How do I track where my leads are actually coming from?

Why most tracking misleads

The lead generation essay in the newer book describes what the previous era taught. Dashboards showed impressions, clicks, open rates and lead scores, and founders could finally point to numbers and say they were making progress. It calls attribution equal to progress one of the illusions that fooled boutique founders: activity replaced outcomes, confidence replaced competence, and the numbers created a story that was not true.

It also explains why service firms are hard to track. Product buyers respond to reminders of needs they already have; service firms usually have to create the need by educating a prospect about a problem they did not know they had. As an inference, that means a buyer often meets the firm several times, through a referral, an article, a talk, a conversation, before they ever fill in a form. Software that credits the last click will usually credit the wrong thing.

Record the source when the opportunity is created

The referral results answer on this site recommends that every opportunity carry the person who made the introduction, the category of source and the date, entered when the opportunity is created rather than reconstructed at year end. As an inference, apply the same rule to every channel. Make source a required field in your customer relationship management system, with a short, fixed list of categories: referral from a client, referral from a partner, outbound, event, content, inbound search, existing client expansion, and other. Add a free-text note for the specific person, event or piece of content.

Ask the buyer, and listen to the answer

As an inference, the most accurate source of attribution is the buyer. Ask on every first call, in a natural way: how did you come across us, and what made you decide to reach out now? The first half tells you the source. The second half tells you the trigger, which the account executive essay says is what makes an opportunity real.

The account executive essay says buyer conversations now happen on recorded platforms, so the voice of the buyer can be captured, transcribed and analyzed. As an inference, let AI pull the answer to that question from each first-call recording and suggest the source category, and have a person confirm it. A buyer will often name more than one touch, such as a colleague who mentioned you and an article they read afterward. Record both.

Do not rely on sellers to fill it in later

The account executive essay describes how records failed in the previous era: critical information lived in conversations, not systems, and CRM fields were updated selectively, often days later, and frequently reflected optimism rather than evidence. It says the burden of record keeping made the seller role heavier without making selling better. As an inference, source tracking fails the same way when it depends on someone remembering to update a field after a busy week. Capture it at the moment the opportunity is created, and let the recording supply the evidence.

Keep the chain, not just the last link

The referral essay says some of the most powerful referral sources never buy; they introduce people who introduce people. It describes second-order mapping to find connectors and influence hubs, the referrers who unlock entire networks. The referral results answer adds that if you credit only the final introducer, the connector who started the chain looks unproductive.

As an inference, record a first source, the original reason the buyer knew your name, and a converting source, the touch that led to the conversation. Over a year, the pattern of first sources usually shows where your reputation really comes from, while converting sources show what turns reputation into meetings.

Judge sources by won work, not volume

The marketing ROI answer on this site recommends counting conversations with buyers who fit, tracing won work back to its first source, and comparing channels on the contribution margin of the clients they bring over time. The referral essay lists the measures for scoring sources: fit with the ideal client profile, strength of expressed need, close rate, deal size, sales cycle and downstream profitability. The marketing essay says boutiques need a few of the right clients, not thousands.

As an inference, review sources every quarter with three numbers for each: qualified conversations, won work, and the margin of the clients won. A source that produces many leads and few good clients is noise. A source that produces one client a year worth keeping may be your best channel.

Track existing clients separately

The account management essay says expansion revenue from existing clients is one of the fastest ways for a boutique to scale, yet most firms build their growth plans on new logo acquisition. As an inference, record work from existing clients as its own source category, and note which engagement or person it grew from. Otherwise expansion is counted as if it came from marketing, and the account management work that produced it stays invisible.

Let the system learn

The lead generation essay describes the memory step of its framework: AI remembers every signal and outcome, including which micro-segments convert, which channels outperform, which triggers predict high-fit leads and which value propositions win. As an inference, once sources are recorded consistently, that pattern becomes visible without a spreadsheet, and it becomes more useful each quarter.

Use win-loss reviews to fill the gaps

The service offering chapter of the 2020 book recommends a quarterly win-loss program in which an objective third party calls recent prospects and asks why they chose you or did not. As an inference, add a question about how they first heard of the firm and what else they considered. Buyers are often more candid with a third party than with the seller, and the answers correct what your records assumed.

What we do not prescribe

Collective 54 names no tracking software and publishes no attribution model or list of required source categories. The published positions are attribution mistaken for progress, activity replacing outcomes, service firms creating demand through education, recording referral sources when the opportunity is created, crediting connectors in referral chains, the referral scoring measures, a few right clients over many, recorded buyer conversations, AI memory of which channels and triggers convert, and win-loss programs.

When this answer flips

If most of your leads come from a handful of people, as an inference, a simple list of who introduced each client may tell you more than any system.

If you run a high-volume, lower-priced service bought online, standard digital attribution becomes more meaningful.

And if your records are empty today, start with the last twelve months of won clients and ask each one how they found you; it is the fastest way to a useful baseline.

The short answer

Make source a required field when each opportunity is created, with a short fixed list of categories and a note of the specific person or event. Ask every buyer on the first call how they found you and why they reached out now, and let AI pull the answer from the recording. Record the first source as well as the converting one, because the referral essay says connectors who start a chain otherwise look unproductive. Review sources each quarter on qualified conversations, won work and client margin, not clicks, which the lead generation essay calls an illusion of progress.