Analytics for builders

Build a two step funnel from traffic you already recorded

By Emily Parker · Nov 4, 2026 · 9 min read

You have two pages and a suspicion. Maybe people read your pricing page and never open the signup page. Maybe they open the signup page and leave. Two page view totals cannot settle it, and starting a new experiment and waiting four weeks is a slow way to find out something your analytics may already hold.

This post walks through one funnel, built on the history you have, with a worked example, the choices that change the answer, and the limits you should know before acting on it.

A crowd of small figures gathered at a first doorway, with only a few continuing through a second doorway further down a hallway.

What a funnel adds to two page view counts

Say /pricing has 1,240 page views this month and /signup has 410. That does not tell you how many people went from one to the other. The 410 could include people who never saw pricing. The 1,240 could be 300 people who came back four times each.

A funnel counts people, not page views. Peeko follows a person rather than a visit: a second hash, built from the address and browser string but without the date, strings one person's page views into a journey. The address itself is never stored, and the hash that powers the People number is re-salted every day so nobody can be followed from one day to the next. The identity a funnel uses is the one without the date, and it is also what goals and revenue use.

So a two step funnel answers one question: of the people who did step one, how many did step two inside the window? Everything else on this page is about asking that question carefully.

The second step is also where the value tends to sit. Ahrefs published a measurement from its own site in Jun 2025: AI search was 0.5% of its traffic over 30 days but 12.1% of its signups. That is one site, one 30 day window, and Ahrefs' own web analytics, so treat it as an example of the pattern and not as a benchmark. A page view count would have called that traffic negligible. Only a count of what people did next shows otherwise.

0%3.75%7.5%11.25%15%Share of trafficShare of signups

Source: Ahrefs, Does AI Search Traffic Convert Better Than Traditional Search? For Ahrefs, Yes: 0.5% of Visitors Drove 12.1% of Signups, Jun 16, 2025. Retrieved Nov 1, 2026.

Pick the pair of pages

Choose two pages where the second is the thing you want people to do after the first. Good pairs for a small product site:

  • /pricing then /signup
  • A blog post then /pricing
  • /signup then the page people land on after they sign up
  • A docs page then your install page

Three checks before you define anything:

  1. Both pages must have existed for the whole period you are reading. If you launched /signup on the 20th, a funnel across the whole month will show a step two of zero for the first 19 days and a false drop-off.
  2. The tag must have been on both pages. If you added the script tag to the signup page last week, earlier signup visits were never recorded. If a page seems to be missing entirely, check the first page view causes.
  3. Spell the path the way it was recorded. A trailing slash, a renamed URL or a localized prefix can split one page into two paths. Our earlier post on following an assistant visit to a signup describes a step as a page path or an existing goal, and says paths can use wildcards such as /docs/*, which is the way to cover a section that moved. Read the path list in your dashboard once and copy from it.

If step two is an action and not a page, such as a click on a checkout button, define it as a goal first. A goal is a definition and not a counter, so one written today counts over months before it was written. The post on a first goal with almost no conversions covers that setup.

Pick the window

The window is how long a person has to get from step one to step two. Peeko offers four: 1 hour, 1 day, 7 days and 30 days. A funnel can cover at most 90 days of history, and a site can hold ten funnels.

Match the window to how people actually decide:

What the second step isWindow to start with
A free tool with a signup form1 hour or 1 day
A paid product people compare first7 days
A purchase that needs a budget or a second person30 days

If you are unsure, build two funnels with the same pages and different windows. The gap between them is information, as the next section shows.

Why this works on traffic you already recorded

Three facts from how Peeko stores data make a same-day read possible:

  • Page views are kept for three years by default, and a site owner can set that anywhere from 30 days to ten years.
  • A funnel covers up to 90 days.
  • Goals are definitions applied to recorded events, not counters that start when you create them.

Put together, a funnel defined today is read against the page views stored over the last 90 days at most, not against traffic that arrives after you click save. The limit is the day your tag went on. If the script has been on your pages for 11 days, you have 11 days of data, and that is the honest size of your sample.

A worked example with illustrative numbers

These numbers are made up to show the arithmetic. They are not a Peeko measurement.

You define /pricing then /signup over the last 60 days.

WindowPeople at /pricingPeople who then reached /signupRate
1 day1,24013110.6%
7 days1,24018615.0%

Two readings come out of this.

The overall drop-off. With a 7 day window, 1,054 of 1,240 people (85.0%) did not reach the signup page. That is the number to look at first. It says the loss happens between those two pages, not on the signup form and not somewhere later.

The gap between windows. 55 people reached /signup between day 2 and day 7. That is 30% of everyone who got there (55 of 186). Many of your visitors come back later to decide. If you change the pricing page and judge the result after one day, you will be reading the wrong window. For a considered purchase, judge on 7 days.

Plausible's funnel documentation gives a similar rule for multi-step funnels: focus on the step with the largest proportional drop, not the lowest absolute number. With only two steps there is one drop to look at, so the work moves to the next section: finding out why.

How much to trust the number

A rate on a small group moves a lot. For 1,240 people at 15.0%, the normal approximation puts the 95% range at about 2 points either way, so 13% to 17%. For 20 people with 5 reaching step two, the rate is 25% and the same approximation gives a range of about 19 points either way, which is 6% to 44%. That is too wide to act on.

A practical rule: if step one has fewer than a few hundred people, widen the date range, up to the 90 day maximum, before you read anything into the percentage. The post on a first goal with almost no conversions goes through what to do when even 90 days is thin.

What the funnel cannot see

Know these before you act on a drop-off.

  • Different browsers are different people. A visitor is a hash of the address and browser string. Someone who reads pricing on a phone and signs up on a laptop counts as one person who dropped off and another who arrived at step two from nowhere. Funnels undercount in that case.
  • Your own visits count as visitors. If you open /pricing forty times while editing it and never open /signup, your step one is inflated. Keep your own visits out of site analytics before you trust a small-site funnel.
  • Step two reached by another route still counts. A person who bookmarked /signup and came back directly is a completion even if your pricing page did nothing. The funnel tells you where the loss is, not what caused it.
  • A funnel does not split by source on its own. To approximate one source, start the funnel on a page that source mostly sends people to. The post on following an AI referral to a signup without cookies shows how.

From a number to a change

Once you know the loss is between A and B, open B the way a visitor would, on a phone and then on a laptop. Check four things:

  1. Is the link or button to B visible without scrolling on a small screen?
  2. Does B load in a second or two, or does it hang?
  3. Does B ask for more than the visitor expected, such as a card number on a free trial? Peeko's own trial asks for none and starts when you create your account, which is the kind of promise worth stating on the page that links to signup.
  4. Does the text on A promise something B does not deliver?

Pick one change. Keep the same pages and the same window so the next reading is comparable. Then read the funnel again after a full window has passed: 7 days for a 7 day funnel, not 2.

When the first funnel is steady, add a third step, up to five, or connect the last step to revenue. The post on reporting revenue from your site or letting Stripe confirm it explains the two ways a sale reaches your numbers.

The short version of the method

  1. Choose two pages where B is the action you want after A.
  2. Confirm both existed and carried the tag for the whole period.
  3. Pick a window that matches how long your buyers take: 1 hour, 1 day, 7 days or 30 days.
  4. Read the overall rate and the gap between a short and a long window.
  5. Check the size of step one before you trust the percentage.
  6. Change one thing and read the same funnel again after a full window.

The point is speed. The page views are already there, and the first answer is a few minutes away.