Which Marketing Channels Bring Paying Customers

To find out which marketing channel brings paying customers, connect each payment to the source of the visitor who made it, then compare channels on customers and revenue per visitor rather than on traffic. In most early SaaS products the answer surprises the founder: the channel that sends the most visitors is rarely the one that sends the most money, and one or two small sources — a niche newsletter, comparison pages, AI assistant referrals — usually punch far above their traffic.

Why does traffic lie about which channel works?

Traffic is the easiest number to get and the least connected to revenue. A front-page Hacker News post can send 10,000 visitors in a day, most of them curious engineers who will never buy a $29 tool. A mention in a 4,000-subscriber newsletter for your exact niche might send 300 visitors and 6 customers. On a traffic chart, the first is a mountain and the second is a bump. On a revenue chart, they swap places.

Signups are better, but still misleading. Some channels bring people who sign up out of curiosity and never return. Others bring fewer signups who almost all convert. The only number that settles the argument is money — and the only way to get money per channel is to connect payments to sources. If you haven't done that yet, start with how to track revenue by traffic source; the rest of this playbook assumes you have.

What should you measure for each marketing channel?

Revenue per visitor is the single most useful number on that list, because it combines audience quality, conversion and price into one figure you can compare across channels of very different size. It also tells you what you can afford to pay for a visitor: if a sponsorship's visitors are worth $0.40 each over 90 days, paying $0.25 per visitor is profitable and paying $1.00 is not. We go deep on it in revenue per visitor: the metric indie hackers should watch, and you can run your own numbers with the revenue per visitor calculator.

How do you find your best marketing channel, step by step?

1. 1.Track the two conversions that matter: signup (or trial start) and first payment. Signups need one event call in your app; payments come from your payment provider's webhook, matched to the visitor.
2. 2.Tag every link you place yourself with UTM parameters, using one naming convention. Without them, email, Slack and app traffic lands in “direct”. The UTM guide has a copyable convention.
3. 3.Group sources into channels you can act on: organic search, AI assistants, communities, social (organic), paid social, newsletters you own, sponsorships, partners and affiliates, direct. Don't let one-off referrers clutter the list — bucket anything under 1% into “other”.
4. 4.Pull 90 days of data per channel: visitors, signups, customers, revenue. Ninety days covers at least two conversion cycles for most self-serve products.
5. 5.Compute signup rate, revenue per visitor and median days to convert for each channel. Add cost and CAC where money or meaningful time was spent.
6. 6.Sort by paying customers first, then by revenue per visitor. Mark any channel with fewer than 10 customers as “directional only”.
7. 7.Make one decision per channel — scale, fix, hold or cut — write it down with the date, and review in 30 days.

How much data do you need before trusting the result?

Less than a statistician would like and more than a founder usually waits for. The honest rule: conversion counts below about 10 per channel are too noisy to compare precisely. If channel A has 3 customers from 500 visitors and channel B has 5 from 500, that difference could easily be chance. Ten to thirty customers per channel lets you see big differences (2x and up) with reasonable confidence. If you want to check whether a gap is real, plug the numbers into an A/B test significance calculator — the math for “is channel A's conversion rate really higher than B's” is the same as for an A/B test.

The practical consequence: at low volume, don't try to rank eight channels. Find the one or two with an obvious lead and the one or two with obvious nothing, and leave the middle alone until more data arrives. A sample size calculator tells you how many visitors a channel needs before a given difference becomes detectable.

How do the common SaaS channels usually behave?

Every product is different, but channels have personalities. This is the pattern we see most often for self-serve SaaS and developer tools.

The AI row deserves attention in 2026. Ahrefs reported that AI search visitors were 0.5% of its traffic but drove 12.1% of signups (Ahrefs, 2025) — a single company's data, but consistent with what many SaaS founders see: people who arrive from an assistant's recommendation have already been told you fit their problem. If your analytics lumps chatgpt.com and perplexity.ai into generic “referral”, split them out. VisitTrack classifies them as their own AI channel automatically, and also shows which AI crawlers read your pages, which is the leading indicator for that traffic (see AI crawler tracking).

What does a channel review look like with real numbers?

Here is an example 90-day review for a B2B SaaS at about $6,000 MRR, plans from $29 to $99 a month. Cost includes ad spend and sponsorship fees; founder time on organic channels is valued at $50 an hour.

Three things are worth noticing. First, X ads look like 6,200 visitors of momentum, but at $0.03 per visitor against a cost of $0.23 per visitor they lose money — unless a targeting change can triple conversion, they get cut next month. Second, the sponsorships and AI referrals are tiny in traffic and outstanding in value; the sponsorship that worked gets booked again before the others. Third, “direct” is the second-largest revenue source, which usually means untagged links: shared docs, emails, Slack messages. Tagging those turns a mystery into data. The CAC calculator and ROAS calculator do the arithmetic for each row.

When should you cut a marketing channel?

Cutting too early is the most common mistake we see, because social and content channels take longest to pay and get the least credit under last-touch reporting. A useful decision rule:

• Scale when revenue per visitor is above your average and the channel can grow — more posts, more pages, a bigger budget, a repeat sponsorship.
• Fix when a channel has meaningful traffic but conversion below average. The audience might be right and the landing page wrong; test a dedicated page before giving up.
• Hold when you have fewer than 10 customers from a channel. Keep going at the current effort and collect data.
• Cut when a channel is below average on revenue per visitor under first touch, last touch and linear attribution, after two full conversion cycles, and a fix attempt didn't move it.

Check the attribution model before deciding, too. A channel that is weak under last touch but strong under first touch is a demand creator — cutting it will hurt slowly and invisibly, as fewer new people enter the top of your funnel. The worked example in first-touch vs last-touch attribution shows how a podcast with zero last-touch credit introduced 20% of customers.

What do analytics miss, and how do you fill the gap?

Some channels are structurally invisible to trackers. Podcasts and YouTube videos where people type your name later, word of mouth in private chats, a conference talk. SparkToro's 2023 research found that visits from TikTok, Slack, Discord, Mastodon and WhatsApp arrived with no referrer at all (SparkToro), so they all look like direct traffic.

Two inexpensive fixes. Add a “How did you hear about us?” field to signup, free text or a short list, and read it every week; it is crude but catches exactly what trackers can't. And use unique, tagged URLs or discount codes for offline and audio channels (a /podcast landing page with its own UTMs, a code read out on the show). When self-reported answers and tracked first touch disagree, trust the pattern over either one alone.

If you work on a small team or alone, the indie hackers use case and SaaS use case pages show how founders set up this review in VisitTrack with nothing more than the script, a signup event and a Stripe connection — the Revenue view then lists sources by the money they brought in, not just the visits.