---
title: "Pipeline and Referrals for Agencies of 3 to 15 People"
url: https://theagencyarithmetic.com/pipeline-crm/
dateModified: 2026-11-03
---

# Pipeline and Referrals for Agencies of 3 to 15 People

For agencies where the owner still does the selling. Where new clients come from, what a referral that goes cold costs, and when a CRM is worth paying for.

In this topic · 3 pieces

- Chart: Five stages: New conversation, First call done, Proposal sent, Waiting, Won or lost. A reminder 3 business days after each proposal.

  Understand · article The 15-minute referral system How do referrals stop going cold without a sales process? See the 15-minute system
- Chart: The example agency's 12 referrals a year as dots; 1.8 go cold. Each referral is worth $28,560 in expected margin.

  Diagnose · calculator What a cold referral costs What is one referral worth, and how many go cold? Run your numbers
- Chart: Revenue in one year, planning example: Grow the book $89,760, New client $81,600, Add-ons $54,000.

  Diagnose · article Growing current clients, with care Which clients should be offered more, and which should not? See the book math

Skip to: when not to buy

- By [Vinicius Diniz](https://theagencyarithmetic.com/about/)
- Published October 5, 2026
- [Methodology](https://theagencyarithmetic.com/methodology/)

An agency that grows on client referrals can easily handle them from an inbox and memory. A referral that waits three days for a reply, or a proposal nobody follows up, can cost more than a tool to track it, as the math below shows. Before choosing a tool, it helps to know what one referral is worth and how often one goes cold.

## Where new clients come from

[Client referrals are the top source of new business for most agencies surveyed](https://theagencyarithmetic.com/benchmarks/#bm-sparktoro-2024-referrals), and relying on them is the norm, not a weakness. In SparkToro's 2024 survey of 612 agency owners and consultants, 66% named referrals from existing and past clients as their top source of new business.

66% of agency owners and consultants name referrals from existing and past clients as their top source of new business.

Source: SparkToro and Paddy Moogan, State of Digital Agencies 2024, 612 agency owners and consultants, 2024 survey, period not stated. [See all benchmarks](https://theagencyarithmetic.com/benchmarks/)

A 2025 RSW/US survey points the same way: 75% cited referrals as their best lead source. RSW/US sells outsourced new business development, so it has a stake in the finding. That survey mixed agencies with other professional services firms and did not state its sample, so it supports the direction, not the number.

Source: RSW/US, 2025 Survey Report: Rolling Toward 2026. [See all benchmarks](https://theagencyarithmetic.com/benchmarks/)

Finding them has become harder. In the Agency Management Institute's 2025 Agency Core study of 778 agencies, 43% strongly agreed that finding new clients is harder than ever, up from 15% in 2023.

Agency leaders who strongly agree finding new clients is harder than ever

- **2023**: 15%
- **2025**: 43%

Source: AMI, 2025 Agency Core, 778 agency leaders, November 13 to December 31, 2024. [See all benchmarks](https://theagencyarithmetic.com/benchmarks/)

## Referrals are worth more, and they cool fast

A client who arrives by referral tends to stay longer and be worth more, and the chance of winning one drops with every hour the inquiry waits. Both facts point to the same job: reply fast and follow up.

In a study of about 10,000 customers of a German bank over almost three years, Schmitt, Skiera and Van den Bulte (Journal of Marketing, 2011) found that referred customers had higher margins and higher retention, and were worth at least 16% more than similar customers who were not referred. The people who refer are not always the biggest buyers: Kumar, Petersen and Leone (Harvard Business Review, 2007) found that a company's best advocates can be worth more than its most enthusiastic spenders. The studies covered banking, telecom and financial services, not agencies; the direction carries over, the percentages do not.

Speed matters more than it seems. In an audit of 2,241 US companies, Oldroyd, McElheran and Elkington (Harvard Business Review, 2011) found that 37% replied to a web inquiry within an hour and 23% never replied. In a companion study of 1.25 million leads, companies that tried to reach a lead within an hour were nearly seven times as likely to have a meaningful conversation as those that tried an hour later. Those were web form leads, not personal referrals, which likely tolerate more waiting. The lesson that transfers is that each day of delay costs something.

