Was this helpful?

Prospect with a real audit and a rebuilt homepage

For agencies and marketers
On this page

I prospect proof first. Before I ask an owner for anything, I show them where they stand on the searches that bring in customers, why each competitor above them wins, and a rebuilt homepage that already fixes what a homepage can fix. This tutorial goes from one business URL to a package worth sending, then through the part that is easiest to skip: sending it and following up until the deal is won or honestly closed.

What you will do

  • Qualify the business and open a deal before building anything.
  • Review the current site and map the real competitor field.
  • Pull real rankings and diagnose why each rival wins.
  • Turn the findings into an improvement register.
  • Rebuild the homepage from the register and QA it on mobile.
  • Put the package behind one link that search engines never index.
  • Send it yourself, follow up on a cadence, and record the outcome.

Time and cost

Qualifying takes about two minutes. Studying the pages that win the prospect's searches takes about thirty, and is not optional. With templates in place the package is a same-day job, and the outreach goes out that day too. The only hard cost is ranking data: at the pay-per-request rates I pay in 2026, these pulls come to under a dollar per prospect. Extras, such as review text or a second city, get their own estimate first.

Stage 1: Qualify and open the record

  1. Confirm someone can receive it

    Check that the business is real, that you could serve it, and that there is a route to whoever decides: a direct email, a mobile number, a form that reaches them. No route at all means stop, because a package nobody can be sent is wasted work. A marginal fit gets a lighter pass with fewer keywords and rivals.

  2. Check for duplicates

    Search your client and prospect records for the business name. Rebuilding a homepage you already built, or pitching a current client as a prospect, costs trust you do not get back.

  3. Open the deal before you build

    Create the deal in your pipeline at a researching stage, with the business name, website, and contact. Work that never reaches the board is work nobody follows up on.

Stage 2: Review the site and map the field

  1. Audit the homepage in place

    Recreate the current homepage and pin each issue to the element it affects: search, accessibility, speed, content, tracking, mobile, and conversion. An owner understands a pin on their own page faster than a list.

  2. Record the baseline and the tech stack

    Title, H1, meta description, word count, internal links, and structured data, plus the platform, whether analytics, a tag manager, and ad pixels are installed, and what each form collects. Missing tracking and broken forms become findings.

  3. Map the competitor field

    Search their services by city and by metro. Label each result a direct rival, an adjacent business, a directory, or a dead listing. Shortlist six to eight direct rivals, local first, plus the regional or national player that owns the broad terms, so the owner sees who holds the top and why that is not their fight.

Stage 3: Pull real rankings

Build the keyword set in four groups: 15 to 30 terms for a full package, or about 8 for a light one.

GroupPatternWhy it is in the set
Money termsService plus city, service plus city and stateBuying intent. These drive the standings.
Near me and categoryService near me, the bare service nameShows who owns the broad demand.
Service-specificOne per service line they sellFinds the gaps service by service.
BrandBusiness name, name plus cityCatches lookalike listings, directories outranking their own site, and forgotten second domains.
  1. Pull search volumes for the area and the country

    Use a ranking data API, not guesses. City-level phrases often return no volume; report that as below the reporting threshold, never as zero.

  2. Pull organic and map pack positions for everyone at once

    Every keyword, to a depth of 20. "Not in the top 20" and "not in the pack" are real values. Build a standings table: keyword, volume, their organic and map positions, the best rival's position, and the winning URL.

  3. Pull footprints and the Business Profile snapshot

    Total ranked keywords for each domain, then the prospect's Google Business Profile: rating, review count, claimed status, primary category, and who sits in the pack above them.

  4. Save every raw response with its date

    Every number in the package traces back to a saved pull. "No data" is a finding. One invented number is the end of your credibility.

Stage 4: Diagnose why each rival wins

For each money term a rival wins (cap it around eight), compare the winning page with the prospect's best page for that search, usually the homepage.

FactorWhat the winning rival usually has
1. Dedicated pageA page for this exact service, against the prospect's homepage or nothing
2. Keyword in title, H1, URLThe phrase in the title, against a title that is just the business name
3. City in title, H1, bodyThe city named where it counts, against a site that never says where it is
4. Content depthReal depth on the service, against a few short paragraphs
5. Structured dataLocalBusiness or Service schema with address, phone, and service area, against none
6. Internal linksA place in the nav and links from related pages, against a handful of pages
7. Domain footprintFar more ranked keywords overall
8. Map pack signalsMore reviews, a better rating, the right primary category, a claimed profile
9. TechnicalFaster load, fewer flagged issues, no mobile problems
10. Link authorityUnknown unless you pulled backlink data, so never cite link counts you did not pull. When nothing above explains the gap, say it is most likely links, which you did not measure.

Then write one verdict per keyword. Each names the rival's URL, cites at least two factors with their numbers, and ends in a fix. "Better content" with no number behind it is not a verdict.

