A roofer texted me last weekend. He had just paid an agency $4,200 to "rebuild" his site. They added 80 location pages, swapped his hero image, and migrated him to a new CMS he could not edit. He sent leads went down 30% the week after launch. He wanted to know what he should do next.
This is now the most common message in my inbox. The agency SEO playbook from 2018 to 2022 is dead and most agencies have not updated their playbook. Contractors keep paying for it and keep wondering why their phone is not ringing.
I have been thinking about what I would do if I started from scratch today, on a brand new domain, in May 2026, with no existing rankings to protect, for a service business. Not "what would I tell a client to do." What would I actually build, page by page, the first week. Here is that.
What Are Most Contractor Sites Getting Wrong in 2026?
Three things, all related.
First, page count. The mass-pSEO era is over for service businesses. Google's helpful content updates have been training a model that recognizes thin location pages and marks the entire domain down for them. A site with 200 location pages and no real content underneath is now a liability, not an asset. The contractors I work with who had massive pSEO setups are slimmer and ranking better now than they did at peak page count.
Second, AI search. ChatGPT, Perplexity, Claude, and Google's AI Overviews are now where roughly 30% of service-business research happens, depending on the niche. AI search engines do not crawl the same way Google does. They favor sites that answer specific questions clearly, cite sources, and have actual entity-level information about the business. A site optimized only for traditional Google ranking is now invisible to a third of buyers.
Third, lead handoff. Most contractor sites still have a contact form that emails a Gmail inbox the owner checks at 9pm. The fastest competitor in the niche is responding inside five minutes via SMS or AI receptionist. The slow handoff is the actual reason the site is not converting, and it has nothing to do with the design.
If I were starting fresh today, I would solve all three of these in week one and skip the rest.
What Pages Would I Build the First Week?
Five. That is it. No service-area pages on day one. No blog yet. No location farm. Just the pages that need to exist for Google, AI search, and a buyer to know who you are and how to hire you.
- Homepage. Hero with one sentence on what you do, where, and one obvious next step. Below the fold: three trust signals (years, license, certs), the actual service grid, three real reviews with names and dates, one project photo per service, and a clear contact block. No carousel. No stock photo. No "schedule consultation" CTA when the buyer wants a phone number.
- About page. Founder photo, real story, license numbers, service area boundary, business hours, and the actual phone number that gets answered. This page is what AI search uses for entity-level identity. Most contractor "about" pages are three sentences and a stock photo. That is the page I would spend the most time on.
- One service page per primary service. Most contractors offer five things and primarily get hired for two. Build one real page for each of the top three services. Real photos, real process, real materials list, real warranty terms, real pricing range, real questions and answers. Not a list of features. Not "why choose us."
- Service area page (one). A single page that lists every town you actually work in, written like a person describes their territory, not a programmatic SEO grid. Include drive time, jurisdiction notes, and the towns where you do not work and why. AI search uses this for geographic relevance.
- Contact / Quote page. Phone number at the top, four-field form, three ways to reach you (call, text, form), the response time you actually deliver, and a calendar embed if you book consults. No more.
Five pages. That is the entire week-one site. Most contractor sites I see have 80+ pages and get fewer leads than a clean five-page site would.
Where Does the Old SEO Playbook Still Apply?
Some of it still works. The parts worth keeping:
- Schema markup. LocalBusiness, Service, FAQPage, Review schema. AI search reads these. Google still uses them. Adding clean schema to all five pages takes a day and pays for itself.
- Real Google Business Profile management. Weekly posts, fresh photos, and aggressive review collection. GBP is still the highest-ROI channel for most local service businesses.
- Real reviews from real customers. Dozens of them. Across Google, Angi, BBB, and at least one industry-specific platform. Reviews train AI search.
- Internal linking with intent. Service pages linking to relevant blog posts, blog posts linking back to service pages, About linking to the service area page. Three or four meaningful internal links per page, not 30.
The parts I would skip on the new site:
- Location-page farms. Build them later as service areas earn coverage, not as a launch tactic.
- Generic "Why Choose Us" content. Replace with specifics: license number, years, warranty, reviews, photos.
- Bloated blog calendars. Quality over quantity. One real post a month beats four generic ones.
- Stock photo heroes. Take the photos with a phone. They will outperform stock every time.
How Should the Site Be Built Technically?
Static-rendered Next.js. That is what I would deploy on day one and that is what I deploy for every paying client at Current Digital. The reasons are concrete:
- Page speed. Static HTML beats every CMS-rendered page on every speed metric. Page speed is now a real ranking factor for AI search engines, more so than for traditional Google.
- Schema control. Hand-written JSON-LD is cleaner and more reliable than any plugin's schema output.
- Edge deploys. Vercel or Cloudflare Pages serve from edge nodes. A buyer in Long Beach Island gets the site from Newark. A buyer in Rutland gets it from Boston. The latency advantage matters when AI search is timing the response.
