Call the Lead With the Website Already Built
She used to find the lead, judge their website, then build a demo site if they were interested. All three now happen before she picks up the phone.
- 2x the leads
- Worked in the same hours
- Day one
- Draft site ready before dialling
- 1 in 9
- Calls closed to proposal
The challenge
WebCare Digital sells websites and CRM setups to small home-services businesses: HVAC contractors, plumbers, trades. The pitch works because of a specific move. Arianna finds businesses whose website is bad or missing, builds them a draft site, and calls to say she has already made one and would they like to see it.
Everything upstream of that call was manual, in three separate stages.
Finding them. A browser extension pulled leads off Yelp, capped at 250 per account. The CSV went into Google Sheets for column cleanup and renaming to match her CRM’s fields, then imported into GoHighLevel.
Judging them. Lead by lead, she opened each business’s site and decided whether it was any good: does it load, is there a call to action above the fold, does it look professional. That judgment sorted the list into who to call and what to pitch.
Building the site. After a call went well, she assembled a demo in a GoHighLevel sub-account, updating text and dropping in a logo she would often have to make in Canva first.
Each stage capped the next. The extension limited how many leads got in, the manual review limited how many got out, and the site build meant the strongest version of her pitch only existed for prospects who had already engaged. Growth meant more hours, and there were no more hours.
What we built
One pipeline that runs all three stages and hands back a CRM ready to dial.
Sourcing. Yelp resists straightforward scraping, and the first finding was a negative one: plain HTTP requests get a 403, because Yelp fingerprints the TLS handshake rather than checking the page. Driving a real browser and reading the structured data the page already publishes works. Leads come back enriched with phone, address, website, rating and review count, cross-checked against Google Business Profile for listings where Yelp has no site on file.
Assessment. Each business’s site is opened and scored against her criteria, not a generic quality metric: call to action above the fold, professional appearance, contact details, service descriptions, trust signals. Every lead arrives tagged with whether it has a site and whether that site is any good, which is the judgment that used to cost her the day.
Site generation. For leads without a decent site, an agent builds a draft one from what the Yelp listing already has: business name, images, services, contact details. Not a mockup of a site: a real one, hosted, ready to show. Where a listing is too thin to personalise, it falls back to a trade-appropriate template carrying the business name.
“I thought it was a beautiful site. And I was actually, like, wow, it’s better than the ones I create. You guys are doing it so quick and fast.”
Delivery into the CRM. Everything lands in GoHighLevel in the right pipeline and stage, tagged by campaign, so she opens her CRM and starts dialling rather than importing a spreadsheet.
Turnaround on a new campaign is hours. Name a city and a trade, and roughly 90 leads come back sourced, assessed, site-built and loaded.
What changed
She works about twice the leads in the same time. Her own estimate, in her own words: twenty leads in the time it used to take for half that. The lead quality is identical because it is the same source; what changed is everything around it.
The site exists before the call, not after it. That is the part that moved her sales motion rather than just her admin. The pitch was always “I’ve already built you a website”, and it is now true at the moment she dials, for every lead that needs one, instead of being work she front-loaded onto prospects who might not buy.
Volume stopped being the constraint. Across campaigns: 304 leads for Tucson and Phoenix, then a Georgia batch of around 90, then roughly 181 for Los Angeles HVAC with 109 contacted. On the Georgia batch she called nine, reached two, and closed one into a proposal.
Her packages run $1,497 to set up plus $297 a month, so the pipeline pays for itself on a small number of closes.
What we got wrong
The first Google Business Profile matching was too loose. It matched on business name and address, and around 40% of matches came back as a different company with the same keyword in its name and a different phone number. Arianna caught it, and her fix was the right one: match on phone number, because two listings with the same name and different numbers are two businesses.
A batch of leads tagged “not interested” turned out to be disconnected phone numbers rather than rejections, which is a data-quality signal we were reading as a sales signal. Duplicate leads from a geography she had already worked herself made an early pipeline messy.
None of that is unusual for a first pass, and it is in this write-up because the weekly feedback loop that surfaced it is the reason the pipeline works now.
Why NimbleBrain
Discovery mapped her whole process before anything was built, including the parts that were working. The manual website assessment was the bottleneck and also the thing that made her pitch land, so automating it against a generic quality score would have produced a faster pipeline that ranked prospects wrongly. The criteria encoded are hers.
The rest was left alone. She did not want an AI dialer, because the personal call is why the pitch converts, and she did not get one. Her CRM, her project management and her sales process all still work the way they did.
Speed is king, and speed is money too. We're scraping the same platform, so the leads are the same quality, but the process is more efficient and faster, so I'm able to close deals faster. I could go through 20 leads where usually it would take me double the time.