michelestaxoffice.com scored 36/100 on the website scan.
Major issues
michelestaxoffice.com · Scanned 2026-10-13
49 passed22 warnings16 failed
- Technical SEO & indexability7/10 of 10 points
- On-page SEO & content8/25 of 25 points
- AI search / AEO / GEO readiness8/35 of 35 points
- Local SEO & entity signals1/5 of 5 points
- Speed, scripts & mobile8/20 of 20 points
- Conversion & lead generation4/5 of 5 points
Why AI search is 35 of the 100 points
Google is sending fewer people to websites. When it answers with an AI Overview, the top-ranked page's click rate is 58% lower than it would otherwise be (Ahrefs, 300,000 keywords, December 2023 against December 2025), and 68% of Google searches now end without a click at all (SparkToro, January–April 2026). The answer comes more and more from Google's AI, ChatGPT, Gemini and Perplexity, and they name the businesses whose sites they can read and quote. That is what this part of the scan measures, so it carries the most points. Every area's points fall faster than its checks: a warning counts a quarter of a pass, and an area half done earns about a third of its points, because search and AI answers go to the sites that finish the work.
What to fix first
-
✗
FAQPage schema +5 points
No FAQ — not in the schema and not on the page. Question-and-answer blocks are what the answer engines lift most readily.
What we’d do: Every substantive page carries a visible FAQ with FAQPage schema built from the same text.
-
✗
FAQs on the inner pages +5 points
The service page read carries no FAQ. The questions people ask about a service are what the answer engines quote.
What we’d do: Every service, area and post page carries its own researched questions — four or more, the ones people ask about that subject, none repeated from another page — visible and in FAQPage schema.
-
!
Questions answered on the page +4 points
Only 1 question asked and answered on the pages read (1 on the home page). Answer engines quote pages that ask the customer's question and answer it in a sentence or two.
What we’d do: Every service page answers the questions people ask about that service — four or more, each asked as a heading and answered in its first sentence — with FAQPage schema from the same text.
-
✗
Schema on the inner pages +3 points
The service page (/services) has no structured data. Past the home page the engines get no structured statement of what a page is or whose it is.
What we’d do: Every page carries the business node, a BreadcrumbList and its own type — a Service on each service page, BlogPosting on each post — so an engine knows what each page is and whose it is.
-
✗
Scripts and third parties +2 points
60 scripts, 867 KB of JavaScript, from 8 third parties (GoDaddy, cdn.trustedsite.com, cdn.reamaze.com and www.trustedsite.com and more). Google CDN 0.3 s held the phone's main thread while the page loaded. A phone downloads, parses and runs all of it before the page settles.
What we’d do: Our pages load one analytics tag and the form's bot check — two third parties and under 100 KB of JavaScript — and nothing runs before the page is drawn.
Every check, by category (103 checks)
Technical SEO & indexability 7 of 10 points
-
!
Canonical URL
No canonical link. With www, trailing-slash and tracking variants of the URL, Google has to guess which one to index.
What we’d do: Every page names itself as its own canonical URL.
-
!
Security headers
Missing: X-Content-Type-Options, Referrer-Policy.
What we’d do: Content-Security-Policy, X-Content-Type-Options, frame protection and Referrer-Policy on every response.
-
!
Sitemap named in robots.txt
robots.txt has no Sitemap: line — Google and Bing have to find the sitemap some other way.
What we’d do: Every robots.txt names the sitemap.
-
!
Lighthouse SEO score
Lighthouse SEO 75/100 — the report names what to fix.
What we’d do: Our sites score 100 on Lighthouse SEO: every page crawlable, titled, described, with valid links and alt text.
-
!
Sitemap dates
Every page in the sitemap carries the same date (2026-01-13) — a build stamp, not the day each page changed. Google ignores dates it has learned not to trust.
What we’d do: Each page's lastmod is the day its own content last changed, written from the content's hash, never the build date.
-
•
HTTPS certificate
The certificate could not be read from this platform, but the page was fetched over HTTPS.
