We read 251 B2B websites across Latin America. Nine out of ten won’t let you book a call with a salesperson.
What each site loads, how it captures leads and what happens when a buyer arrives with a hot problem. Measured by machine, with the method in the open.
251 sites read · 400 attempted · October 16, 2026 · built with Switchy’s Mirror engine
What we found
A B2B buyer in LATAM who lands on a site today ready to buy almost always finds no way to book time with anyone: there’s a form, a WhatsApp button or nothing. It isn’t a design problem: the sales process isn’t connected to the website.
has no way to book time with a salesperson
229 of 251 sites
of the sites that say “request a demo” have nowhere to book it
33 of 42
of the sites running paid ads have nowhere to book
34 of 40
shows no form, no booking link, no chat and no WhatsApp on its homepage
106 of 251
The six findings
- 34 of 40
They pay for ads and don’t let you book
40 sites load ad pixels (mostly Meta). 6 have a calendar. The other 34 buy the buyer’s attention and don’t offer a way to book: 22 of them don’t even show a form in what we read.
- 31 of 34
They installed HubSpot and added no calendar
34 sites load HubSpot. Only 3 let you book. The CRM is installed; the front door isn’t.
- 92 sites
They measure visits, not conversations
92 load analytics with no detectable CRM. As far as we can see, they know how many people came in, not who they talked to.
- 3 to 1
WhatsApp beats the calendar
30% offer WhatsApp and 8.8% let you book (76 vs 22). The channel exists; the process doesn’t. 38 sites have WhatsApp with no form and no calendar: a conversation that depends on someone looking at their phone.
- 67 of 83
They publish the price and don’t catch the buyer who already decided
83 sites show a price. Only 16 of them let you book. Whoever reached the pricing page has already walked most of the funnel alone, and in most cases has no one to talk to.
- 2 of 251
Only two have the full machine
Two sites out of 251 load a CRM, have a booking link and run ads at the same time. The rest have loose pieces. It’s not a criticism: it’s the measure of how much room there is.
The cross-cuts
| Sites read | 251 of 400 attempted |
|---|---|
| With a booking calendar | 22 (8.8%) |
| Ask for a demo | 42 · nowhere to book it: 33 |
| Load ad pixels | 40 · no booking link: 34 |
| Load HubSpot | 34 · with a booking link: 3 |
| Analytics with no detectable CRM | 92 |
| Public price | 83 · with a booking link: 16 |
| WhatsApp with no form and no calendar | 38 |
| CRM + booking + ads at once | 2 |
How it was done, and what we couldn’t see
We publish the method so anyone can repeat it. If you repeat it and get something different, we want to know.
- The sample. 400 domains from our own database of Latin American B2B companies, filtered by type of business and fit with our customer profile. 251 were read; the other 149 didn’t answer in time or blocked automated reading. We don’t count them as anything.
- What is read. The homepage and, if it was linked, the pricing page. From there come the signals: the scripts the page loads (CRM, analytics, ads, chat, booking), whether there’s a form and whether it asks for an email, whether there’s a WhatsApp link and whether a public price appears.
- Deterministic signals, no model. No interpretation and no natural language: each signal is the presence or absence of a specific marker in the page’s code. The same scanner on the same site gives the same result.
- What we couldn’t see biases the numbers down. Tools loaded inside a tag manager are invisible to this read. The real CRMs, pixels and chats are more than we report, not fewer. Our 13.5% with HubSpot is a floor, not a ceiling.
- Three clicks deep doesn’t count. A site may have a calendar hidden on a subpage. For the buyer who arrives today and wants to talk today, it may as well not exist. That is the study’s yardstick, and it is deliberate.
- What it doesn’t measure. How good the product is, how fast they reply by email or what happens after the form. That’s another study.
Five fixes for Monday
In order of effort. The first three cost nothing and need nobody from tech.
- 01
Put a calendar where you already ask for the demo
If your button says “request a demo” and leads to a form, you’re asking someone with intent to wait. 79% of the sites that ask for a demo have nowhere to book it.
How: check whether your CRM includes a booking link; if not, there are free options like Calendly or Cal.com. Paste the link into the same button.
- 02
Give WhatsApp an owner and office hours
38 sites have WhatsApp as their only channel. A number with no assigned owner and no published hours is a conversation that goes cold just like a form.
How: define who answers in each time slot, publish the hours next to the button and measure how long it takes.
- 03
Set a one-hour alarm on the form
Whoever leaves their details today expects an answer today.
How: a rule in your inbox that flags form messages, and an alarm on your phone. No new software.
- 04
Connect the form to where the pipeline lives
92 sites measure visits with no detectable CRM. If the form lands in an inbox and then in someone’s memory, the lead exists only as long as that person remembers it.
How: make every submission create a record in the same place as the rest of the pipeline, with an owner and a date.
- 05
If you publish a price, catch the buyer who already decided
83 sites show a price and only 16 let you book.
How: next to each plan, a button to talk to someone today. Not a form: a calendar.
And what does your site say?
The same engine that read the 251 sites can read yours in about 20 seconds and tell you what you sell, where the sale leaks and what the lever is. Free, no signup.
Read my site →And if your problem is the operation behind the site: our guide “From experimentation to an operation that sells” (in Spanish) →Who made this. We made this at SwitchON, which builds this scanner and sells a related product (Switchy, a sales copilot). Read it with that in mind. If you repeat the study and get something different, write to Nicolás: nicolas@switchon.dev.