SaaS Conversion Funnel: Benchmarks From Visit to Paid User (2026)

Most SaaS financial models start with a cell that says "MRR grows 15% per month." Revenue does not work that way. Revenue arrives through a funnel: someone visits your site, signs up, reaches value, pays, and stays. Each of those steps has a conversion rate, each rate differs by channel, and each one is a lever you can actually pull. A growth-rate cell has no levers — it is a wish with formatting.

This guide gives you 2026 benchmarks for every stage — visit→signup, signup→activation, trial and freemium→paid — broken down by conversion model and by channel. It carries one fictional startup's numbers through the whole funnel so you can see how the math compounds, and it shows why a 1-point improvement at signup→paid can match a 20% increase in traffic spend.

The five stages: visit → signup → activation → paid → retained

Every self-serve SaaS funnel has the same skeleton:

1. Visit. Someone lands on your site from search, an ad, a referral, or a shared link.
2. Signup. They create an account — free plan, trial, or waitlist.
3. Activation. They reach the first moment of real value: the imported project, the first report, the connected data source. Not a login — an outcome.
4. Paid. They enter a card and a subscription starts.
5. Retained. They are still paying in month 2, month 6, month 12.

Modeling revenue funnel-first instead of growth-rate-first matters for one reason: auditability. When your forecast is built as traffic × conversions, every number in it can be checked against something you measure. When someone asks "what happens if paid conversion drops 3 points," the model answers. A blended MRR growth assumption answers nothing, hides channel mix, and produces cash timing that is wrong by a full trial period.

The core identity of the whole funnel:

New paid users = Visits × Visit→Signup × Signup→Paid

Multiplicative, which cuts both ways. Rates compound when they improve — and a single weak stage caps everything upstream of it.

Visit→signup benchmarks for 2026

Visit→signup is where the widest spread lives, because it depends almost entirely on traffic intent. The same landing page converts a branded-search visitor and a cold paid-social click at rates 5x apart.

Traffic type	Visit→signup	Notes
Cold paid social (Meta, TikTok)	1–3%	Interruption traffic; nobody was looking for you
Paid search, non-brand	2–5%	Intent exists, trust doesn't yet
Dedicated landing page, self-serve average	2–5%	The standard planning range
High-intent organic search	5–10%	"best X tool" and comparison queries
Branded search and direct	8–15%	They already decided; don't get in the way
Referral / word of mouth	8–12%	Borrowed trust converts

Two planning rules. First, a blended 3–4% is a reasonable starting assumption for a self-serve product with mixed traffic; below 1.5% on a dedicated landing page means the page or the targeting is broken. Second, sending paid traffic to your homepage instead of a matched landing page typically costs you a third to half of the conversion — the homepage is built for everyone, which means it is built for no one.

Signup→paid benchmarks by conversion model

The signup→paid rate is not one benchmark. It depends on which conversion model you run, and the ranges barely overlap:

Conversion model	Signup→paid	Notes
Freemium	2–5%	Great products with strong upgrade triggers hit 5–10%
Opt-in free trial (no card)	15–25%	The default for most self-serve SaaS
Opt-out trial (card required)	40–60%	High rate, but the card wall cuts signups 50–75% upstream
Demo / sales-led	20–30% of held demos	Add demo-request→held at 60–70%

Activation is the stage hiding inside signup→paid

Signup→paid is really two conversions multiplied together:

Signup→Paid = Signup→Activation × Activation→Paid

For self-serve products, signup→activation typically runs 25–40%, and activated users convert to paid at 40–60% — versus low single digits for users who signed up and never reached value. Multiply the midpoints: 0.32 × 0.5 = 16%, right inside the opt-in trial range. That decomposition tells you where to work. If activation is 25% and activated→paid is 55%, your problem is onboarding, not pricing. Skip the split and you will optimize the wrong page for a quarter.

Freemium vs free trial vs demo-led: pick the model that fits the math

Freemium

Freemium converts 2–5% of signups, so it only works when the other numbers are enormous or free: a huge addressable top-of-funnel, near-zero marginal cost per free user, and ideally free users who do marketing for you through sharing or collaboration. Notion and Figma can run freemium; a niche tool with 8,000 visits a month cannot. At 3% conversion you need 33 signups to mint one customer.

Opt-in free trial

The default answer for most self-serve SaaS. A 14-day trial with no card converts 15–25% of signups and keeps the top of the funnel wide. It fits products whose value is demonstrable inside two weeks. If your product needs 30 days of accumulated data before it gets interesting, the trial clock is your enemy — fix time-to-value or pick another model.

Opt-out trial (card up front)

The card wall roughly triples the conversion rate and cuts signups by half to three quarters. Run the math on 1,000 visitors at a 4% signup rate: no-card gets 40 signups × 20% = 8 customers; card-required gets maybe 14 signups × 50% = 7. Volume is nearly a wash. What differs is quality — card-entering users are pre-qualified and churn less — and risk: opt-out conversions include forgotten cancellations, which come back as refunds, chargebacks, and month-1 churn. Choose it when you trust your onboarding, not to flatter a conversion slide.

