Find local businesses that just told the internet they need you

A list of 20 is a start. A list of 20 where each one has a reason to say yes this month is a different game. These tactics take longer per lead, which is the point: most people won't bother, so the businesses you find this way rarely hear from anyone else.

1. Read the job ads, not the directorySearch Indeed and LinkedIn Jobs for marketing assistant, social media or marketing executive plus your niche and town, posted in the last 14 days. A dental practice advertising for a part-time marketer has admitted the problem and set aside a budget. You're cheaper than a salary and can start next week. The line: pitch the business through its allowed route (Ltd by email, otherwise phone or visit). Never apply to the job as a fake candidate to get a meeting.

2. Catch them in their first 90 daysNew businesses need everything at once. Two free, public sources: Companies House advanced search ↗ lets you filter by incorporation date, location and SIC code (for example 86230 dental practices or 43220 plumbing and heating). For anything selling food, the food hygiene ratings site ↗ lists places marked Awaiting inspection, which usually means newly opened. New company leads goes deeper. The line: Companies House shows directors' names and sometimes home addresses. Contact the business, never the person at home.

3. Let their own customers write your openerOn Google Maps, open a prospect's reviews and use the search reviews box for website, find, phone and prices. A 5-star review that says "great job, but hard to find your number" is the best first line you'll ever get, because the owner has already heard it from a customer. The line: quote it accurately and leave the reviewer's name out of anything you send.

4. Spot the dead website linkGoogle shut down its free Business Profile websites (the business.site addresses) in 2024. Some businesses never updated their profile, so the Website button on Maps still points to a dead page. Tap every Website button in your list. A dead link, a link to a Facebook page, or a free wixsite.com address all mean the same thing: nobody is looking after this.

5. Find the orphaned clientsSmall web agencies close and leave their clients with nobody to call. Note the "website by" credit in each prospect's footer, then look the agency up on Companies House. Dissolved or in liquidation means every site carrying that credit has lost its support. Google "website by [agency name]" to find the rest of them.

6. Check ads without costing them moneyThe Google Ads Transparency Center ↗ shows which businesses run ads and what the ads say, with no clicking. A business paying for ads that land on a slow or broken page is spending money right now, and you can show them where it leaks. The line: never click a prospect's or competitor's ads to check them. Repeated clicks drain their budget and count as click fraud.

7. Run 20 speed tests while you make a teaThis short Python script runs Google's own PageSpeed test on every site in your sheet, one at a time, and writes the mobile scores to speed.csv. Download speed_check.py and the sample targets.csv (same columns as step 3), or export your own sheet as targets.csv. It needs the Business and Website columns.

Get a free key firstWithout one, Google refuses almost straight away. Here's what a keyless run of the sample file returns:

Example Domain (test row)  |  error  |  HTTP Error 429: Too Many Requests
GOV.UK (test row)  |  error  |  HTTP Error 429: Too Many Requests

Open the PageSpeed Insights API get-started page ↗, click Get a Key, and pick or create a Google Cloud project. Copy the key, then set it in the same terminal you'll run the script from:

$env:PSI_KEY="paste-your-key"

export PSI_KEY=paste-your-key

Then run python speed_check.py targets.csv in the folder with both files. Here's the script, if you'd rather paste it:

import csv, json, os, sys, time, urllib.parse, urllib.request

API = "https://www.googleapis.com/pagespeedonline/v5/runPagespeed"
with open(sys.argv[1], newline="", encoding="utf-8") as f, \
     open("speed.csv", "w", newline="", encoding="utf-8") as out:
    w = csv.writer(out)
    w.writerow(["Business", "Website", "Mobile score", "Largest paint"])
    for row in csv.DictReader(f):
        url = row["Website"].strip()
        if not url:
            continue
        q = {"url": url, "strategy": "mobile"}
        if os.environ.get("PSI_KEY"):
            q["key"] = os.environ["PSI_KEY"]
        try:
            data = json.load(urllib.request.urlopen(API + "?" + urllib.parse.urlencode(q), timeout=120))
            lh = data["lighthouseResult"]
            score = round(lh["categories"]["performance"]["score"] * 100)
            lcp = lh["audits"]["largest-contentful-paint"]["displayValue"]
        except Exception as e:
            score, lcp = "error", str(e)[:60]
        w.writerow([row["Business"], url, score, lcp])
        print(row["Business"], score, lcp, sep="  |  ")
        time.sleep(5)

Then paste speed.csv and your "biggest problem" column into Claude or ChatGPT and ask for one plain-English opening line per business, using only the numbers given. Twenty tailored openers in five minutes instead of an evening.

8. Time it to their busy seasonPitch the month before demand jumps, when owners are already thinking about it: heating engineers in late summer before the boiler rush, roofers and gutter cleaners before the autumn storms, wedding venues before the January booking season. Avoid accountants in January.

What a perfect lead looks likeHere's an example row, with every column drawn from one of the signals above:

Business: Example Dental Care (Ltd)
Website: Maps button points to a dead business.site page
Google reviews: 4.9 from 212
On page 1?: No (page 3 for "dentist [town]")
Running ads?: No
Biggest problem: Advertising for a part-time marketing assistant (Indeed, 6 days ago). Review: "Brilliant dentist, shame the website never loads". Mobile speed score 23 on the old site
Contact by: Email
Status: Pitch this week

Three signals, one row. That's a business that has already admitted the problem, set aside money for it, and heard about it from a customer.