Video Workshop: Using Analytics to Improve Form Conversion

Which metrics to analyse to pinpoint user friction on forms

‍

Want to know how to identify the issues that are causing people to abandon your online form or checkout? This short video workshop takes you through a few simple techniques and analyses that help you to quickly diagnose user problems on your forms and develop hypotheses for improvement.

We first cover basic metrics as a starting point:

• Abandon volume & rate
• Error message triggers
• Corrections / returns
• Time spent

And then run through more advanced methods that let you uncover issues that can't be seen from the headline stats alone:

• Behaviour difference - Abandoners Vs Completers
• Failed submissions - what happens next?
• Field level segmentation of abandonment rates

‍

‍

‍

If you'd like to learn more, read our in-depth white paper on using data to optimize forms.

‍

Using analytics to improve form conversion rates (Video workshop transcript)

What we’re covering today

Today we’re talking about using analytics to improve your form conversion rates — specifically, which metrics you should be looking at to identify user friction.

Before we get into the data, I always like to start with a quick question:

Put your hand up if you enjoy filling in forms.

I’m going to raise my hand… but I’m assuming (because of the delay) I won’t get many hands up. Usually when I ask this in a crowd, there’s maybe one person in a hundred. Most people don’t like filling in forms — they fill them in to get what the form promises them. That’s a key thing to remember when you’re designing forms and asking people to give you information.

People dislike forms largely because they’ve had bad experiences — like:

• a huge, overly long form with a CAPTCHA at the bottom to frustrate them even more
• a massive dropdown where you have to scroll past endless options (who has time to pick between “Reverend Mother” and “Wing Commander”?)

We can laugh at these examples and assume “of course people drop out”… but do we truly know the scale of the problem? And do we know where it’s happening?

That’s where analytics comes in.

We’re going to split this into two parts:

• Base metrics: the building blocks and best starting point
• Advanced metrics: the ones that help you pinpoint the true causes

Why listen to me?

I’m Alun, Managing Director of Zuko. Zuko is a form analytics platform and we’ve been doing this for around 10 years. We provide data and help teams analyse their forms to identify issues and fix them.

Over that time, we’ve seen pretty much every form problem there is — and we’ve quantified them. So today I’m sharing some of the “wisdom” we’ve built up through that experience.

Base metrics

Abandonment

The first metric you should look at is the most obvious: abandonment.

By abandonment, I mean: the last interaction on a form before someone leaves and doesn’t come back.

A lot of people start by looking at abandonment volume:

• Which fields do people drop off on most often?
• Where is the biggest “pile” of drop-offs?

You might have a table showing:

• number of sessions that interacted with each field
• abandonment count (i.e., how many dropped off there)

And that’s useful — because high volume means more scope for improvement.

But you also need to look at…

Abandonment rate

You should always look at abandonment rate too.

That’s the percentage of people who abandon at a field, out of everyone who interacted with that field:

• abandonments at that field
  ÷
• total interactions with that field

Why does that matter?

Because sometimes a field has fewer drop-offs in total, but a much higher propensity to cause abandonment.

For example:

• A “Title” field might have a high abandonment count
• But an “Age check” field might have a lower abandonment count and a 45% abandonment rate

That tells you the “Age check” is a serious friction point for the people who reach it.

Why percentages matter so much

Two reasons:

1) Not everyone sees every question.
Conditional logic can change the path. Answer one question, and the form shows a different next question. So a lot of users may never even see a particular field.

2) People can only abandon once.
Users who drop out early can’t drop out later — so later fields will naturally have lower raw abandonment volumes. That doesn’t mean they’re not problematic. If a later field has a high abandonment rate, it can become a major issue as you improve earlier steps and drive more people to it.

So: look at volume and rate.

Error messages

Next metric: error messages.

At this stage, you’re not necessarily judging whether the error messages are “good” or “bad” — that’s a separate topic. You’re asking:

• Which error messages are triggering?
• How often?
• When are they triggering?

Errors are an indicator someone is struggling.

But you must be careful: not all errors are equally related to abandonment.

A practical way to look at this is to compare:

• errors across all sessions
  vs
• errors among sessions that don’t successfully complete (abandoners)

A common pattern is:

• the most frequent error overall might be something like “Please tick the agreement box”
• but among abandoners, the most important errors might be things like:
  • “Card number is required”
  • “This is not a valid card number”

Why? Because a missed checkbox is easy to fix — users recover quickly. Card-related errors are often harder and correlate far more with drop-off.