-
•
Structured data Google accepts
Skipped — no structured data on the pages read.
-
✓
Indexable by search engines
No noindex on the page or in the headers, and robots.txt lets Googlebot in.
-
✓
HSTS header
Strict-Transport-Security is set — browsers remember to use HTTPS.
-
✓
Mixed content
Every script, stylesheet, image, frame and form on the page is loaded over HTTPS.
-
✓
Sitemap
Sitemap at https://michelestaxoffice.com/sitemap.xml — an index of 1 sitemap.
-
✓
Missing pages return 404
A made-up URL returns 404, so Google never indexes ghost pages.
-
✓
HTTP redirects to HTTPS
http:// redirects (308) to https://michelestaxoffice.com/.
-
✓
Internal links resolve
All 4 internal links read on the home page answer.
-
✓
Language declared
<html lang="en-US">.
-
✓
Redirect chain
The URL answers directly, with no redirect.
-
✓
robots.txt
robots.txt is served (1 group).
-
✓
www and apex agree
www.michelestaxoffice.com redirects (301) to michelestaxoffice.com.
-
✓
Sitemap lists the live URLs
The sitemap's 5 URLs are all on https://michelestaxoffice.com, and the pages read from it answer directly and are open to indexing.
-
✓
Lighthouse best practices
Lighthouse best practices 96/100.
On-page SEO & content 8 of 25 points
-
✗
Meta description
The home page has no meta description, so Google writes its own from the page.
What we’d do: Every page gets a 120–160 character description written to earn the click.
-
✗
Titles across the site
the service page's title is 8 characters.
What we’d do: Every page gets its own 30–60 character title, checked in the build.
-
✗
Image alt text
1 of 1 images have no alt attribute — screen readers and image search get nothing.
What we’d do: Every image gets an alt that says what is in it; decorative ones get alt="".
-
✗
Descriptions across the site
the service page has no description.
What we’d do: Every page gets its own 120–160 character description, checked in the build.
-
!
Page title
“Michele's Tax Office” is only 20 characters. There is room for the service and the place, which is what people search.
What we’d do: We write titles of 30–60 characters with the service, the place and the name.
-
!
Pages that repeat each other
3 of the 8 sentences on the service page also appear on the home page (38%).
What we’d do: Every page is written once, for its own subject.
-
!
Canonical and share card on every page
The service page (/services) has no canonical.
What we’d do: Every page names itself as its canonical, carries its own og:image and twitter:image, and opens with one H1 — checked on every page before the site ships.
-
!
A page for each service
Only 1 service page in the sitemap (/tax-info). A service with no page of its own has nothing to rank for it.
What we’d do: A page for every service the business sells, each with its own title, its own words, its own FAQ and Service schema, so a search for that service lands on the page about it and an engine can quote it.
-
!
Not a stock template
Built with GoDaddy's Website Builder, from templates thousands of businesses share. Nothing is left over from the template, but the layout says “GoDaddy's Website Builder” before it says anything about the business.
What we’d do: A site designed for this business — its own layout, type and colors, written from its own facts — with nothing left of whatever it was built from.
-
•
No stock or AI-marked pictures
No pictures on the pages read.
-
•
Pictures named and described for what they show
No pictures on the pages read.
-
•
Service pages link each other
Skipped — fewer than three service pages to link between.
-
•
Pages read
Read with the home page: service page https://michelestaxoffice.com/services.
-
•
Article schema on the post
No post found in the sitemap to check.
-
✓
Social share cards
Open Graph title and image and a Twitter card with its own image. A share on any platform unfurls with a picture.
-
✓
No stock phrases
None of the phrases copy falls back on when it has no facts (“look no further”, “we pride ourselves”, “one-stop shop”, a run of “top-notch” and “peace of mind”) on the 2 pages read.
-
✓
No placeholder or template text
No template or placeholder text on the 2 pages read — no lorem ipsum, sample address or number, “add your text here”, or “undefined” where a value should be.
