Projects Data · Sales
Website auditor for winning clients
504 local business websites audited by a script and ranked by sales opportunity.
- Role
- Creator
- Year
- 2026
- Tools
- Python · requests · BeautifulSoup · openpyxl · Excel
- Status
- Finished
How to use it
Three made-up example websites go through the audit live, check by check, and land in their place within the real distribution of the 504 audited. Below, the most frequent technical failures. The names and phone numbers of the audited businesses are not published.
The best client for a new website isn’t the business with the worst website. It’s the one with hundreds of five-star reviews and a website that is losing it the customers it has already won.
- 504
- Madrid business websites audited
- 119
- Different business categories
- 50+
- Technical and aesthetic signals checked on each website
- 189
- Websites without a security certificate (“Not secure”)
The question
Which businesses should you call first to sell them a new website? Cold calling costs hours and wears you down. If you have to make a hundred calls, better that they are the hundred most likely to end in a meeting.
The hypothesis
A bad website isn’t enough. A business with a bad website and no customers has neither the budget nor the reason to invest. The ideal candidate combines three things: a website that makes a bad impression, a business that works (many reviews and a good rating) and a sector where the website matters in the buying decision. People choose a tax adviser for the trust their website conveys; a neighbourhood hairdresser, much less so.
How it scores
A Python script visits each website, downloads it and analyses it with BeautifulSoup. The final score adds up five blocks:
- Technical. More than twenty checks, each with its own weight: security certificate (20 points if missing), site down (25), legal notices and cookie banner, loading speed, basic SEO tags, mobile friendliness, outdated jQuery, a copyright from years ago, or whether the “website” is really a directory listing.
- Aesthetic. Another thirty signs of an outdated website, read straight from the code: table layouts,
<font>or<marquee>tags, Bootstrap 2, BMP or GIF images, no web fonts, no favicon, fixed width or Comic Sans. - Reputation. Number of reviews (3 to 20 points) and average rating (up to 8 more).
- Sector. How much the website matters when selling that kind of business, from 10 to 25 points.
- Urgency. Problems the owner can see today, such as the browser’s “Not secure” warning.
The higher the score, the worse the website and the better the candidate. For each business, the script also estimates a budget range according to how much work its website needs.
What came out
- 504 websites in 119 categories, sorted into a full ranking and a top 50 with the main argument for each call.
- The most common technical failure: 189 websites without a security certificate. The browser shows “Not secure” next to their name, in front of every customer.
- The most common aesthetic failure: almost 200 websites without modern image formats, which makes them slow on mobile.
- Among the largest categories, those with the highest average opportunity were administrative agencies, nutritionists, tax advisers and immigration lawyers: sectors where clients choose on trust and the website is the first thing they see.
It turns a list of phone numbers into a prioritised call queue with a script for each call, which is the part of selling that takes the most time.
What I take away
A little code turns a blind sales task into a list ranked by probability of success. The key isn’t the script but the business hypothesis behind each weight.