So always segment error analysis by outcome (completers vs abandoners).

Field returns and corrections

Another strong signal: field returns / corrections.

This is when someone:

• fills a field
• moves on
• then comes back and changes it

Often this is driven by an error message.

Returns/corrections are a great indicator of friction — but again, you should check whether they’re genuinely associated with abandonment (we’ll cover how in the advanced section).

Time spent

The final basic metric: time spent.

A general rule of thumb:

• the more time people spend on a field, the more friction it may be causing

But be careful: time spent can reflect legitimate effort, not frustration.

A name field should be quick.
A “1000-word essay” field should take longer — that’s expected, not necessarily friction.

So time spent is useful, but it needs context and segmentation.

Advanced techniques

Compare abandoners vs completers

The first advanced technique is one you should always do:

Segment behaviour between:

• people who complete the form
  and
• people who abandon the form

If you see a meaningful behaviour difference between those two cohorts, it often points directly to what’s driving abandonment.

You can do this using several metrics — I often use returns/corrections, but you can apply the same thinking to time spent.

Example: returns split by abandoners vs completers

If you look at raw returns alone, you might see something like:

• “Email field has returns around 55%”

That sounds bad. But when you split it:

• if abandoners and completers both return to email at similar rates, it may not be driving abandonment

Now compare that to a field like phone number:

• 25% of abandoners return to fix it
• 17% of completers return to fix it

That difference is your clue:

• it’s friction
• and it’s friction that’s more closely associated with abandonment

You can do the same with time spent:

• abandoners spend 16 seconds
• completers spend 10 seconds

That’s a strong signal there’s a problem and it’s related to drop-off.

Analyse what happens after failed submissions

The second technique is: what happens after failed submissions.

These are users who:

• make it through the form
• click submit
• but don’t successfully complete

These are high-intent users — they’ve invested time, they’re ready to proceed, and something goes wrong. That’s money left on the table.

So you want to analyse:

• what happens after the submit click?
• where do they go next?
• what are they trying (and failing) to fix?

A good analytics tool should show this clearly — for example, as a flow chart from the submit button to the next field(s).

A pattern you often see:

• a chunk of users abandon immediately after failed submit (they see errors and give up)
• but many jump back to specific fields trying to fix issues

In the example shared:

• after failed submit, users most commonly jump back to:
  • Set password
  • Phone number

And because the cohort is specifically “people who did not successfully complete,” you know they struggled with those fields and then abandoned.

This is one of the fastest ways to pinpoint where your form is truly failing.

Segment at field level for different audiences

The final technique: field-level segmentation by audience.

Averages can hide major problems.

You might see a field with a “reasonable” average abandonment rate — but when you split by device or platform, you uncover something completely different.

Example: voucher field split by device

Voucher fields are notoriously tricky in checkout.

You might see:

• overall abandonment rate ~21%
• mobile abandonment ~22%
• desktop abandonment ~15%

That tells you: mobile is driving the issue.

And it also helps explain behaviour:

• desktop users can easily open a new tab and hunt for a voucher code
• mobile users often try, get frustrated, and abandon

Go deeper than “mobile vs desktop”

Even “mobile” can be too broad.

Example:

• mobile abandonment 18.5%
• desktop abandonment 5.8%

That looks like “mobile is the problem.”

But split mobile by OS:

• iOS: 6.1% (close to desktop)
• Android: 35% (a third of users drop)

Now you know where to focus:

• not “mobile” generally
• but Android users, potentially down to:
  • specific Android devices
  • specific Android browsers
  • specific UX or performance issues on that combination

This kind of drilling down is how you find the cohort that’s struggling and build targeted hypotheses to improve the UX.

Wrap-up and next steps

Those are the main metrics and techniques:

• base metrics to get started (abandonment, errors, returns, time spent)
• advanced methods to pinpoint true causes (abandoner vs completer differences, post-failed-submit behaviour, field-level segmentation by cohort)

If you want more depth, we have a white paper on using data to optimise forms on the Zuko website (easy to find).

I’ll pause briefly for questions, but because of the delay I may not get many live. If you do have questions:

• ping me on LinkedIn
• or drop a comment later and I’ll respond.

‍