-
✓
Single H1
One H1: “Your Friendly Neighborhood Tax Office.”.
-
✓
Readable content in the HTML
323 words of readable text in the HTML.
-
✓
Share image loads
The share image loads (image/webp, 63 KB).
-
✓
Favicon
A favicon is linked — it shows in the tab and beside the site in Google's results.
-
✓
Title and H1 differ
The title and the H1 are different, each doing its own job.
-
✓
About and privacy pages
The home page links an About page (/about) and a privacy policy (policies.google.com).
-
✓
Titles lead with the subject
The inner pages' titles lead with their subject, not the business name.
-
✓
A menu a thumb can use
4 links in the header and menu, and the service page read is one tap from the home page.
AI search / AEO / GEO readiness 8 of 35 points
-
✗
FAQPage schema
No FAQ — not in the schema and not on the page. Question-and-answer blocks are what the answer engines lift most readily.
What we’d do: Every substantive page carries a visible FAQ with FAQPage schema built from the same text.
-
✗
FAQs on the inner pages
The service page read carries no FAQ. The questions people ask about a service are what the answer engines quote.
What we’d do: Every service, area and post page carries its own researched questions — four or more, the ones people ask about that subject, none repeated from another page — visible and in FAQPage schema.
-
✗
Schema on the inner pages
The service page (/services) has no structured data. Past the home page the engines get no structured statement of what a page is or whose it is.
What we’d do: Every page carries the business node, a BreadcrumbList and its own type — a Service on each service page, BlogPosting on each post — so an engine knows what each page is and whose it is.
-
✗
JSON-LD present and valid
No JSON-LD on the page. Google and the answer engines get no structured statement of what this is.
What we’d do: Every page carries Organization or LocalBusiness, WebSite, BreadcrumbList and the page's own type.
-
✗
The first paragraph answers
The home page's headline is followed by buttons and pictures, not a sentence; the service page's headline is followed by buttons and pictures, not a sentence. An answer engine quotes the sentence right under the heading.
What we’d do: The first paragraph under the headline answers the question the page is for — what, where, for whom — in two or three sentences, so a searcher sees it without scrolling and an engine can lift it whole.
-
!
Questions answered on the page
Only 1 question asked and answered on the pages read (1 on the home page). Answer engines quote pages that ask the customer's question and answer it in a sentence or two.
What we’d do: Every service page answers the questions people ask about that service — four or more, each asked as a heading and answered in its first sentence — with FAQPage schema from the same text.
-
!
Specifics an engine can cite
The service page (/services): 272 words, 14 numbers, 19 names (Taxes, Local, Prices, Subject, …) — thin for a page an answer engine should quote. They quote prices, timelines, counts, and named places, products and people.
What we’d do: Every page is written from the business's own facts: real prices or ranges, timelines, how many and how big, the products and brands used, the towns and neighborhoods worked in, the people by name — 300 words or more that no competitor could publish unchanged.
-
•
FAQ questions visible on the page
Skipped — no FAQ schema to compare.
-
•
Freshness dates in schema
No dateModified or datePublished in the schema. Not expected on every home page; a freshness date helps the engines trust the page is current.
-
•
Speakable specification
No speakable specification. Google's speakable is a news-only beta, so this is noted, not scored.
-
✓
llms.txt
llms.txt is published (1 KB), headed with the name and linking 5 pages of the site — a plain statement of the business for the AI engines to read.
-
✓
Readable without JavaScript
The page's 323 words are in the HTML the server sends, so a crawler that runs no JavaScript reads them.
-
✓
AI answers may quote the page
Nothing on the page stops an answer engine from quoting it: no nosnippet, max-snippet limit, noarchive or noai.
-
✓
Extractable structure
Structure an engine can lift: 5 H2s, 3 H3s, 30 list items, 1 table, 1 question heading.
-
✓
AI crawlers allowed
robots.txt lets every answer engine in: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User and the rest. GPTBot, ClaudeBot, OAI-SearchBot and PerplexityBot each get the home page from the edge, as a browser does.
