Exchange trader retention: why the second trade matters

A first trade shows that a new trader tried the product. A second trade shows they came back. That makes the second trade one of the earliest retention signals an exchange has. It is tracked with second-trade rate, time to second trade, and D7 and D30 trading retention by cohort. When traders stop after one trade, the cause usually traces to an information gap, a cost surprise, or how the first trade turned out.

This guide is for exchange and broker product and growth teams. It covers the metrics, the drivers of drop-off, and how to measure them.

Why the second trade matters

The first trade is often pushed by onboarding. A deposit bonus, a guided flow or a referral can carry a new user through it. That makes the first trade partly a measure of the funnel.

The second trade has less help. The trader returns, opens the app and acts again. It is the first sign that the product is earning a place in their routine.

It also arrives early. D30 retention takes a month to read. A second-trade rate within seven days can be read within a week, which helps when testing changes.

The core cohort metrics

Build cohorts by the date of first trade, usually by week. Then track:

Metric	Definition	Formula
Second-trade rate	Share of a cohort making a second trade within a window	Traders with a 2nd trade in window ÷ cohort size
Time to second trade	Time from first to second trade	Median, and the 75th and 90th percentiles
D7 trading retention	Share trading again in the day-7 window	Traders active in window ÷ cohort size
D30 trading retention	Share trading again in the day-30 window	Traders active in window ÷ cohort size
Trading days in first 30	Distinct days with a trade	Count per trader, then distribution

Arithmetic example, illustrative numbers only, not a benchmark. A weekly cohort has 1,000 first-time traders. Of those, 380 make a second trade within 7 days. The 7-day second-trade rate is 380 ÷ 1,000 = 38.0%. If 250 of the cohort trade during the day-30 window, D30 trading retention is 250 ÷ 1,000 = 25.0%.

Define the windows once

Retention numbers are easy to make incomparable. Pick definitions and fix them:

• Does “D7” mean any trade in days 1 to 7, or a trade on day 7 itself?
• Do conversions between assets count as trades?
• Do copy trades, bots or automated recurring buys count?
• Which time zone sets the day boundary?

Write the definitions down and apply them to every cohort.

What drives drop-off after one trade

Information gaps

The trader cannot answer a question about what happened or what to do next.

• “Why is my position showing a different price than I paid?”
• “What is this funding charge?”
• “How do I close half of this?”

A trader who does not understand their first position is unlikely to open a second. The questions behind first-trade stalls are covered in why new traders stall before their first trade.

Cost surprises

The trade cost more than the trader expected.

• Fees higher than they assumed, often the taker rate on a market order. See maker vs taker fees.
• Slippage on a market order in a thin book. See slippage explained.
• Funding on a perpetual held longer than planned.
• Spread on an instant-convert flow.

A cost the trader saw before confirming is a decision. A cost they discover afterwards is a surprise, and surprises erode trust.

First-trade outcomes

How the first trade ended shapes whether there is a second.

• A quick loss, especially a liquidation, often ends the relationship.
• A confusing partial fill or an unfilled limit order can read as “the product did not work”.
• Even a gain can end activity if the trader does not understand why it happened.

The outcome itself is the market’s doing. Whether the trader understood it is partly the product’s.

How to measure the drivers

You can trace each driver with data most exchanges already hold.

1. Segment second-trade rate by first-trade outcome. Split cohorts by profit, loss, liquidation, partial fill and unfilled order. Large gaps between segments point to the outcome as a driver.
2. Compare quoted and realised cost. For each first trade, compare the fee and price shown before confirming with what was charged. Measure second-trade rate by the size of the gap.
3. Look at sessions between first and second trade. Did the trader return to the app without trading? Which screens did they visit? Repeated visits to a position or history screen suggest an unanswered question.
4. Track help-link taps and exits. Taps on help content from a position screen, followed by a session ending, mark information gaps. See where traders go with questions.
5. Segment by product and channel. Spot, convert, futures and perpetuals often retain differently. So do paid, referral and organic users.

What tends to help

These are design patterns, not certain fixes. Each one is testable.

• Show the full cost before confirmation. Estimated fill, fees and, for leverage, margin and estimated liquidation, all labelled as estimates.
• Explain the position after the trade. What it is worth, what it costs to hold, what closing it would do.
• Explain outcomes plainly. Why a limit order did not fill, why a fill differed from the shown price.
• Make the second order easy to draft correctly. Quantity echoes like “Sell 0.5 ETH (50% of your 1.0 ETH)” remove a common source of error. See AI order ticket fields.

Where an embedded agent fits

An embedded conversational trading agent answers questions at the moment they arise, inside the app, with the venue’s live data. It can explain a first position and draft a second order for the trader to confirm. It does not advise on what to trade. For how one works, read what an embedded conversational trading agent is.

Whether it lifts second-trade rate on a given venue is a question for a controlled pilot. The metrics above are the ones to use; measuring an AI agent’s impact covers pilot design.

For exchanges and brokers exploring a pilot, askthehippo.com has more.