Fashion Retailer Cuts Google Ads Spend by 40% After SEO Overhaul

StyleFlow is a UK-based fashion e-commerce brand selling occasion and workwear, generating approximately £2.1M in annual revenue at the time of engagement. Around 78% of that revenue was coming through paid channels — primarily Google Shopping and Performance Max campaigns.

The founder had been increasing ad spend each quarter to maintain revenue as CPCs rose. By the time we audited the account, cost-per-acquisition had climbed from £18 to £31 over 18 months. The brand had a blog, 200+ indexed product pages, and a functioning Google Search Console account — but organic was contributing less than 300 sessions a day.

The brief was simple: build organic so we can pull back on paid without losing revenue. The underlying problem was more layered than it appeared.

The technical audit identified three compounding issues. First, mobile LCP was averaging 6.2 seconds — Google's threshold for 'poor' is 4 seconds. The culprit was a combination of unoptimised product images served in JPEG (not WebP), a chat widget loading synchronously, and a size chart modal injecting layout shifts. Second, 140 of their 200+ product pages were flagged as 'thin content' by Google — most had under 80 words of unique copy, the rest being size guides and shipping policies shared across the site.

Third, and most critically, zero schema markup was implemented anywhere on the site. Competitors had Product schema, Review aggregation, and Breadcrumb markup — meaning they were winning rich snippets that StyleFlow was completely invisible for.

The keyword strategy was also off. The brand was targeting 15–20 broad terms with monthly search volumes over 5,000. These are terms dominated by ASOS, Marks & Spencer, and John Lewis. There was no realistic path to page 1 for these terms in a 90-day window — and even if there were, conversion rates on broad fashion terms are low. The opportunity was in the long tail.

Week 1–2 was entirely technical. Images were converted to WebP and lazy-loaded below the fold. The chat widget was deferred. Layout shift issues were resolved by setting explicit dimensions on dynamic elements. By the end of week 2, mobile LCP was down to 2.8 seconds across product pages.

Schema markup was implemented across all 200+ pages within the first three weeks — Product schema with price, availability, and SKU data; BreadcrumbList on every page; Review aggregation pulling from Trustpilot. Rich snippets appeared in Google Search Console within 11 days of implementation.

Keyword work focused on building a 340-keyword map organised around buying stages and occasion types. We identified 89 long-tail queries with clear purchase intent, search volumes between 100–800/month, and current ranking positions between 11–40 — the 'low-hanging fruit' band where targeted optimisation produces the fastest movement.

Thirty product pages were rewritten with these keywords. Copy was structured to answer the most common pre-purchase questions in the first paragraph, followed by product specifics, then styling context. Average word count moved from 65 words to 380 words per page — but more importantly, the content was structured around what buyers actually search for.

AEO implementation ran in parallel. We audited 50 product-category questions being answered by ChatGPT and Perplexity — e.g. 'what to wear to a garden party UK', 'best midi dresses for petite women'. The brand was absent from all of them. We created 12 supporting content pages structured as direct answers to these questions, cross-linked to the relevant product categories.

By day 90, 34 keywords had moved to page 1. Eight of those were category-level terms with over 1,000 monthly UK searches. Organic sessions grew from approximately 2,400/month to 9,200/month. Revenue attributed to organic (at the same 2.3% conversion rate as paid) reached approximately £18,500/month.

The founder reduced Google Ads spend from £6,000 to £3,600/month. Crucially, the remaining spend was restructured — retargeting and branded terms were cut, and budget was concentrated on prospecting campaigns for new customer acquisition. Paid ROAS improved as a result.

The brand now appears in ChatGPT and Perplexity responses for 12 product category searches. Traffic from AI referral is currently 140 sessions/month — small but entirely new, and growing.

One honest note: the 90-day timeline was tight and required fast client sign-off on technical changes. Brands with slower internal approval processes should expect the same results over 120–140 days, not 90.