.svg)
We took 1,332 conversations between buyers and RepX across five B2B software companies, then analyzed the 4,017 messages exchanged. Where all five companies show the same pattern, we've said so. Where one of them moves the average on its own, we've given the range instead of a single number.
In summary
- Pricing and capability questions account for 42% of everything buyers ask. Security and compliance accounts for 1.5%.
- Qualification rises with conversation length at every company, from 17% of one-message conversations to 60% of the longest.
- Email capture follows the same curve, from 1% to 33%. The median converting conversation runs five buyer messages.
- 51.4% of conversations start outside the buyer's own working hours, and those conversations are no shorter and no less qualified.
- Buyers who bring up a competitor have the longest and most qualified conversations in the dataset, and are the least likely to hand over an email.
1. What buyers ask RepX about
Of the 4,017 messages, 2,225 were questions or requests for information. The rest were buyers answering something the agent had asked them.
Pricing and capability together come to 42%. The categories that absorb most of a marketing team's effort come to much less: customer proof at 2.0%, ROI and business case at 2.0%, competitive differentiation at 5.1%.
Security and compliance came to 1.5%, or 34 questions out of 2,225, and it sits between 0.5% and 3.0% at all five companies. That figure is worth holding next to a separate analysis of 300 recorded sales calls with around 150 software companies, where those same vendors named compliance and regulation as the biggest structural reason their buyers can't evaluate the product independently: 43% of all the reasons they gave. The obstacle sellers describe as their largest is not one buyers raise.
Pricing itself is the least consistent number here. At one company it's 6.0% of questions and at another it's 37.4%, which tracks with how self-serve the purchase is rather than with anything about the buyers.
What this means for you
Check what your agent does with a pricing question, because a quarter of everything it receives is one. If the answer is a form or a calendar link, you're deflecting the single most common thing anyone asks you.
Then look at where your pricing and capability answers actually live. Both are usually written for a human to deliver on a call, which means the agent either can't reach them or paraphrases them badly.
2. Conversation length predicts qualification
This is the most consistent result in the set. It holds at all five companies, and the movement is similar in each: 13% to 44%, 29% to 55%, 6% to 50%, 17% to 59%, and 26% to 56%.
The duration figures explain what the short conversations are. A one-message conversation lasts a median of 14 seconds. An eleven-message conversation lasts 1,043 seconds, or a little over 17 minutes.
What this means for you
Score inbound on conversation depth rather than on whether a form got completed. You already collect the number and it predicts better than anything on the form does.
Six messages is where qualification crosses 50%, which makes it a reasonable handoff threshold. At 14 seconds, your single-message traffic is bouncing rather than browsing, so there's little point building nurture around it.
3. Buyers convert once they're invested
Email capture follows the qualification curve almost exactly. A buyer in a long conversation is around 30 times more likely to share an email than one who sends a single message. The median conversation that captures an email reaches five buyer messages first, against three for conversations overall.
The direction holds at all five companies, though the steepness varies. One runs from 0.5% to 39%, another only from 1% to 6%.
What this means for you
Anything that asks for an email before the buyer has asked three or four questions is being answered by about one person in a hundred, and most of the other ninety-nine haven't disqualified themselves. They just haven't spent anything yet.
Let the agent answer freely through the early exchanges and ask afterwards. You'll ask fewer people and a much higher share will say yes.
4. Half of conversations start outside of working hours (and quality holds up)
We converted every timestamp into the buyer's own local time rather than the vendor's.
51.4% start outside weekday business hours. Across the five companies that runs from 38.5% to 55.9%, so no single deployment is carrying it.
The quality question matters more than the volume, and the answer is that nothing much changes. Mean conversation length is flat between in-hours and out-of-hours traffic at every company. Qualification is flat at three of the five and somewhat lower at two, which is not a consistent penalty. The composition is also less dramatic than the headline suggests, since the largest off-hours band is weekday mornings before 9am at 19.6%.
What this means for you
Convert your own inbound timestamps into buyer-local time. Most teams have only ever looked at this in their own timezone, which hides the pattern completely.
Then decide whether a form and an autoresponder is the experience you want half of your inbound to get.
5. Buyers who raise a competitor go deepest
When a buyer named a rival product or asked for a comparison, the conversation looked different from every other kind.
Those conversations run more than twice as long and qualify at 53.2% against 36.0%. The pattern holds at all four companies with enough of them to measure, on both length and qualification.
They're also the least likely to share an email, at 9.7% against 17.9%. The buyers doing the most direct evaluation work are the ones most determined to stay anonymous while they do it. They ask to speak to a human more often too, at 8.1% against 3.6%.
This rests on 62 conversations, which is small. The effect is large and consistent across four companies, but treat the exact figures as indicative rather than precise.
What this means for you
Make sure your agent can handle a competitor question without stalling, because the buyer asking it is one of the most qualified people to reach your site that week. An agent that routes comparison questions straight to sales is routing away your best conversations.
Don't gate comparison content either. The people who want it are already the least willing to identify themselves, and a form in front of it removes them rather than capturing them.
What this all adds up to
Almost every signal worth having in this data depends on how long the conversation runs. Qualification, identification and willingness to speak to a person all climb with message count, and none of them are visible in the first exchange.
Most B2B sites are built the other way round. The form comes first, the calendar second, and the answers to the two questions buyers actually arrived with sit behind both.