Where traders go with questions, and what it costs exchanges

During a session, traders ask what moved a market, what a term means, what an order will cost and what happened to their position. Most of the answers come from outside the exchange: a search engine, a social feed or a general-purpose AI chat. Each of those exits pulls attention away at a decision point. Exchange teams can measure the cost with data they already hold.

This guide is for exchange and broker product and growth teams. It maps the questions, the places traders take them, and how to measure what it costs.

The four kinds of mid-session question

Type	Typical question	Data needed for a good answer
Market	“Why is SOL down today?”	Live prices, recent moves, sourced news, timestamp
Concept	“What does funding mean?”	Accurate explanation, ideally with the venue’s own terms
Cost	“What will this order cost after fees?”	Venue fee schedule, account tier, order book depth
Position	“Why is my P&L negative if price went up?”	The trader’s own position, entry, mark price, funding

Concept questions can be answered from general knowledge. The other three need data that only the venue has: its live book, its fee schedule, the trader’s account.

Where traders take those questions

Search engines

Fast and familiar. Good for concepts, weaker for the rest. The results do not know the venue’s fee tier, contract rules or the trader’s position. Search results also include other venues’ content.

Social feeds

Good for “what is happening right now” in broad terms. The trade-offs are accuracy and noise. Commentary is mixed with promotion, and opinion is mixed with fact. It is also the place where a trader is most likely to meet a pitch to trade elsewhere.

General-purpose AI chat

Increasingly common for concept and “explain this” questions. It can explain well, but it has no view of the venue’s live data or the trader’s account unless connected to them. Answers about current prices or the venue’s own rules may be out of date or generic.

The exchange’s own help centre

Accurate for the venue’s rules, but often hard to reach mid-order and written as static articles. Traders rarely go looking in the middle of an order form. Help content also tends to cover concepts and account issues, not market or position questions.

Community chats and forums

Fast answers from other traders. Quality varies, and the answers are not grounded in the trader’s account.

What each exit costs

The cost is not the question. It is where the question happens.

1. The exit is near a decision. A question about cost or a position usually arrives with an order form open or a position on screen.
2. Attention moves elsewhere. Off the venue, the trader meets other content, other offers and other venues.
3. The answer may be wrong for this venue. Generic fee figures, another venue’s contract rules or a stale price can lead to a different decision than accurate data would.
4. Some sessions do not resume. The trader may come back, or the order may never be placed.

None of this needs an industry statistic to evaluate. Each venue can measure its own version.

How to measure it

Most of these signals come from standard product analytics. Define each one, then compare sessions with and without the event.

1. Help-link tap, then exit. Share of taps on help or external links from trading screens followed by the session ending within a set window, such as five minutes.
2. Backgrounding during an open order form. Share of order forms where the app goes to the background before submit or cancel. On mobile, this often marks a lookup in another app.
3. Return to the same screen. Of those who leave mid-order, how many come back to the same pair or form within a set time, and how many then submit.
4. Order abandonment by last field touched. Abandonment at the order-type selector, the leverage control or the quantity field points to specific questions.
5. In-app search queries. Where the app has search, the terms traders type are a direct list of their questions.
6. Short surveys. A one-question prompt after a help-link exit, such as “What were you trying to find?”, collects the questions in the trader’s words.

Arithmetic example, illustrative numbers only. In a week, traders tap help links from the order form 2,000 times. In 700 of those sessions, the app goes idle within five minutes without an order. The help-tap exit rate is 700 ÷ 2,000 = 35.0%. Compare that with the exit rate on order-form sessions with no help tap to see the difference the question makes.

Segment every metric by product, device, country and trader tenure. Questions differ sharply between a first-week trader and an experienced one.

The case for answering in-app

If the questions are going to be asked anyway, the argument is about where they get answered.

• The data is already there. The venue has the live book, the fee schedule and the trader’s position. An in-app answer can use all three, with sources and a timestamp.
• The decision stays in one place. The trader asks, reads the answer and returns to the order on the same screen.
• The answer can show the effect of an action. “What would closing half cost?” can be answered with the exact quantity, estimated fill and fees.
• The venue keeps the relationship. The answer comes from the venue’s own product, not a third party with its own interests.

The boundary matters. An in-app answer explains what the data shows and what an order would do. It does not tell the trader what to trade. See information vs investment advice.

Where an embedded agent fits

An embedded conversational trading agent is one way to answer in-app. It sits next to the trading flow, answers with the venue’s live data, and can draft an order for the trader to confirm. It is not a support chatbot and does not replace ticketing; see trading agent vs support chatbot.

Hippo arrives as a docked chat panel delivered by a thin-client SDK, so it adds this without changing the venue’s native screens. Whether it reduces exits on a given venue is measurable in a controlled pilot. Measuring an AI agent’s impact covers the design.

For the activation side of the same problem, read why new traders stall before their first trade. More on exchange growth, or see askthehippo.com.