We Pointed Our Own Audit Tool at Our Own Website. The Blog Page Scored 'Poor.'

There is a chiropractic practice in Assonet I think about a lot. Not a specific one. A composite of the kind that exists in every village in this part of Massachusetts: twenty or thirty years in the same building, a patient list built entirely on the fact that somebody's mother went there in 1998 and got better, and a website somebody's nephew put up during the pandemic.

That practice does not have a marketing problem. It has the opposite of a marketing problem. It has demand that arrived without being asked for, which is the hardest kind to manufacture and the easiest kind to quietly lose.

And when a practice like that finally decides to expand onto the internet properly, the thing it deserves is not a sales pitch. It deserves to see someone else's numbers first. Including the ugly ones.

So here are ours.

What we did#

Avalon runs an audit tool. It measures websites the way Google measures them: mobile Largest Contentful Paint through the PageSpeed Insights API, accessibility and SEO scoring through Lighthouse, plus direct checks for structured data, tappable phone numbers, contact forms, booking systems, analytics, and certificate validity. We point it at local businesses, and then we write up what it found.

On 31 July 2026 I pointed it at avalonpartner.com.

The numbers#

Google's thresholds, not ours: an LCP of 2.5 seconds or less is "good," and anything over 4.0 seconds is "poor."

Read that table again. Our homepage scores 95 out of 100 on performance and still misses Google's good threshold. Our blog index — the page this article lives on — is over the line into "poor."

A 95 that fails is the most useful number in this entire article, and it is why I wanted to publish this one rather than a client's. The score is not the goal. The score is a weighted average of a lot of things, most of which we got right. The one thing that determines whether a person on a phone in the Stop & Shop parking lot sees your page or leaves is the LCP, and that single number does not care how good the average was.

No one really wants the fastest website. If someone wanted the fastest website, they'd let me delete all of their CSS, delete all of their JavaScript, use Times New Roman. It's a balancing act and ultimately what we want is the most effective website.

Roberts is right, and it cuts both ways. We are not going to strip this site to Times New Roman to win a number. But "effective" has to mean something measurable, and for a blog index sitting at 4.1 seconds on a phone, it does not yet.

What else it found, including the parts that embarrassed us#

No online booking. Our tool checks for the common booking platforms. It found none, because there are none. We have a contact form and a phone number. We have told a great many local businesses that "call us during business hours" is a tax on the customer, and we are currently charging it ourselves.

Our tool said we had no analytics. Our tool was wrong. The detector looks for a fixed list of tracking patterns — Google Tag Manager, Google Analytics, Meta Pixel, Clarity, Hotjar, Plausible, Fathom, Matomo, PostHog and a few others. It found zero and reported "no analytics installed at all." That reads as damning in an audit, and we have written that exact sentence about other people's sites.

It is a false negative. We run Vercel Analytics, which loads from a first-party path on our own domain rather than a third-party script, so it does not match any pattern in the list. I confirmed it by hand: the script returns HTTP 200 from our own origin.

That is a bug in our tool, found by pointing our tool at ourselves, and it means every past audit that reported "no analytics" carries a small chance of the same error. It is now on the fix list. I would rather write this paragraph than not write it.

The things that passed. LocalBusiness schema present. Meta description present. Mobile viewport set. Tappable phone number. Contact form reachable on five of the eight pages crawled. Accessibility 100, SEO 100, best practices 100 on both pages where we scored those categories. Cumulative layout shift of exactly zero — nothing jumps under your thumb as the page loads. TLS certificate valid, issued by Let's Encrypt, expiring 19 October.

What we could not measure. Google's field data — CrUX, the record of what real visitors on real phones actually experienced over the trailing 28 days — is not available for our origin. Google only reports it for sites with enough traffic. Every speed number in this article is therefore a lab estimate, not a record of real visits. That is a genuine limitation and I am not going to dress it up. Our HTML delivers in 0.29 seconds, but that figure is HTML delivery only and is not load time; anyone quoting it as load time is misleading you, including if it is us.

