Funnel analysis maps the steps people take toward a goal, such as visiting, adding to cart and paying, and counts how many reach each step. The gaps between steps show where customers drop off. You then investigate the biggest drop, form a reason, test a fix and measure whether the step improves.
- A funnel is an ordered list of steps toward one goal, measured step by step.
- Fix the biggest and most fixable drop-off first, not the most visible page.
- Segment by device, source and new versus returning users to find hidden patterns.
- Funnels show where people leave; session views and interviews show why.
What is a funnel in analytics?
A funnel is the sequence of steps a person goes through to reach a goal. For an online store: view a product, add to cart, start checkout, enter address, pay. For a service business: land on the site, view services, open the enquiry form, submit it. The funnel counts how many people make it from each step to the next.
It is called a funnel because numbers narrow as you move down. Some narrowing is natural, since not everyone who views a product will buy it. The question is whether the drop at any step is larger than it should be, or larger than before.
How do you build a funnel?
Start with one goal and list the steps in order. Make sure each step is something you can actually record, such as a page view or an event. Define the time window within which steps must be completed, since some purchases take minutes and others days, and decide whether people must follow the exact order.
Check the data before trusting it. If the number of people at step three exceeds step two, tracking is probably broken or steps are not required in order. Compare the final step with real orders or enquiries in your database. A funnel built on faulty tracking leads you to fix problems that do not exist.
- Choose a single goal, such as a paid order or a submitted enquiry
- List the steps and how each is recorded
- Set a time window for completion
- Validate the end of the funnel against real records
- Use the same definitions each time you compare periods
How do you read the drop-offs?
Look at the conversion from each step to the next, not only the overall number. A small improvement at an early step with many visitors can be worth more than a big improvement at a late step with few. But late drop-offs, like payment, are often the most painful because those users showed strong intent.
Compare against your own history before comparing with others. A sudden fall after a release points to a bug or a change in design. A gradual decline may reflect lower quality traffic. Without a baseline you cannot tell whether a number is good or bad.
Why should you segment the funnel?
An overall funnel can hide big differences. Mobile users may drop at the payment step because the form is awkward, while desktop users do not. Visitors from one ad campaign may never get past the first page while organic visitors convert well. New customers behave differently from returning ones.
Segment by device, traffic source, location, new or returning status and plan. In India, differences in network speed and payment preferences such as UPI versus card can show up clearly in the segments. The pattern guides you to the fix: speed, form design, trust signals or targeting.
- Device type: mobile, tablet, desktop
- Traffic source: search, social, ads, direct, WhatsApp
- New versus returning users
- Payment method used or attempted
- City or region, where network speed varies
What do you do after finding a drop-off?
Form a hypothesis about why. Watch session recordings, check error logs, test the flow yourself on a mid-range phone, read support chats and ask a few customers. Typical causes include surprise costs, required account creation, slow pages, confusing forms, missing payment options and failed validation.
Then change one thing, measure the step again, and keep the change only if it helps. If traffic is low, run the test for longer or rely on direct user testing. Record what you tried so the team does not repeat failed ideas.
What are the limits of funnel analysis?
Funnels assume a fairly linear path, but real people wander, return later and switch devices. If tracking cannot join sessions across devices, some users appear to drop off when they have simply continued on a laptop. Remember also that people who never started the funnel are not counted, so it says nothing about demand you never reached.
Use funnels as one tool alongside retention analysis and qualitative research. A Plus Solution sets up product analytics, funnels and dashboards, and in practice the best results come from pairing them with regular improvement cycles, not from a single report.
Frequently asked questions
How many steps should a funnel have?
Enough to find the problem but few enough to read at a glance, often between three and seven. Split a long journey into separate funnels.
What is a good conversion rate for a funnel step?
It depends on the product, traffic quality and step. Compare with your own history and your own segments first.
Can we build a funnel without special tools?
Yes, if your database or analytics records the events. A simple count of users reaching each step in a spreadsheet works for a first pass.
How often should we review funnels?
Weekly for key funnels, and after any release that touches the flow.
Need help with this? Ask us a question about it — we reply within one working day.