## The math

In the example below, one referral that becomes a client is worth $81,600 in margin, and the referrals that go cold cost $51,408 a year. The example uses a made-up agency; the calculator takes real numbers.

Assumptions

- **Average monthly retainer Example agency**: $6,800
- **Months a client stays Planning assumption**: 24
- **Delivery margin Planning assumption, below Parakeeto's 60% to 70% guideline for a single client, to keep the example conservative**: 50%
- **Referrals per month Example agency**: 1
- **Referrals that become clients Planning assumption**: 35%
- **Referrals that go cold from a slow reply or no follow-up Planning assumption; nobody has measured this for agencies**: 15%

Result

- **Margin from one client**: $6,800 × 24 × 50% = $81,600
- **Expected margin from one referral**: $81,600 × 35% = $28,560
- **Referrals going cold per year**: 12 × 15% = 1.8
- **Margin lost per year**: 1.8 × $28,560 = $51,408

Margin guideline: Parakeeto (agency profitability consultancy), 2024. [See all benchmarks](https://theagencyarithmetic.com/benchmarks/)

At these assumptions, keeping one referral a year from going cold is worth far more than the tool. Every input above is an assumption except the arithmetic. [Change them in the calculator.](https://theagencyarithmetic.com/pipeline-crm/cold-referral-cost-calculator/)

## What a CRM does, and what it doesn't

A CRM keeps every open conversation in one place and puts a date on the next follow-up. It does not reply to anyone, and it does nothing if nobody opens it.

The warning is old and still accurate. In 2002, Rigby, Reichheld and Schefter of Bain wrote in Harvard Business Review that CRM projects failed when companies treated CRM as software that manages relationships on its own. Where people actually use it, the research is positive: Mithas, Krishnan and Fornell (Journal of Marketing, 2005) linked the use of CRM applications to better customer knowledge and satisfaction, and Ahearne and colleagues (Management Science, 2008) found that salespeople's use of technology improved their service and adaptability, and through them their results.

The HBR study on response times even names a CRM habit among the causes of slow replies: pulling new leads from the system once a day instead of as they arrive. For an owner who sells alone, what a CRM needs to do is short: show every referral that has not had a reply, and every proposal without a follow-up date.

## Before hiring someone to sell

A salesperson rarely fixes a pipeline that loses referrals, because the referrals usually ask for the owner. The smaller fix comes first: a reply within a day and a follow-up date on every proposal.

In a small agency, clients usually buy the person who will do the work, and a new salesperson does not carry that credibility on day one. Hiring before there is a record of where referrals come from, and how fast they get a reply, leaves the new hire without the one thing a pipeline needs: history.

## Current clients, with care

Not every current client should be offered more. Selling more to a client that already loses money makes the loss bigger.

In databases from five companies, in both consumer and business markets, Shah, Kumar and colleagues (Journal of Marketing, 2012) found that 10% to 35% of customers who bought more than one service were unprofitable, and accounted for 39% to 88% of the companies' total losses from customers.

Before offering more to a client, check what that client already costs in hours. The client hours calculator in [agency operations for 5 to 15 people](https://theagencyarithmetic.com/agency-operations/) does that in a few minutes.

## When not to buy

A CRM is the wrong buy when there is too little to track, or when the problem is that no referrals arrive at all.

Under one referral a quarter, a shared list in a notes app is enough.

Keep one line per referral: who sent it, the date it arrived, the date of the first reply, and what happened.

If no referrals come in at all, a CRM will not create them. The work is asking current clients, which [the referral system](https://theagencyarithmetic.com/pipeline-crm/referral-system/) covers.

## Next step

This afternoon, list every referral from the last 12 months: who sent it, when it arrived, when the first reply went out, and what happened. The gaps in that list are the case for, or against, a tool.

Pipeline and referrals are one of the topics in Agency Arithmetic's [benchmarks and tool math for agencies of 2 to 15 people](https://theagencyarithmetic.com/).

Correction, November 3, 2026: The 50% margin in the example was attributed to Parakeeto as a guideline. Parakeeto's 50% line is for the whole agency; for a single client its guideline is 60% to 70%. The 50% stays as a planning assumption.
