Audit referral

Referral is the channel your dashboards show least and your revenue may depend on most. At Deux.io, 73% of our clients came through referrals. A prospect who hears about you from a peer types your name, lands on your homepage and shows up as direct or branded search, so a last-click report credits the brand and the real origin vanishes. The audit starts by making that origin visible. Only then can you ask whether it can be produced on purpose.

Throughout this chapter, say a B2B SaaS at 2M ARR adds 50 paying accounts a month and has no referral program. The numbers are invented to show the method.

Measure what last-click cannot see

Add a required free-text field, “How did you hear about us?”, to the signup or demo form, store it on the CRM record, and have sales ask the same question on the first call. Free text beats a dropdown because a dropdown suggests the answer. Classify the responses into about ten values, then put them next to your attribution data.

Origin	Last-click share	Declared share
Direct and branded search	46%	12%
Colleague, client or peer	2%	34%
Non-brand organic	24%	16%
Paid	18%	12%
Community, LinkedIn, podcasts, events	6%	20%
Other or unknown	4%	6%

Keep both columns. Memory is imperfect and last-click is partial, so the useful reading is the gap between them. Here, a third of new customers credit a person, and the analytics give that person 2%.

A few other traces complete the picture: the trend of branded search in Search Console, referring domains from review sites and partners, new signups whose email domain matches an existing customer, and customers who changed company and signed up again from the new one.

Then check whether referred customers are worth more:

SELECT
  CASE WHEN declared_source IN ('colleague', 'customer', 'community', 'partner')
       THEN 'referred' ELSE 'other' END AS origin,
  COUNT(*) AS signups,
  ROUND(AVG(CASE WHEN paid_at IS NOT NULL THEN 1.0 ELSE 0 END), 3) AS paid_rate,
  ROUND(AVG(first_mrr) FILTER (WHERE paid_at IS NOT NULL), 1) AS avg_first_mrr,
  ROUND(AVG(CASE WHEN churned_at IS NULL OR churned_at > paid_at + INTERVAL '12 months'
                 THEN 1.0 ELSE 0 END)
        FILTER (WHERE paid_at < CURRENT_DATE - INTERVAL '12 months'), 3) AS retained_12m
FROM accounts
GROUP BY 1;

If referred accounts convert better and stay longer, you can afford to spend more to produce them. Compare them with the retention cohorts you already built.

Find out why people talk, and when

Interview 8 to 10 customers who sent someone your way. Ask whom they told, what triggered it and which words they used. Those words are often better than your homepage, and they feed straight into your positioning. An NPS survey can identify who to call, and its comments are worth more than its score.

Then map the moments when asking makes sense, and check which ones exist today.

Moment	Signal in your data	The ask	Format
First clear result	Activation event reached	Who else on your team needs this?	In-product invitation
Quantified win	Result shown in a customer review	Permission to publish the story	Call, then a one-page case
Renewal or expansion	Contract renewed or plan upgraded	A review on the platform your buyers read	Email with a direct link
Champion changes company	Job change on LinkedIn	An introduction to the new team	Personal message
Support case closed well	High satisfaction rating	A short testimonial	Reply in the same thread

The honest result is often a column of blanks. That is a good finding, because each blank is cheap to fill.

Audit the referral program, or the case for one

Read a program as a funnel: eligible customers, customers who see the offer, customers who send an invitation, invited people who sign up, invited people who pay. The viral coefficient K = invitations per customer × invitation-to-paid conversion is a useful diagnostic. In B2B, judge the program on its cost per paying account against your other channels, including rewards, tooling and time.

Item	What to check	Warning sign
Reward	Whether it suits the person referring and the company paying	Cash that a company buyer can’t accept personally
Timing	When the reward is earned	Paid on signup, which invites abuse
Visibility	Where the offer appears	Buried in settings, never at a moment of success
Friction	Clicks to send an invitation	The customer has to write the pitch alone
B2B fit	Whether an introduction is possible	A consumer-style link in a market that sells through conversations
Partners	Agencies, consultants and integrators with their own terms	They recommend you with nothing to support them

A program multiplies word of mouth that already exists. If the interviews show nobody has much to say, the work belongs in the product and in activation, and the program can wait. Product-led loops deserve the same audit: teammate invitations, shared documents, public templates, a discreet “made with” mention on what your customers publish.

Reviews and proof

Pick the platforms your buyers read in each market: software directories, app marketplaces, Google Business Profile for local services. Compare yourself with the same three competitors you used earlier.

Platform	Metric	You	Competitor A	Competitor B	Competitor C
Directory 1	Reviews, total
	Reviews, last 12 months
	Average rating
	Share answered by the vendor

Search <your brand> reviews and <category> alternatives as well, and note which of these pages rank. Read the negative reviews for recurring themes, then pass them to product and to the positioning work. Follow each platform’s rules on incentives, and never buy reviews or filter out the bad ones: fake reviews are a legal risk in the EU as well as a reputational one.

On your own site, audit where proof sits and whom it speaks to. Logos and testimonials should be segmented by country, since a Spanish buyer learns little from a wall of French logos. Place proof next to the decision, on pricing and feature pages, and aim for one success story a month so the stock keeps growing.

Community and distribution

Audit a community on activity: members who post in a month, questions answered by someone other than your team, members who later signed up, retention of members against non-members. Note what you own and what you rent, since a LinkedIn audience and a Slack workspace carry very different risks. A community is also a secondary conversion, a way to keep someone close before they are ready to buy, so check that your pages offer it in context. Don’t launch one until you have enough people and someone with time to run it. An empty channel is worse proof than no channel.

Content travels through people when distribution is a workflow per type, written once and followed every time.

Content	Your channels	People who carry it	Reuse
Success story	Newsletter, proof page by country	The customer posts it, sales sends it to similar accounts	One quote per persona page
Webinar	Dedicated page with transcript, email to registrants	Guest speakers invite their audience	Short clips, replay in nurturing
Benchmark from your data	Landing page, press pitch	Customers quoted, partners	One chart per social post
Free tool	Tool page, in-product link	Communities where the problem comes up	Results that people want to share

What to deliver

A finding here reads like this: “A third of new paying accounts credit a person, last-click credits 2%, and no moment in the journey asks for a referral. Action: add the first-result invitation and the renewal review request. Metric: share of paying accounts with a referred origin, read each month.”

Deliver the last-click versus declared table, the quality comparison, the moments grid, the review benchmark and one distribution workflow per content type.

Your next step: add the free-text question to your signup form today, then read the first 100 answers yourself before classifying anything.