Inquiry Response Time vs. Close Rate: A Data-Driven Analysis
1. Align on definitions before you analyze
The first step in this kind of quantitative analysis isn’t charting — it’s nailing down definitions. Otherwise everyone on the team means something different by “response time” and “close rate,” and the curve you draw is useless.
| Metrics | Definition | Common mistake |
|---|---|---|
| first response time | Time of first human reply minus form submission time | Counting auto-replies as the first response |
| Inquiry base | Deduplicate after removing repeats, tests, and spam | Counting spam inquiries in the denominator |
| Close rate | Won deals ÷ valid inquiries | Counting in-progress orders as closed |
| Observation window | Full 12-month funnel | Using only the last month — sample too small |
| Close flag | Use the final CRM status | Sales reps fill it in inconsistently |
The rule that matters most is “first response excludes auto-replies.” Auto-replies are sent by the system and don’t reflect your response capability. Counting them would wrongly sort a flood of 3 a.m. auto-replies into the “fast response” bucket and distort the curve. First response timestamps come from Website leads to CRM‘s automatic sync — don’t rely on sales reps to log them by hand.
2. How to compute this curve from historical data
The method has four steps: pull the data, bucket it, compute rates, then segment. Pulling the data means exporting every deduplicated inquiry from the past 12 months out of the CRM, each with submission time, first response time, source channel, intent score, and close status. If you don’t have an intent score field, use Lead Scoring and Prioritization for Inquiries‘s model to fill one in.
| Steps | Action | Tool / field | Note |
|---|---|---|---|
| 1 Pull data | Export 12 months of deduplicated inquiries | CRM export | Remove duplicates and spam |
| 2 Bucket | Split first response time into 6 ranges | 0-5 / 5-30 / 30-120 / 120-480 / 480-1440 / >1440 minutes | At least 50 per bucket |
| 3 Compute rates | Closed deals ÷ inquiries per bucket | Excel or BI | Label the sample size |
| 4 Segment | Slice again by channel / score / time of day | Grouped pivot table | This is where you see the truth |
Sample size per bucket is a hard threshold. If a bucket has only 15 inquiries, a 20% or 0% close rate means nothing — random noise can bury the trend. Rule of thumb: at least 50 valid inquiries per bucket. Below that, merge adjacent buckets.
Example: an industrial equipment site received 1,200 deduplicated inquiries over the past 12 months. Bucketed by first response time, the 0-5 minute bucket had 180 inquiries and 22 closed deals, a close rate of about 12.2%; the over-24-hour bucket had 90 inquiries and fewer than 1 closed deal, about 0.5%. On the surface that’s a 24x gap, which makes it look like response time decides everything.
3. The surface curve: the slower the response, the lower the close rate
With no stratification at all, the curve slopes down with remarkable consistency. This matches the direction of the empirical range in First-response SLA and conversion rate for inquiries: faster is better. But the point of this section is that the curve itself cannot be used directly as causal evidence.
| First response time bucket | Number of inquiries | Deals closed | Close rate |
|---|---|---|---|
| 0-5 minutes | 180 | 22 | about 12.2% |
| 5-30 minutes | 260 | 24 | about 9.2% |
| 30 minutes-2 hours | 300 | 18 | about 6.0% |
| 2-8 hours | 240 | 7 | about 2.9% |
| 8-24 hours | 130 | 2 | about 1.5% |
| Over 24 hours | 90 | 0 | about 0.5% |
Across the six buckets from top to bottom, the close rate falls from 12.2% to 0.5%, with most of the decline concentrated between 30 minutes and 8 hours. This trend shifts up or down by industry, but the direction — faster is better — holds steady. That’s also why you still need an SLA.
4. Confounding variables: why the curve isn’t causal
Here’s the counterintuitive part. Slow inquiries don’t close at a low rate purely because “slow loses the deal.” There’s also self-selection bias: buyers who are genuinely interested and deciding today will keep following up during your working hours and are easier for you to catch quickly; inquiries submitted at 2 a.m. on a whim and forgotten right after already had a low chance of closing, and they happen to land in the “slow response” bucket.
| Confounding variable | How it misleads you | How to control for it |
|---|---|---|
| Submission time | Late-night inquiries are naturally slow to answer and low in intent | Compare fast vs. slow only within the same time window |
| Channel source | Ad traffic has low intent and gets slower replies | Group by channel first, then compare |
| Intent score | Low-score inquiries are both slow and unlikely to close | Compare after controlling for score |
| Buyer type | Competitors fishing for pricing send slow inquiries that never close | Exclude known competitors |
The approach is to compare within the same stratum: for example, look only at inquiries submitted on weekdays between 9 a.m. and 6 p.m. with an intent score of 70 or above, then split them into fast and slow piles by first response time. The fast pile still closes at a higher rate, but the gap narrows from the surface-level 24x to about 2-4x — that’s the part response speed can actually explain.
Example: isolate inquiries submitted “on weekdays during working hours, score ≥70.” The fast pile closes at 18%, the slow pile at 11% — a 7 percentage point gap. Those 7 points are the real return you buy by adding on-call staff and tightening your SLA, not the 24x curve on the surface. For attribution methodology, see How to calculate lead attribution correctly.
5. How to use this conclusion: allocate resources, don’t make everyone chase speed
The ultimate goal of analysis is allocating resources. Since the marginal return on speed is greatest for high-quality inquiries, you shouldn’t require a 5-minute response for every inquiry alike — that’s costly and unsustainable. The right approach is to allocate response speed by value tier.
| Inquiry tier | How to identify it | Response Target | Resources needed |
|---|---|---|---|
| High value | Score ≥80, business hours, specific model | Within 5 minutes | Senior reps get first pick |
| Medium | Score 60-79, vague requirements | Within 30 minutes | Standard on-duty sales rep |
| Low value | Score <60, late night, price-shopping signals | Same day | Template reply + nurture |
For the specifics of intent tiering and BANT assessment, see Identifying inquiry intent and responding differently. Use the quantified findings, but don’t treat them as gospel: they tell you speed has value and who benefits most, not that speed fixes everything.
FAQ
How do you tell whether response time and close rate are actually causally linked?
How many inquiries do I need for this analysis to be reliable?
Why shouldn’t auto-replies count toward first-response time?
Why do slow inquiries tend to be lower quality to begin with?
How do I apply these findings to actual staffing and resources?
How should the close rate metric itself be calculated to be accurate?
Final thoughts
The correlation between response time and win rate is the pair of numbers most easily turned into a “causal myth” in direct-to-consumer site data. The curve is steep, it looks great, and it is easy to show your boss to prove “just add another person on shift” — but half of that steep line is self-selection bias, not the work of speed alone.
Three steps to put this into practice: first, lock down the three definitions — first response time, deduplicated inquiries, and deal status — so the team is not telling different stories; then slice the 12 months of data into 6 time buckets and layer it by channel and scoring period to see the real gap once variables are controlled.
Finally, turn the analysis into a tiered response strategy — 5 minutes for high-value inquiries, same-day for low-value ones. That way you get a resource allocation table you can actually use, not a curve chart meant to move yourself.
One line worth remembering: faster response does give you an edge, that part is true, but do not treat the 24x surface gap as a promise — what you can really buy by adding shifts is probably only two to four times of it, and spending that on your most valuable inquiries is enough.