Pilot Your Referral Program Before You Buy a Tool: What to Track

A director of marketing told me something on a call this week that most vendors would not want to hear. She has to bring in twice the business next year. Customer advocacy is already one of her strongest channels. Referrals are an obvious next step. And she is not buying a referral tool yet.

Her reason: "We do sort of track referrals that come through. But I don't think it's something you can quantify." So before adding software, she is launching a manual employee and customer referral program, tagging every referral at the source, and running it as a pilot. If the numbers hold, she automates. If they don't, she has lost nothing but a few weeks.

She is right, and it is the order I would recommend to anyone, including people who will eventually buy from us. Here is how to run that pilot so it actually answers the question.

Why most referral programs can't prove anything

Ask a sales team where their best deals came from and "a referral" comes up constantly. Ask for the number and it falls apart. Three reasons:

• Referrals arrive through side doors. A customer emails the AE, a board member texts the CEO, a CSM mentions a name in Slack. None of it touches a form, so none of it gets a source.
• The source field is overwritten. The lead later downloads an ebook, and marketing attribution claims it.
• Nobody records the asks that went nowhere. Without the denominator, you cannot tell a channel that works from a few lucky intros.

A pilot fixes all three by deciding, before the first ask, exactly what gets recorded.

The five fields to tag on every referral

Add these to the lead or contact record in your CRM. Keep them as picklists where you can, so the reports are clean.

• Referral source type: Employee, Customer, Investor or board, Partner. The four groups behave differently and should be measured separately.
• Referred by: The person's name (link to their contact record). Lets you see who your real connectors are, and thank them.
• Asked or unprompted: Asked, Unprompted. Separates what your program caused from what would have happened anyway.
• Date of ask: Date. Lets you measure time from ask to meeting.
• Outcome: Declined, No reply, Intro made, Meeting held, Opportunity, Won. The funnel. Update it as the referral moves.

Also log the asks that do not turn into referrals. A simple sheet with the date, who was asked and the result is enough. That is your denominator.

The three numbers to report

At the end of the pilot, these three answer the question leadership will ask, which is "should we put more money into this?"

1. Ask-to-intro rate. Introductions made divided by asks sent, by source type. This tells you whether people will help when asked.
2. Intro-to-meeting rate. Meetings held divided by introductions made. This tells you whether the intros are to the right people.
3. Cost per meeting. The time spent by CSMs, executives and marketing, plus any incentives, divided by meetings held. Compare it with what a meeting costs you from paid channels or outbound today.

Report them separately for employees and customers. In most programs the two look very different, and blending them hides which one deserves the investment.

How long to run it, and how big

Sixty days is usually enough to see a pattern without losing momentum. Pick a slice you can actually manage by hand:

• Customers: the accounts with a recent good moment, such as a strong QBR, a renewal or a go-live, not the whole base. Asking only happy customers is the point, not a bias to correct.
• Employees: start with the leadership team and your most connected sellers rather than an all-hands announcement.

Make every ask specific. "Do you know anyone?" produces very little. "Would you introduce us to the VP of operations at this company, whom you worked with at your last job?" produces a decision.

The two signs it is time to automate

Run the pilot by hand and you will hit one of two walls. Both are good news.

The CSMs run out of time. The asks work, but finding who each customer knows, writing each note and chasing each reply is eating the hours your CSMs need for retention. The work is real and it is repeatable, which is exactly what software is for.

You run out of obvious asks. The first round used everyone's best-known relationships. The next round needs someone to work out which of your 2,000 customers or 200 employees knows which of your target accounts, and nobody can do that by hand.

If you hit neither wall, the pilot told you something too: the channel is small for you right now, and a tool would not change that.

What to ask a vendor once you get there

When the pilot earns the budget, the questions that matter come straight out of it:

• Does it keep my five fields? The tool should write referral source, connector and outcome back to the CRM so the reporting you built during the pilot carries on.
• Who has to share what? Your customers and partners should not have to upload contacts or install anything for you to find paths through them. Ask exactly which data the tool uses for whom.
• Who sends the ask? The note should go from the person who has the relationship, from their own account. An introduction sent by software on someone's behalf is a cold email with a borrowed name.
• Can it target only happy customers? The pilot taught you that timing and sentiment matter. The tool should let you ask after good moments, not blanket the base.

That last set is how we built Boomerang. Rudy, our agent, maps who across your employees, customers, investors and partners knows each target account, mostly from signals you already have and public ones, so your customers and partners do not have to share anything. He drafts the ask for the right person to send from their own account, and writes the outcome back to your CRM. Here is how connectors and Primary Super Connectors differ.

But run the pilot first. A referral program you can measure is worth automating. One you can't is just a faster way to not know.