We Measured 337 SouthCoast Businesses. The Ones Ranked Highest Have the Slowest Websites.

If you own a nail salon in Fall River, you have almost certainly been told some version of this: your website is slow, Google punishes slow websites, speed it up and you will climb the map pack.

The first half is probably true. The second half, in this region, is not something I can find in the data.

I went back through every measurement run Avalon has done since July 12th — 337 businesses across 28 local markets in ten towns, from Fall River and Somerset out to New Bedford, Providence and Tiverton — and asked one question of it. Does a business's position in Google's local results have anything to do with how fast its website loads on a phone?

The myth, stated plainly#

"Make your site faster and you will rank higher in the local pack."

It is repeated constantly, usually by someone selling a speed optimisation package. It sounds obviously right. Google publishes speed thresholds, Google talks about Core Web Vitals, therefore Google must be ranking the map pack on speed.

How I tested it#

Every business Avalon audits gets two numbers recorded side by side: its position in the list Google Places returns for " in ", and its mobile Largest Contentful Paint measured through Google PageSpeed Insights. LCP is the moment the biggest thing on the page actually appears — the number a customer feels. For 58 of these sites Google had enough real traffic to report field data, meaning what actual visitors on actual phones experienced over a trailing 28 days, not a simulation.

249 businesses had a measured LCP. Of those, 205 also had a recorded position in their own market's results — a business with no website has a position but no load time, and a business that blocked our checker has neither. Those 205 are what the correlation runs on. I ranked both columns and compared them.

You can repeat this. Paste any local business URL into pagespeed.web.dev, then search its category and town on Google and note where it sits. Do it thirty times and you will get what I got.

What came back#

The medians are taken over the middle column, not the left one — a business with no website contributes a position but no load time, so it cannot be averaged into a speed figure. That is the whole reason those two columns differ.

The direction is backwards from the myth. Rank correlates with LCP at Spearman's rho = -0.18 (p = 0.0097, n = 205) — a weak but real relationship in which the businesses further down the list load faster. Against Google's 0–100 Performance score, the relationship disappears entirely: rho = -0.06, p = 0.38. That is noise.

And the gap between the top three and the bottom of the list is not itself statistically significant (Mann-Whitney p = 0.135) — which is the actual finding. There is no speed advantage at the top. Whatever is putting those businesses in the top three, it is not their websites.

That last line is the one that should end the argument. Twenty businesses in this dataset hold the number one position in their category and town. Four of them do not have a website. Not a slow website — no website. Six of the 58 businesses in the top three across all markets have none either. (Of the twenty number ones, fourteen had a site we could actually measure: four have no site at all, and two blocked our checker and are recorded as unmeasured rather than counted against them. Hence "11 of 14.")

I also went market by market and found the single fastest website in each one, then looked up where it ranked. Sixteen of the 28 markets had at least six businesses with both numbers, which is the minimum I would trust for this; the other twelve are excluded. Median position of the fastest site in its own market: 8.5. In only 3 of those 16 markets was the fastest website even in the top three. In one Fall River category the fastest site in town — 1.4 seconds, genuinely good — sat at position ten.

Why this happens#

Google says so, in writing, on a public help page most people have never opened:

Relevance, distance, popularity. Read that list again and notice what is not on it. Your website is not on it. Your load time is not on it. The map pack is ranking a Business Profile, and your website is one input among many into the "popularity" bucket, competing with proximity to the person searching — which you cannot change without moving.

The most useful independent work on this comes from Joy Hawkins at Sterling Sky, who ran 8,186 businesses across 200 cities through the same question. Her finding on reviews matches what I see locally: it is not the lifetime total, it is the recent flow. Her closing line is the right instruction for anyone reading this:

So should a Fall River nail salon ignore page speed?#

No. And this is where most contrarian posts get it wrong, so let me be careful.

Speed did not predict rank in this data. It absolutely predicts whether the customer who found you is still there ten seconds later. Those are different problems with different owners. Ranking is Google's decision. Staying on the page is the customer's decision, and the customer is much less patient than Google.

