SEO auditJune 14, 2027

Five fixes worth about 2,300 more visits a month

Greta, Google sent Sorrel Street Cycles 6,180 visitors in May, down from 7,420 in December. Nothing here is beyond repair. Five things are holding the site back, and they're listed by what each one costs you.

Prepared for Greta Eklund, owner of Sorrel Street Cycles, by Idris Vaughan at Longtail Search.

6,180visits from Google searches in May 2027

≈ 2,300more a month once all five are fixed

≈ $5,150a month in orders, if 1.6% buy at $140 as they do now

The five, biggest first · visits a month

1Sold-out bikes vanish from the site700
2Category pages are slow on phones500
3Filters make thousands of copies450
4Workshop pages never say Madison350
564 pages share one title300
2,300 in all. These are estimates; how I made them is at the end.

Visits from Google, last six months

Sold-out bikes vanish, and take their visitors with them

≈ 700 visits a month

When a bike sells out, its page is deleted. 312 addresses that used to be bike pages now say “Page not found”. People still search for those models, and 87 of the pages still have links from reviews and club sites that now lead nowhere.

The evidence: in January those 312 pages brought 940 visits. In May they brought none. The fall on the chart above begins the week of the February clear-out.

The fix

Keep a sold-out bike's page up, marked “Sold out”, with three similar bikes below it. If there's a newer model year, send the old address straight to the new one.

Effort

1 developer day, plus an hour from whoever knows the stock.

What to expect

About 700 of the 940 visits back within six to eight weeks. The rest were for models nobody stocks now.

For your developer: 312 product URLs return 404. Restore each with a 200 and a sold-out state, or 301 it to the successor model. The list is in the spreadsheet sent with this link.

Category pages take 6.8 seconds to appear on a phone

≈ 500 visits a month

71% of your visitors are on phones. On a mid-range phone with a 4G signal, a category page such as Road Bikes takes 6.8 seconds to show its main content. Google counts anything up to 2.5 seconds as good, and speed is one of the things it ranks on.

The cause is photos. Each category page loads a 2.4 MB banner and 48 full-size product photos, including the ones far below the screen.

Category6.8 s

Product3.9 s

Home3.1 s

Guides2.4 s

Workshop2.2 s

Dashed line: 2.5 seconds, Google's “good”

The fix

Resize and compress the photos, and load the ones below the screen only as the visitor scrolls down to them.

Effort

2 developer days.

What to expect

Category pages under 2.5 seconds and about 500 more visits a month within three months. Phone visitors should buy more often too; I haven't counted that.

For your developer: LCP is 6.8 s on category templates. The hero is a 2.4 MB JPEG with no srcset, and grid images aren't lazy-loaded. Target: LCP under 2.5 s on 4G.

Google is reading 9,400 pages. You have 1,430.

≈ 450 visits a month

Every combination of filters (size, color, brand, price) makes a new address with almost the same bikes on it. Google has found 9,400 addresses on your site. Your real pages number 1,430: 1,180 products, 64 categories and 186 guides and other pages. Google spends most of its visits on the copies.

1,430 real pages7,970 near-copies made by filters

The evidence: a new bike takes 19 days on average to show up in search after you list it.

The fix

Tell Google that a filtered page is a copy of its main category, and stop it from visiting combinations of filters at all.

Effort

Half a developer day.

What to expect

New bikes in search within a few days, and about 450 more visits a month as attention moves to the real pages.

For your developer: filter URLs (?size=, ?color=, ?brand=) are indexable. Add rel=canonical on single-filter pages pointing to the parent category, and disallow URLs with two or more filter parameters in robots.txt.

Your workshop pages never say Madison

≈ 350 visits a month

People who need a repair search with a place: “bike repair madison”, “bike tune-up near me”. Across 11 phrases like these there are about 2,900 searches a month. You appear 14th on average, which is the second page, and got 41 clicks from them in May.

The evidence: the word Madison is on your contact page and nowhere else. Your listing on Google's map still shows winter hours and names no services.

The fix

Put the city in the title and heading of all four workshop pages, add a price and turnaround for each service, and update the map listing with current hours, services and photos.

Effort

1 day of writing. No developer.

What to expect

A top-five place for the main repair searches in about three months, and roughly 350 more visits a month. These are the people who book a $95 tune-up.

For your developer: nothing. This one is words.

64 category pages share one title

≈ 300 visits a month

The title is the blue line people click in search results. All 64 of your category pages have the same one: “Shop | Sorrel Street Cycles”. Google can't tell Road Bikes from Kids' Helmets by it, and a searcher can't either. Separately, 410 product pages use the maker's description word for word, the same text other shops show.

The evidence: your category pages get 1.0 click for every 100 times they're shown. Your guides, which each have their own title, get 3.4.

The fix

Give each category its own title, such as “Road Bikes from $649 | Sorrel Street Cycles, Madison”. Then rewrite the descriptions of your 40 best sellers in your own words: who the bike suits and what you'd change on it.

Effort

3 days of writing: half a day for the titles, the rest for the descriptions.

What to expect

About 300 more visits a month within three months, mostly to category pages.

For your developer: no code needed; titles are editable in the shop admin. A title template per category would save the retyping.

Already working

So nobody charges you to fix it.

Every page loads over a secure connection.
The layout holds up on phones. Nothing is cut off or too small to tap.
Google has issued no penalties or warnings against the site.
Your buying guides are strong. Three sit in the top three results, including “How to measure a child for a bike”.

Order of work

Four weeks, then a re-measure in September.

When	What	Who	Time
Week 1	Fix 1, sold-out pages, and fix 3, filter copies	Developer	1.5 days
Week 2	Fix 2, photos and speed	Developer	2 days
Weeks 1 to 4	Fix 4, workshop pages and map listing, and fix 5, titles and descriptions	A writer	4 days
Sept 13	I measure all five again and send you the before and after	Idris	Included

What it costs. At your developer's $800 a day, 3.5 days of code is $2,800. The writing is free if you do it in-house, or $1,800 if I do. Either way the total is less than one month of the orders it should bring.

How I measured

I crawled the whole site on June 8 and 9, 2027, and read six months of your own search data, December to May. Speed figures are the middle result of five runs per page type. Each visit estimate comes from what a page used to get, or from the share of clicks that pages usually receive in the position I expect yours to reach. They are estimates. Rankings move, and so do your competitors.

Next step. Reply to my email with a time this week, and I'll walk your developer through fixes 1 to 3 in half an hour.