The one cause Lighthouse actually named. Unused JavaScript: roughly 450ms and 101 KB of savings available on the blog index, 320ms and 115 KB on the homepage. That was the only opportunity it flagged. Separately, the blog index carries 930 KiB against the homepage's 464 KiB — we have not yet measured where that extra weight sits, so I am not going to tell you it is the images even though that is my guess.

Now the practice in Assonet#

Here is why any of this matters to a chiropractor off Route 79.

The health practices we have measured around here are considerably worse off than we are, and they are worse off while sitting on enormous reserves of goodwill. When we audited dentists in Fall River earlier this month, five practices held 4,487 Google reviews between them and averaged 4.72 stars. Their median mobile LCP was 6.4 seconds. All five failed Google's good threshold. All five were rated poor. The slowest took 10.0 seconds. Two of four live sites had no contact form at all.

Those are not bad businesses. Those are excellent businesses with a website problem, which is a completely different diagnosis and a much cheaper one to treat.

A chiropractic practice in Assonet expanding onto the internet is not starting from zero. It is starting from thirty years of people in Freetown, Berkley and Lakeville who already trust it. The internet's only job is to not lose them, and to let the next generation of them book at 9pm on a Sunday from the couch.

What to do Monday, cheapest first#

1. Measure yourself. Free. Ten minutes. Go to pagespeed.web.dev, paste your website address, choose Mobile, and read the LCP number. That is the same tool that produced every speed figure in this article. If it comes back over 4.0 seconds, that is Google itself telling you a share of people who tapped your listing never saw your page. Do not fix anything yet. Just know the number. You cannot argue with a supplier about a problem you cannot name.

2. Make your phone number tappable. Free. Five minutes. On a phone, a phone number that is plain text is a phone number nobody calls. It needs to be a link. Whoever maintains your site can do this in one line; if it is Wix or Squarespace, it is a checkbox in the link settings. Six of ten jewellers we measured in Fall River failed this one. It is the single highest-value five minutes on this list.

3. Compress your images. Free. An afternoon. Almost always the cause. If your homepage is carrying photographs straight off a camera, that is your LCP. Run them through Squoosh, save as WebP, target under 200 KB each. The photographs on this site are WebP, and the full-size versions average about 205 KB — slightly over the target I just gave you, which is exactly the kind of thing you only find out by measuring.

4. Add online booking. Roughly $0 to $30 a month. The one we do not currently practise. For a chiropractic practice this is the difference between capturing a new patient at 9pm Sunday and hoping they still remember you at 9am Monday. Several scheduling tools have free tiers that are entirely adequate for a single-practitioner office.

5. Put LocalBusiness schema on the site. This tells Google your hours, address and phone in a format it does not have to guess at. Nine of ten jewellers we measured did not have it. It is one block of code on one page, and most site builders now have a field for it.

Do one and two this week. They cost nothing and they are the two most likely to produce a phone call.

Why we published this#

Because the audit only means something if it applies to us. We measured our own site with the same tool, published the failures alongside the passes, admitted a bug the exercise exposed, and told you plainly which numbers we could not get. If you would rather work with someone who only ever shows you the 95, they are not hard to find.

If you run a practice or a shop around here and want to know your actual numbers before anyone tries to sell you anything, I will run the audit and send you the results. No charge and no obligation, and if the answer is that your website is fine, I will tell you that too.

774.559.8992 · Joshua.Amado@AvalonPartner.com

---

Method: All figures measured 31 July 2026 via the Google PageSpeed Insights API, mobile strategy, on avalonpartner.com. Thresholds are Google's published Core Web Vitals thresholds, not ours. No field (CrUX) data was available for this origin, so all speed figures are lab estimates and are labelled as such. Dentist figures are from Avalon's Fall River dentist audit, July 2026. Any of it can be re-run by anyone at pagespeed.web.dev.