Demo-led

When ACV is above roughly $5,000, self-serve checkout stops carrying the load and a demo funnel takes over: visit → demo request (1–3% of visits) → demo held (60–70% of requests) → closed-won (20–30% of held demos). Slower and more expensive per deal, justified by contract size.

The hybrid: reverse trial

A reverse trial starts every signup in the full product for 14 days, then downgrades non-payers to a free tier instead of locking them out. You get trial urgency plus a freemium pool that keeps nurturing — the users who downgrade remain reachable and convert later at rates well above cold email. Teams moving from pure freemium to reverse trial commonly report 1.5–2x paid conversion. It is the best default for products with a viable free tier and a clear premium hook.

Every channel runs its own funnel

The single most expensive modeling mistake in acquisition: one blended conversion rate across channels. Channels deliver different intent, and intent shows up twice — once in conversion, again in retention. The cohort a channel produces keeps behaving like that channel long after the click.

Channel	Visit→signup	Signup→paid (opt-in trial)	Monthly churn of resulting cohort
Organic search	5–8%	18–25%	2–4%
Referral	8–12%	20–30%	2–3%
Paid search (non-brand)	2–5%	12–20%	4–6%
Paid social	1–3%	8–15%	5–8%

Notice the pattern: the channels that convert best also retain best. An organic visitor searched for the problem you solve; a paid-social visitor was interrupted mid-scroll. That difference persists for the life of the cohort, which is why channel mix belongs in your churn model, not just your acquisition model — the mechanics of measuring it are covered in our guide to SaaS churn rate.

Blended numbers hide all of this. A company with great organic conversion and terrible paid conversion shows a respectable average — right up until it scales paid spend and the average collapses. Averages don't scale; channels do.

Time lags: when signups become cash

Conversion rates say how many. Lags say when, and "when" is what your cash flow runs on.

Visit→signup is effectively instant: 80–90% of signups happen in the first session or within 24 hours. Signup→paid is not. On a 14-day opt-in trial, conversions cluster on days 14–17 as trials expire, with a small early spike from annual-plan buyers. Freemium is far slower — upgrades spread over 30–90+ days, with a long tail that never fully closes.

The lag matters twice. For cash forecasting: dollars spent on ads in June produce subscription revenue in July, so a model that books conversions in the spend month overstates near-term cash by a full trial cycle. For measurement: dividing June's ad spend by June's new customers mixes cohorts and produces a CAC that is simply wrong — the same-period attribution error covered in CAC explained. Lag the denominator by your median time-to-convert, or compute CAC per cohort.

Worked example: Inkwell's funnel, channel by channel

Inkwell is a fictional self-serve writing app: $20/month, 85% gross margin ($17 monthly contribution per user), 14-day opt-in trial. Two channels.

	Organic search	Paid social	Total
Monthly visits	40,000	60,000	100,000
Visit→signup	5.0%	2.5%	3.5% blended
Signups	2,000	1,500	3,500
Signup→paid	22%	12%	17.7% blended
New paid users	440	180	620
Monthly channel spend	$8,000 (content)	$30,000 (ads)	$38,000
CAC	$18	$167	$61 blended

So far paid social looks acceptable: $167 CAC against a $17 monthly contribution is a 9.8-month payback, inside the under-12-months bar for self-serve. Now add retention. Inkwell's organic cohorts churn at 3% per month; paid-social cohorts churn at 6%.

• Organic LTV = $17 ÷ 0.03 = $567 → LTV:CAC = 31:1
• Paid social LTV = $17 ÷ 0.06 = $283 → LTV:CAC = 1.7:1
• Blended ≈ $485 LTV against $61 CAC ≈ 8:1

The blended 8:1 looks spectacular. The channel view shows paid social sitting below the 3:1 floor — and at 6% monthly churn, 0.94^10 ≈ 0.54, meaning nearly half the paid-social cohort is gone before the channel even pays back. Every incremental dollar Inkwell shifts into paid social makes the business worse while the blended dashboard keeps smiling.

The 1-point improvement that beats a 20% traffic budget

Inkwell wants more customers and has two options on the table.

Option A: fix onboarding, gain 1 point of signup→paid on both channels. Organic goes to 23%, paid social to 13%. New paid users: 460 + 195 = 655, a gain of 35 per month. Marginal cost: one-time engineering work. Blended CAC falls to $58.

Option B: buy 20% more paid traffic. Ad spend rises $6,000 to $36,000; 72,000 visits produce 1,800 signups and 216 paid users, a gain of 36 per month. Blended CAC rises to $67, and every incremental user comes from the channel that churns at 6% and sits at 1.7:1.

