audit scored the site 40/100. After the client's team shipped an interim rebuild, our technical re-audit found the score had moved to only 47/100 — real progress on one front, but most of what actually drives discovery hadn't changed at all.
The problems clustered into four areas:
Five connected pillars, not five separate vendors.
Real server-rendered content as the foundation, a genuine XML sitemap, true 404 handling, corrected titles/meta descriptions across every page, and a 47-redirect plan to protect search equity through the platform move.
FAQPage, LocalBusiness/Bakery and Product schema sitewide, with FAQ answers server-rendered so both search engines and AI systems can actually read them — not just a human in a browser.
An AI-orientation page (/pages/llms), deliberate review of AI-crawler access, and GA4 tracking that treats AI assistants as their own measurable traffic source rather than lumping them into "Direct."
Google Business Profile correction and photo strategy, a documented Instagram content system, and consistent NAP (name/address/phone/hours) as a standing discipline, not a one-time cleanup.
A full platform move off the limitations exposed by our audits — onto Shopify, with a custom fulfillment-aware filter system, a dedicated Custom Cakes hub with live booking, a Corporate Orders channel, and real GA4 event tracking (12–18 events) so the business can see which channel actually drives an order.
| Dimension | Before | After |
|---|---|---|
| Website experience | Blank JS shell; 3 of 4 product photos broken; $0 price bug | Full Shopify store, real content, working checkout |
| Technical SEO health | 578 bytes raw HTML; no real 404s; fake sitemap | Real sitemap, true 404s, structured data sitewide |
| Search discoverability | 89 clicks before launch; generic titles everywhere | 624 clicks in 23 days; page-1 visibility (~#7) for 3 key local terms |
| AI search accessibility | robots.txt blocked GPTBot/ClaudeBot/9 AI crawlers | AI-orientation page live; AI assistants now a tracked, measured channel (3% of sales) |
| Content structure | FAQ answers invisible to non-JS readers; zero schema | Schema sitewide; FAQ + blog content indexed and ranking |
One item carried forward honestly: the Frisco/Hosur brand-name collision flagged in our original audit is not yet resolved on the production site — it's an open item, not a claimed fix.
| Metric | Before | After |
|---|---|---|
| Organic clicks (Search Console) | 89 before launch | 624 / 23 days |
| Organic impressions | 7,920 | 17,800 |
| Click-through rate | 1.1% | 3.5% |
| GA4 active users (24 days) | — | 1,658 (19,032 tracked events) |
| Sales channel mix | Unmeasured | 70% Google Search, 3% AI assistants |
| Weekly order volume | ~4/day benchmark | Matched benchmark in week one; 27–28 orders/week |
One number worth explaining honestly: average search position slipped slightly from 6.2 to 7.9. That's not a ranking loss — new content is now showing up for far more, often harder, search terms than the old site ever attempted, which pulls the average down even as total clicks and impressions grow.
Still being measured: AI-citation rate on live GEO queries for the production site (not yet re-tested since launch), backlink/referring-domain count, and review-funnel conversion once that program launches.
A technical rebuild and real growth are two different things. We proved that by re-auditing a developer's rebuild and finding the highest-weighted problems — entity clarity, reputation signals, AI-crawler access — hadn't moved at all. Then we built the version that actually did: real rendering, real schema, real measurement, and a platform that could carry all three. That's the difference between a site that looks fixed and a business that's actually easier to find, understand and choose.
Get your Online Visibility Report from SumukhAI — search visibility, AI presence, competitors and reputation, in one report.
Get Your Free AI Search Visibility Audit →Tell us where things stand today — we'll tell you honestly what's working, what isn't, and what we'd actually do about it.
Talk to SumukhAI →