"[service] [city]" ([local volume] a month)
You: [organic position], [map position]
Winner: [rival], #[n] organic, #[n] map, at [URL]
Why: their page is a [page type], [n] words, phrase
in the title and H1, [schema]. Yours is the [page]:
[n] words, title "[title]", city named [n] times.
They have [n] reviews at [rating]; you have [n].
Fix: [what they did, which you will do]

This diagnosis also works on its own, for your own business or a client's: see find out why competitors outrank you.

Stage 5: Build the improvement register

The register turns separate reports into one package: every finding becomes a row, and everything downstream (the fix list, the homepage, the plan, the outreach hook) is read from it.

Finding and evidence
What is wrong, the number, and the saved pull it came from, with the date.
Who does it better
The rival and the URL that proves it, or "nobody" when the ground is open.
Fix
The specific change, usually the thing the winning rival did.
Scope
Homepage, new page, site-wide, Business Profile, off-site, or social.
Priority
1 for money, 2 for foundation, 3 for polish.
Applied
The demo section that implements it, or the real reason it was deferred.
  1. Freeze it

    Merge duplicates, and set scope and priority on every row.

  2. Pick the one thing

    The single most damaging verified finding that the owner can check on their own phone in thirty seconds. It leads the package and the outreach.

  3. Pick the supporting facts

    A few headline numbers and the three searches that matter: keyword, their position, who wins, and why, in one line each. Nothing downstream invents its own.

Stage 6: Rebuild the homepage and QA it

  1. Study the winners first

    Open the two or three pages that win the priority searches. Note the type family, the hero visual, section order and density, how they show proof, and where the phone and form sit. A page that reads like the winners reads to the owner like a real website.

  2. Build the search skeleton from the register

    A title with the money term, city, and brand in under 60 characters; a meta description with the offer, city, and phone; an H1 a person would actually say; LocalBusiness schema with name, phone, address, hours, service area, and profile links; one section per money keyword.

  3. Use their brand and their facts

    Their colors and logo, not yours. A few of their own photos where they exist, an illustration where they do not. Real services, cities, years, and phone numbers. Reviews quoted word for word with the platform and date. A tap-to-call phone in the header, and a form.

  4. Trace every change

    Tag each section with the register rows it implements, and update the register. Every homepage and site-wide row ends as built, or deferred with a real reason such as "needs their price list."

  5. QA on mobile and desktop

    At 375 pixels wide: the menu opens and closes, nothing scrolls sideways, the hero reads, tap targets are at least 44 pixels, and tables work. Capture full-page screenshots at phone and desktop widths with a headless browser, and actually look at them.

An owner will not open five links. Build one cover page in your own brand (the demo is theirs; the cover page is your letter on top of it) and send only that URL. In order, it holds:

  1. A cover naming who it was prepared for, with a two-part headline: something true they are good at, then the problem.
  2. The one thing to look at first, with "check it yourself" steps and the date you verified it.
  3. The evidence: a few numbers, each captioned with what it means to the business, and the searches that matter, linked to the full audit.
  4. What you built, homepage first, then one card per report with a reason to open it.
  5. The plan, in the order the work would happen, with at least one step they can do free without hiring anyone.
  6. A one-line footer: you did the research and built it personally, plus the data source and pull date.

Stage 8: Send it and follow up

  1. Write the outreach around the one thing

    Lead with the finding they can verify in thirty seconds, include the cover page as the only link, make one ask, and keep a thirty-second phone version. Write it in your own voice.

  2. Send it yourself, the same day

    Tools can draft. A person reads it, personalizes the first line, and sends it through the channel the owner actually uses: email, a text, a call, or walking in. Log the send on the deal and move it to outreach sent.

No reply? Follow up on a fixed cadence, rotate channels, and log every touch on the deal.

  1. Day 0Send the package.
  2. Day 3A short email to check it arrived.
  3. Day 7A text or a phone call.
  4. Day 14One specific finding from the audit, as a last useful touch.

If your cover page logs visits, treat a visit as a reason to follow up the next day, not as a reply, and exclude your own visits or your QA will look like interest. Only a reply moves the deal forward, to a conversation and a discovery call. After three follow-ups and fourteen days of silence, mark it lost with a one-line reason. A lost deal with a reason is data; a deal whose stage no longer matches reality is fiction.

Ethics and hygiene

You are building on someone else's name without being asked. These rules keep that fair to them and safe for you.

  • Every page in the package carries <meta name="robots" content="noindex, nofollow">, stays out of your sitemap, and is not linked from your public site.
  • Nothing says or implies the business is your client: no logo in your portfolio, no "client work" label.
  • Any recreated copy of their site has their third-party scripts removed, so their chat widgets, analytics, and ad tags never run on your domain.
  • Any working form on the demo sends to your own inbox, never to theirs.
  • Their photos, logo, and copy appear only inside the noindexed package, used sparingly, with the demo labeled as a sample.
  • The package claims nothing that did not happen: not that they asked for it, and not that you have already spoken.
  • The owner gets one clear link, not a pile of attachments.
  • If they ask you to take it down, it comes down that day.

What's next

Stuck on something this guide does not cover?

I run my own practice on exactly this system. If you do marketing for clients and have a specific question, send it to me.

Ask me a question

Last updated 2026-09-13 UTC. Written by Tucker Shively.