- Cost. Hosting is $20 a month total. No CMS license. No plugin maintenance. No security patches.
What I would not build with: WordPress (slow, brittle, security exposure), Wix (poor schema control, expensive at scale), Webflow (good for design, expensive for the value), Squarespace (template lock-in, weak local SEO).
For most contractors, the right answer is a developer-built static site with a small CMS layer for blog updates. If they want to write their own posts, fine. If they do not, even better.
What About AI Receptionists, Chatbots, and Lead Routing?
This is where the actual ROI lives in 2026 and where most contractor sites are losing every week.
The handoff problem is simple. The buyer fills out a form on your site. The form sends an email. You see the email three hours later. The buyer has already gotten three responses from competitors and made a decision. You lost the lead before you saw it.
The fix is mechanical, not creative:
- Form submission goes to SMS, not email. The owner gets a text within 30 seconds. Use Twilio, Make, n8n, or a pre-built tool. The cost is under $20 a month.
- Auto-reply to the buyer within 60 seconds. "Got it. We will call within an hour. If urgent, call this number now." That single message buys you 90% of the lost leads back.
- AI voice answering for missed calls. When a contractor is on a roof and cannot pick up, an AI receptionist takes the call, qualifies the lead, captures the details, and texts the owner. We covered this in detail in our AI voice receptionist guide.
- CRM logging from day one. Every lead, every source, every status. Even a free Airtable base beats nothing. Without this you cannot tell which channel is actually producing.
This is the part most agencies still skip. Building a beautiful site without fixing the handoff is the contractor equivalent of building a storefront with no door.
What Would I Do for SEO in the First 90 Days?
Order matters. Here is what I would actually run, in sequence.
Days 1 to 14: Launch the five-page site, GBP claimed and verified, all schema in place, AI receptionist routing tested, SMS lead handoff live. Submit sitemap to Google Search Console and Bing Webmaster Tools. Verify entity in OpenAI's GPTBot and Perplexity's PerplexityBot crawlers (yes, you can verify your business profile in some AI search tools now).
Days 15 to 45: Real review collection. Aim for 20+ Google reviews in the first 30 days from real recent jobs. Photo updates to GBP every week. One blog post answering the highest-volume question you get on phone calls. Start a referral conversation with three adjacent service businesses (electrician with HVAC, plumber with general contractor, etc.).
Days 46 to 90: Add service area depth. One real city page per week for the top five towns you serve. Each page is 800+ words, real photo from a job in that town, specific local reference (zoning, permitting, materials common to that area). Add one blog post a week answering a real question. Start a paid local search campaign on Google Ads for the highest-value service term, capped at $300 a month, with conversion tracking from form to closed job.
Day 90: Full audit. What is ranking? What is converting? What is not? Cut anything that is not earning its place.
How Much Should All of This Cost?
Real numbers. Not what I charge clients, what I would actually spend if I started from zero.
| Item | Cost | |------|------| | Domain | $15/year | | Vercel hosting | $0 to $20/month | | Email (Google Workspace) | $7/month | | SMS (Twilio + Make) | $20/month | | AI receptionist | $0 to $200/month depending on call volume | | Photography (one day shoot) | $400 to $800 once | | Schema and site build (DIY with AI tools) | 30 to 60 hours of your time | | Schema and site build (hire a dev) | $2,500 to $6,000 once | | Google Ads first 90 days | $900 | | Total first 90 days, DIY route | Under $2,000 plus your time | | Total first 90 days, hire-it-out route | $5,000 to $9,000 |
That is the honest budget for a site that will outperform 90% of contractor sites in your area in 2026. Most of the contractors I talk to are spending three to four times this on agency retainers and getting less.
What Should You Actually Do This Week?
If you are a contractor reading this and your site is older than two years, or you have never seen a real lead-to-close report from your agency, run this audit:
- Pull up your site on a phone. Time how long it takes to load. Anything over three seconds is hurting you.
- Search "best [your service] near [your town]" in ChatGPT and Google's AI Overviews. Are you mentioned? If not, that is the gap.
- Submit a contact form to your own site as a test. How fast did you actually respond? Be honest.
- Pull your last 30 days of Google Business Profile insights. Calls, direction requests, website clicks. That is your actual demand signal.
- Look at your last 12 months of leads. How many came from the website versus referrals, GBP, ads, or repeat customers? If the website is under 20%, the site is not pulling its weight.
If any of those tests went badly, the fix is not a redesign. It is the playbook above. Five pages, real content, fast handoff, weekly GBP, real reviews. That is what works in 2026.
If you want me to look at your current site and tell you exactly what I would change, run our free site grader and send me the report. I read every one. I do not always reply, but if there is something obvious to fix or you are clearly being taken for a ride by an agency, I will tell you. The point of all of this is fewer contractors paying for the wrong thing.