	Baseline	A: +1pt signup→paid	B: +20% paid traffic
New paid users / month	620	655	656
Extra monthly spend	—	$0	$6,000
Blended CAC	$61	$58	$67
Churn profile of added users	—	Mirrors existing mix	6%/month cohort

Same growth either way — but Option B costs $72,000 a year, repurchased every month forever, while Option A is free after the build and compounds with every future visitor from any channel. Conversion improvements are assets; traffic purchases are expenses. That is the asymmetry a funnel-first model makes visible and a growth-rate model cannot.

For completeness, the biggest single lever in Inkwell's funnel is upstream: lifting organic visit→signup from 5% to 6% adds 400 signups and 88 paid users a month — a 14% jump in new customers, at organic retention. Sensitivity analysis, stage by stage, is how you find these.

How to improve each stage

Visit→signup: message match

The ad's promise and the landing page's headline must be the same sentence. Dedicated landing pages per campaign, one call to action, no navigation bar, load under 2.5 seconds. Matched pages routinely convert 2x a generic homepage.

Signup→activation: shrink time-to-value

Find the one action that separates retained users from churned ones, then rebuild onboarding so a new user completes it in the first session. Cut every signup field you can; each extra field costs measurable conversion. Products that move activation from day 3 to minute 10 see trial conversion move by points, not decimals.

Activation→paid: design the trial

Shorter usually wins — 14 days beats 30 for most products because urgency drives engagement, and users decide in the first week anyway. Send a day-before-expiry email with a clear annual offer; let users extend once on request rather than lose them silently.

Pricing page: three plans, one anchor

Three tiers with a highlighted recommendation outperform both a single plan and a five-plan grid. Show annual pricing with the discount stated in dollars, and answer the top three objections directly on the page.

Checkout: remove friction, add wallets

Every form field is a toll. Apple Pay and Google Pay typically lift checkout completion 5–10%; local payment methods matter more the further you sell from the US.

Payment: recover involuntary failures

5–10% of first charges fail for boring reasons — insufficient funds, expired cards, bank declines. Smart retries on a decline-code schedule and a card-updater service recover a third to half of them. This is the cheapest conversion improvement in the entire funnel because the customer already said yes.

Five mistakes that corrupt funnel math

1. One blended rate across channels. Inkwell's blended 17.7% signup→paid describes neither channel — organic converts at 22%, paid social at 12%. Scale either channel and the blended forecast is wrong on day one.
2. Counting signups as customers. Signups are inventory, not revenue. A model (or a pitch) that quietly treats registrations as paying users overstates everything downstream by 4–6x at trial-conversion rates.
3. Ignoring activation. Treating signup→paid as one opaque number means you cannot tell an onboarding problem from a pricing problem, so you fix neither.
4. Optimizing the top while the bottom leaks. Doubling traffic into a 1.5% checkout completion doubles your ad bill and your disappointment. Fix stages bottom-up; every downstream point multiplies all upstream spend.
5. Baselining on your best month. One good month is weather. Use trailing 3–6 month medians for model assumptions, and treat a Product Hunt spike as an outlier, not a baseline.

How funnel assumptions feed your financial model

In a cohort-based model, the funnel is the front door of the entire revenue engine. Each month, each channel produces a cohort:

Cohort inflow (channel, month) = Visits × Visit→Signup × Signup→Paid, shifted by the conversion lag

That inflow then flows through the channel's retention curve to become MRR, month by month — the full mechanics are in our guide to cohort-based revenue forecasting. Because the stages multiply, ARR sensitivity is easy to state: a 10% relative improvement at any single stage lifts steady-state ARR by roughly 10%. But percentage points are not equal across stages — going from 2.5% to 3.5% visit→signup is a 40% relative gain, while 17% to 18% at signup→paid is a 6% gain. Run the sensitivity per stage and per channel, and the model tells you where a week of work buys the most ARR.

This is exactly how CashQuil is built: traffic channels feed funnels (visits → registrations → paid), each cohort carries its own retention curve, and the output rolls up to MRR, P&L, cash flow, and payback — with scenario and sensitivity analysis on any assumption. The complete guide to CashQuil walks through the whole chain.

Next steps

1. Instrument all five stages per channel — visits, signups, activation, paid, retained — even if the tracking is a spreadsheet.
2. Compute your channel-level LTV:CAC the way Inkwell did; find out whether a blended average is hiding a losing channel.
3. Decompose signup→paid into activation × activated-to-paid and fix whichever number is embarrassing.
4. Rebuild your revenue forecast as traffic × funnel × retention, with conversion lags, and retire the MRR-growth-rate cell.
5. Run a 1-point sensitivity on each stage and rank the results before planning next quarter's roadmap.

CashQuil does this out of the box — channel-level funnels, cohort retention, CAC and payback per channel, sensitivity analysis, and XLSX export — with a 3-day free trial.

Start free →

---