Find First Hundred Users: Ask One Question, Read the Missing Source

You open the dashboard and see a row of direct visits. The user says a friend sent the link, and the dashboard misses the friend. For the first hundred, that friend is the channel.

These paths are often a direct message, a group chat, a referral, a community channel, or an AI answer copied into a browser. They leave little tracking. The answer is outside the analytics.

The dashboard is missing the source that matters

Attribution tools record what they can tag. They miss private channels, direct messages, and recommendations that happen before the click. In one completed-order index, 51.8 percent of purchases had no usable tracked source. Among the untracked customers, 91.7 percent still named a discovery channel the dashboard did not record.

Word of mouth is the common answer. In post-purchase survey data, word of mouth was the top stated discovery channel, making up 19.88 percent of responses year to date. It affected at least 25 percent of purchases in every vertical tracked. In a September 2026 index, word-of-mouth purchases were untracked 58.7 percent of the time and YouTube purchases were untracked 60.2 percent of the time.

AI referral is newer, but it deserves the same check. The share of completed-order shoppers naming an AI assistant rose from 0.23 percent in December 2025 to 1.10 percent in August 2026, a 4.8-fold increase. It is small. It is moving.

Direct traffic is the label for a missing tag. It often covers the referral, the direct message, or the community thread that actually brought the user in. Named answers fill the missing source field.

One question builds the first-user channel list

• Put one question on the confirmation screen. Use a single free-text field: 'How did you find us?' Keep it short. Done looks like every new user leaves a one-line answer before the confirmation screen closes. Do not turn it into a survey. If you add too many fields, the first hundred will not finish the form, and the source list becomes unclear.
• Tag the answer as a first-user channel. Use plain labels: friend, group chat, forum, video, AI referral, newsletter, direct. If the answer is vague, mark it as direct and keep the raw text. The output is a table with the tracked source and the named source.
• Compare the named channel to the tracked source. Line up the sources. If the tracked source says direct but the named source says a friend, the difference is your real channel. If the tracked source says a campaign but the named source says a community channel, the campaign may be the click, not the reason. You finish with a short list of differences, ranked by how often they appear.
• Spend more on the untracked channel with the highest named share. Pick the channel that appears most often in the named answers, not the channel with the best dashboard number. Then act on that channel: make word of mouth easier to forward, answer daily in the community channel, or write the answer you want the assistant to give. The result is a repeatable process for the next hundred.

The manual work is the distribution

For the first hundred, no distribution is the starting condition. Send the link manually. Watch the first use. Ask what confused them. Then ask how they found you. A confirmation-screen question records that manual work as a source list.

Do not wait for scale. First hundred users test whether one person will tell another. If the answer is yes, the source list tells you where to be when the next hundred arrive.

Spend the next round on that channel, with the same manual care, until the next users arrive.