Which ad actually brings customers? How small businesses track their leads

A tradesperson spending SEK 5,000 a month on Google ads and the same on Facebook, without knowing which channel brings the jobs, is effectively paying double for half. The question “how did you find us?” rarely yields better answers than “Google”, and that answer does not distinguish between the ad that cost SEK 40 per click and the organic search result that was free. Lead attribution is about replacing that guess with data.

Three traces a customer leaves behind

A prospective customer reaches you in three ways: clicks and reads, submits a form, or calls. The first two are easy to track. The foundation is UTM parameters, small additions to the ad link like ?utm_source=facebook&utm_medium=cpc&utm_campaign=springcampaign. The analytics tool reads them and shows which campaign the visitor came from. Google recommends always setting source, medium and campaign, and warns of a classic trap: the values are case-sensitive, so “Facebook” and “facebook” become two different sources in the report. Therefore decide a naming standard, write it down and stick to it.

Calls are the hard trace. The customer sees your number on the site, calls, and the attribution chain breaks. That is exactly the problem call tracking solves, more on that shortly.

One analytics tool and a little UTM discipline goes a long way

Google Analytics 4 is free and shows campaign data directly in the Traffic acquisition report. But free has downsides. In 2023 IMY decided, following noyb complaints, that four Swedish companies had used Google Analytics unlawfully; Tele2 was fined SEK 12 million and three companies were ordered to stop. The decisions concerned the old version, but the fundamental problem remains: GA4 data is processed on Google’s global server park and legality rests on a data transfer agreement that noyb has already announced it will challenge. In addition, the free version limits data retention in explorations to a maximum of 14 months, making year-on-year comparisons shaky.

If you want away from that there are two EU alternatives. Matomo runs either self-hosted (free, data on your own server) or as a cloud service from EUR 29 per month with storage in Frankfurt. The catch hits self-hosters. The finer multi-channel attribution is included in the cloud plan but not in the free version, where it requires the paid Business package at EUR 1,450 per month. And there is no Swedish exemption from the consent banner; that only applies to certain countries such as France. Plausible is simpler and cheaper, from USD 9 per month, entirely without cookies and banners, with all data at European-owned providers. Phone clicks and form submissions are tracked as own goals with a CSS class on the element. The limitations Plausible states itself. There is no deep attribution modelling, the tool measures nothing backwards in time and goals only start counting once created.

Calls need their own tool

If you get leads mainly by phone none of the above helps. Then dynamic number insertion is needed. WhatConverts reads each visitor’s traffic source and swaps the number on the site for a tracking number from a pool. When the visitor calls, the call is tied to ad, campaign and ideally keyword. The tool can then send qualified leads back as conversions to Google Ads, so Google’s bidding is trained on real customers instead of wrong numbers. For an ad-driven business this is the difference between guessing and knowing. Read our detailed WhatConverts review for prices, GDPR status and when the tool fits.

The product is excellent, 4.9 out of 5 on both G2 and Capterra. Yet we give it 3.1 in our review, and the reasons belong here too. All data is stored in the USA on AWS under US law, and the privacy policy names no transfer mechanism for EU data, even though the pricing page markets GDPR compliance in all plans. Prices are also in USD. The entry-level plan at USD 30 (approx SEK 290) only tracks calls and SMS; form tracking requires Plus at USD 60 (approx SEK 580). On top of that come usage fees, and Swedish numbers are more expensive than American ones, about USD 3 per number and 6 cents per minute. The Sweden page says “up to 100 calls” within the plan’s credit, while the USA page says 148. The monthly cost is therefore hard to predict exactly.

The recommendation is therefore split. If you advertise for tens of thousands of SEK per month, live on incoming calls and handle no sensitive personal data, the tool pays for itself quickly. Try WhatConverts free for 14 days (affiliate link, we may receive commission), but read our review with reservations first. If you run a healthcare clinic, law firm or anything else where the calls themselves are sensitive data you should abstain. Recorded patient calls on American servers without a named transfer mechanism are the wrong place to be if the Schrems III question is decided badly.

Or manage without tools

A moment of self-reflection before you pay for anything. If you advertise in a single channel no attribution is needed; everything new comes from there. If the budget is under a couple of thousand per month the tool cost eats up the profit of knowing. And the manual trick still works: a separate phone number in the Google ad, another on Facebook and a column in the booking calendar. Blunt, but free and often fully sufficient until the ad budget grows past the point where every misplaced thousand really hurts. Only then is it time for UTM discipline and an EU-friendly analytics tool. Call tracking comes last, and only for the right kind of business.