How a product becomes a choice
My first growth article began with 10,000 people visiting a financial app’s website. I followed what happened as they enrolled, used the product, and decided whether to return.
But how did the product become one of their options in the first place?
Imagine someone sees a creator describe an account without monthly fees. They are content with their current account, so they keep scrolling. A month later, their bank charges them $12. Now they remember the video and start looking at alternatives.
At the time of the video, they did not visit a website or begin an application. But the creator gave them a name to recall when the fee made their current account worth reconsidering.
If we begin the story with the website visit, we begin too late.
Coming to mind
The video connected an account to a problem they did not have yet: if I ever get tired of paying fees, this account might be worth a look.
That is different from getting someone to apply on the spot. Most people who see the video will not need a new account that day. Some may never need one. When the $12 fee gives the user a reason to look, the account is already familiar. If they eventually find it through search, that click tells us how they arrived, not why they recognized the name.
The same gap can appear at a much larger scale. After sharply reducing marketing spending during the pandemic, Airbnb reported that more than 90% of its traffic in the final quarter of 2020 was direct or unpaid. That tells us how people arrived in that quarter. It does not tell us when Airbnb first became an option they knew about, or prove that its earlier marketing was unnecessary.
The account is now on the user’s list. Next, they have to decide whether it fits.
Working out what to choose
They ask an AI tool about accounts without monthly fees. It suggests a few, giving them a shortlist before they visit a company’s website. They start checking the details. Does avoiding the fee require direct deposit? How would they access cash? What would moving their paycheck involve?
The answer may draw on official pages, comparisons, and reviews. They can rule an account in or out without opening its website. Their decision moves forward, but the company may see no visit that tells it what they learned.
They may still ask a friend or check an official site. They are not moving neatly from one marketing channel to the next. They are deciding whether changing accounts is worth the trouble.
Eventually, they search for the account by name and open its website.
Following through
The user reaches the landing page with a claim they want to verify: an account without monthly fees. They have heard the creator’s claim and seen other options in an AI answer. The page should state plainly whether avoiding the fee has conditions, how the account works, and what enrollment involves. It also needs to load quickly enough for them to get those answers.
Suppose a redesign makes the “Start” button more prominent. More people click it, but fewer finish opening an account. The page may have drawn them into enrollment before they understood the terms. The reverse can happen too: fewer starts, but more completions because the page answered their questions first. I would judge the page by completed enrollments, not the button click alone.
The landing page should carry forward the decision they have already begun, not ask them to start over.
What gets credit
Once the user finishes enrolling, the team sees a completed signup in its report. The report credits search because it was the last source recorded before enrollment. This is last-touch attribution.
That tells the team how the user arrived. It does not tell them why they chose the account, and it may miss the creator’s video, a conversation with a friend, or what they learned from an AI answer.
A team might look at the report and move money from creators to search. Perhaps that is right. Or perhaps many people search for the product because they heard about it earlier. The report alone cannot tell the team which is true.
One signup. Many possible paths.
Change the path. See what the report credits.
Reason to look
$12 feefrom their current bankThe report credits Search. Earlier moments stay out of view.
Illustrative paths. The final arrival is assumed to be recorded; this does not establish cause.
The report can miss an earlier influence. It can also give too much credit to a click it did record. In a 2015 study of eBay’s paid-search experiments, ads shown when people searched for eBay by name produced no measurable short-term sales benefit. Almost all the traffic that stopped coming through those ads arrived through unpaid search instead. Ads on other search terms did help bring in new and infrequent customers. A sale that follows an ad click is not necessarily a sale the ad caused.
One response is to split credit among the steps a team can observe. That captures more than the final click, but it still cannot show what would have happened without a particular ad or creator campaign. Withholding a campaign from a comparable group can help estimate that contribution. Delayed decisions and creator content make those tests harder to run cleanly.
The user’s path does not prove the creator caused their signup. But it shows how much can happen before the final click: the account came to mind, held up against alternatives, and offered a clear next step. Search recorded their arrival, not the whole decision.
To improve acquisition, a team needs to learn which efforts help at each of those moments, and which bring in customers who would not have come otherwise. Enrollment is where acquisition hands off to the product. The next question is whether the product delivers what brought the user in.