Local SEO & entity signals 1 of 5 points
-
✗
Business entity schema
No Organization or LocalBusiness schema. The engines have no entity to attach the reviews, the phone and the service area to.
What we’d do: LocalBusiness (or the trade's own type) with name, phone, address, hours, service area and sameAs on every page.
-
✗
Service area stated by town
The home page does not say which towns the business serves — visitors and engines both need it.
What we’d do: The towns the business works in, named — a service-area line in the footer of every page and areaServed in the schema with each town — so a search from any of them finds it.
-
✗
Reviews on the page
No reviews, rating or testimonials on the home page, and no link to a profile that has them — the proof visitors look for first.
What we’d do: The Google rating and count, three or more reviews quoted word for word with the reviewer's first name, and a link to the profile they live on — in the HTML, so the engines read them too.
-
✗
Hours shown and in schema
No hours on the home page or in the schema. “Are they open now?” is the first thing map search and a visitor want to know.
What we’d do: The hours on every page, in the footer or beside the number, and the same hours in openingHoursSpecification, as the Business Profile shows them.
-
!
Linked to its profiles
The page links Facebook, but there is no business node to carry a sameAs naming it, so the engines have to guess that they are the same business.
What we’d do: The business node's sameAs lists every real profile — the Google Business Profile first, then Facebook, Yelp, BBB and the trade's own directories — and the footer links to them.
-
!
Site name for Google
Only og:site_name (“Michele's Tax Office”) names the site. Google asks for WebSite schema on the home page to choose the name it shows above each result.
What we’d do: WebSite schema on the home page with the business's name, and alternateName for any other name people use, as on the Business Profile.
-
!
Trust signals
Only years of experience, 25 years. Noted, not scored: a visitor comparing three sites reads whether the work is insured and guaranteed.
What we’d do: Insurance, a written guarantee and the trade's credentials, stated plainly where the visitor decides.
-
•
Business schema completeness
Skipped — no business node to read.
-
•
The trade's own schema type
Skipped — no business node to read.
-
•
A named person
No named person in the schema. Not required: a home-service company is known by its brand, and the business node carries what the engines tie the reviews and the pages to. A Person belongs on the About page only when the owner wants to be named.
-
✓
Address and phone on the page
The phone ((724) 846-4272) and where the business is (3420 4th Avenue) are on the page as text.
Speed, scripts & mobile 8 of 20 points
-
✗
Scripts and third parties
60 scripts, 867 KB of JavaScript, from 8 third parties (GoDaddy, cdn.trustedsite.com, cdn.reamaze.com and www.trustedsite.com and more). Google CDN 0.3 s held the phone's main thread while the page loaded. A phone downloads, parses and runs all of it before the page settles.
What we’d do: Our pages load one analytics tag and the form's bot check — two third parties and under 100 KB of JavaScript — and nothing runs before the page is drawn.
-
✗
Nothing blocks the first paint
4 files load before anything is drawn — styles and scripts in the head — and PageSpeed puts 1.9 s of the first paint on waiting for them.
What we’d do: The styles the first screen needs are inlined and the rest load after it, every script is deferred, and the hero image is preloaded: nothing stands between the HTML and the first paint.
-
✗
Largest Contentful Paint (LCP)
Largest Contentful Paint 7.8 s in the lab (no real-user data yet; lab runs vary, so we pass up to 4 s).
What we’d do: The hero image sized and preloaded, no render-blocking scripts, a fast first byte: our pages paint in about a second.
-
!
Lighthouse accessibility score
Lighthouse accessibility 83/100 — the report names what to fix.
What we’d do: Contrast, labels, landmarks and focus order are checked on every page before it ships.
-
!
Fonts
3 font files (56 KB) from GoDaddy and Google Fonts. 1 face without font-display, so the words wait for the file.
What we’d do: Two font families in the weights the page uses, from one host, with font-display set so the words show at once in a fallback face.
