How to size your inside sales team when leads suddenly surge, and where AI agents come in
At a B2C subscription business, paid media reached a new level and lead volume surged. The bottleneck moved from marketing to the inside sales team. We treated the funnel like a production line: we measured each salesperson's capacity, the queue and response time, hour by hour. The queue formed in the hours with no coverage. The company moved from a single business-hours shift to two shifts covering the whole day and reinforced the team. Lead response time fell by 40%. Today, a good part of that queue can also be handled by AI agents.
Profile
B2C subscription · Series A
Problem
More leads should have meant more sales, but the team only worked business hours and couldn't keep up with the new demand. Outside those hours, leads waited and went cold. Media spend was buying leads that sales couldn't reach in time.
Solution
Look at the funnel as a production process with several stages and several owners: marketing, inside sales and operations. Each stage has capacity, a queue and a cycle time. From there, measure the team's real capacity throughout the day and design shifts and headcount around where the queue forms, not around the daily average.
What we did
- We measured each salesperson's handling speed: how many leads they handle per hour and how long each contact takes.
- We crossed, hour by hour, incoming leads with available capacity. We calculated lead time, the time between a lead arriving and the first contact, to see in which hours queues formed.
- We read the conversion rate throughout the day, as capacity varied, to show how much the queue cost in sales. We tied this work to the S&OP model, so that media, sales and operations planned for the same volume.
Result
With two shifts covering the whole day, lead response time fell by 40%. The higher volume of leads coming from media started being handled in time, instead of going cold in the queue. The analysis also showed that the new demand required more salespeople.
Key learnings
- When marketing accelerates, the bottleneck moves to sales: a chain is only as strong as its weakest link. Plan team capacity together with the media plan, not after it.
- The daily average hides the queue. Look hour by hour, using the median rather than the average, and treat each funnel stage as a production station, with capacity, a queue, a cycle time and an owner.
- Today, part of the queue can be solved with AI agents. They respond in seconds and at any hour, qualify and book meetings. Humans step in where the conversation is worth most: higher-value leads, tough objections, sensitive topics and closing.
Do it yourself
Map the funnel as a production line and measure each stage:
- list the stages (lead, first contact, qualification, proposal, sale), the owner of each and the expected SLA;
- export from the CRM, for the last 4 to 8 weeks, each lead's arrival time, the time of each stage change and who handled it;
- calculate, per stage, volume, pass-through rate and cycle time. Use the median and the 90th percentile; the average is only useful for comparison, because a few extreme cases distort it.
Map capacity hour by hour and split the work between AI and humans:
- for each hour of the day, compare incoming leads with possible contacts (salespeople × contacts per hour for each). Where arrivals exceed capacity, there's a queue;
- put an AI agent on the hours and volumes humans don't cover: first contact, frequently asked questions, qualification and booking;
- size the human team for the leads the AI qualifies and passes on, at peak hours. Set clear hand-off rules (lead value, purchase intent, customer request) and measure conversion separately for each type of contact;
- to connect this to results, multiply conversion by lead-time band by the number of leads and the average ticket. You'll see how much each hour of queue you remove is worth in sales.
Ask an LLM, with the export stripped of names and contact details:
Act as a sales operations analyst. I'll paste a CRM export with, for each lead, the date and time of arrival, the date and time of each stage change (first contact, qualification, proposal, sale), the owner and the source channel. 1) Clean the data and tell me what you removed. 2) For each stage, calculate volume, pass-through rate and cycle time with the median and 90th percentile, and show the average alongside only for comparison. 3) Hour by hour and by day of the week, calculate lead arrivals, handling capacity, queue and lead time, and tell me which stage is the bottleneck. 4) Calculate conversion by lead-time band and what it's worth per month with an average ticket of [X]. 5) Propose a shift schedule and a split between an AI agent and the human team: what the AI handles on its own and when it hands off to a human. 6) Build a dashboard in a single HTML file, with these charts, filters by channel and week and an assumptions area I can edit.