Look at what happens after somebody taps your listing. Across all 249 measured sites in the region the median mobile LCP is 6.6 seconds, and 197 of them — 79% — fail Google's "good" threshold of 2.5s. The closest comparison I have to a nail salon is hair salons, and I have measured four of those markets — Fall River, Dartmouth, Somerset and Swansea. In Fall River the median was 6.8 seconds across four live sites; in Dartmouth 8.1 seconds across ten. A woman standing in a parking lot deciding where to get a fill is not waiting seven seconds for your gallery to render. She is going back to the map and tapping the next one.

What to do Monday, cheapest first#

Free, twenty minutes — fill in your Google Business Profile properly. This is the one that actually touches rank, because it feeds "relevance" directly. Add every service by name: gel manicure, dip powder, acrylic full set, pedicure, nail art, repairs. Add your real hours including the holiday hours. Add attributes — parking, appointment required, wheelchair access. Then upload four recent photos of actual work. Hawkins's study found photos move the needle specifically in visual industries, and yours is the most visual industry there is. Google's own page says businesses with complete and accurate info are more likely to show up.

Free, ten minutes a week — ask for reviews now, not eventually. Recent review flow beat lifetime review count in Hawkins's data, and reviews with written text beat bare star ratings. Ask the client in the chair, while she is looking at her hands and happy. Two a week, consistently, beats forty in a burst last spring.

Free, one hour — fix your hero image. This is the speed fix, and it is a conversion fix rather than a rank fix, which is exactly how you should think about it. In almost every slow site I measure, the single biggest cost is one enormous photograph at the top of the homepage, often straight off a phone at 4 MB. Drag it into squoosh.app in your browser, export it as WebP at about 1600px wide, and re-upload. That one swap routinely takes seconds off LCP and costs nothing.

Paid, and worth it before a redesign — put booking on the page. Across the four hair salon markets I have measured, 11 of 17 live sites gave a visitor no way to book without calling during business hours. A booking link that works at 9pm converts customers your competitors are losing. Nobody who ranks above you is being rewarded for their website, so you are not fighting uphill here; you are just being better where it pays.

What I could not measure#

Straight, because it matters more than the headline.

• No nail salon was measured for this article. The Google Places API returned HTTP 403 for every discovery request today, so the nail salon audit for Fall River did not run. Every number above comes from the 28 markets already in Avalon's dataset — chiropractors, florists, HVAC, movers, restaurants, hair salons, vets and others. The closest analogue to your business is the hair salon markets, and I have labelled those explicitly rather than pretending they are yours.
• This is observational, not an experiment. I did not speed up a website and watch its rank. I measured many businesses once and compared them. Correlation of this kind cannot prove that speed has no effect on local rank — only that across 28 markets I cannot find one, and that the effect I can find points the wrong way.
• "Rank" here means the order Google Places returns for "category in town." It is not the personalised three-pack a specific person sees on a specific phone in a specific parking lot. Distance shifts that for every individual searcher, which is precisely one of the three factors Google names.
• Social presence was not part of this test. Facebook and Instagram cannot be verified reliably from a server, so Avalon reports those as unmeasured rather than guessing.

The short version#

If somebody quotes you for a speed optimisation package and the pitch is "this will get you to the top of Google Maps," ask them for their data. Mine says otherwise, and mine is 337 businesses in your own towns.

Speed up your site because the woman deciding between you and the next salon on the map will not wait. Fill in your Business Profile and keep your reviews coming because that is what the ranking is actually built on. Those are two different jobs. Do both, but stop paying for one and calling it the other.

If you want the measurement run for your own market — nail salons, Fall River, your five nearest competitors, with the numbers and no sales theatre — I will run it.

774.559.8992 · Joshua.Amado@AvalonPartner.com

Method: 337 businesses across 28 local markets audited between 2026-07-12 and 2026-08-17. Mobile LCP via Google PageSpeed Insights; field data used where Google reports it (58 sites), lab estimate otherwise. Thresholds are Google's own — 2.5s "good", 4.0s "poor". Businesses that blocked our checker are recorded as unmeasured, never as "no website". All businesses anonymised.