-
!
The hero image loads first
The hero image is not preloaded or marked high priority, so the browser finds it late.
What we’d do: The hero picture is the first thing the HTML asks for: sized, in a modern format, preloaded with fetchpriority=high and never lazy-loaded, so the first screen paints in about a second.
-
!
HTTP/2 or HTTP/3
Google's PageSpeed test loaded the page over http/1.1 — HTTP/1.1 loads one resource at a time per connection.
What we’d do: Every host we use speaks HTTP/2 and HTTP/3.
-
!
Page weight
2.3 MB downloaded to show the home page on a phone. On a cellular connection that is seconds of waiting, and data the visitor pays for.
What we’d do: Our pages weigh 300–700 KB in all: 20–50 KB of HTML, two fonts, pictures sized to the box they fill in WebP or AVIF, and almost no JavaScript.
-
!
What would speed it up most
PageSpeed's biggest savings on this page: Render-blocking requests (about 1.9 s); Use efficient cache lifetimes (about 1.6 s); Reduce unused JavaScript (about 1.2 s).
What we’d do: These are the first three things we would fix.
-
•
Image hygiene
No raster images on the page.
-
•
Interaction to Next Paint (INP)
No real-user INP yet — Google reports it once a page has enough visitors on Chrome. It measures how quickly the page reacts to a tap.
-
✓
Mobile viewport
The viewport meta tag is set for phones.
-
✓
No popup on arrival
Nothing covers the page on arrival: no popup, no email gate.
-
✓
Readable and tappable on a phone
Every piece of text Google's test measured on a phone clears WCAG AA contrast, and the buttons and links are big enough, and far enough apart, to tap.
-
✓
Mobile PageSpeed score
Lighthouse mobile performance 71/100. Google's own bands: 90 and up is good, 50–89 needs improvement.
-
✓
No hero slider or autoplay video
One hero: no slider and no autoplaying video in the first screen.
-
✓
Cumulative Layout Shift (CLS)
Cumulative Layout Shift 0.02 (lab). Google's thresholds: 0.1 good, 0.25 poor.
-
✓
Pinch-to-zoom allowed
Pinch-to-zoom is allowed.
-
✓
No fixed-width overflow
No inline widths of 400px or more that would push past a phone screen.
-
✓
Time to first byte
First byte in 106 ms from our server.
-
✓
HTML weight
About 25 KB compressed (137 KB of HTML).
-
✓
Compression
Responses are compressed (gzip).
-
✓
One chat widget at most
No chat widget scripts on the page.
-
✓
Static files cached
9 files served with a short or missing cache lifetime, 0 of them the site's own.
Conversion & lead generation 4 of 5 points
-
!
Analytics installed
No analytics tag found — nothing counts the visits, the calls or the forms, so nothing can be improved on evidence.
What we’d do: First-party traffic logging on every site, and GA4 as the second source.
-
•
Chat or text option
No chat or text option. Not a fault, but a share of visitors will not call; an assistant that answers in seconds, day and night, is the cheapest lead a site can add.
-
✓
Contact form
1 form on the home page.
-
✓
Call to action in the first screen
A phone link, a form or a quote button appears in the top of the page.
-
✓
Lead form wiring
The form posts to back to the page, where a script sends it, guarded by reCAPTCHA.
-
✓
Tappable phone number
2 tel: links — the number dials from a phone.
-
✓
A lead form that fits a phone
The lead form asks 3 fields, labeled, and fits a phone screen.
-
✓
No CAPTCHA puzzle
No CAPTCHA puzzle on the page.
Read with the home page: service page michelestaxoffice.com/services
Want these fixed?
Every fix above is part of the work Zorva Labs does. Tell us the business and what you want from the site, and a written, priced proposal comes back in one business day.
We can fix every issue on michelestaxoffice.com.
Zorva Labs builds, hosts and grows websites for businesses across the United States — including the fixes flagged above. Every engagement is priced to the project, with a written proposal in one business day.