ARPU: The Hidden SaaS Metric That Sets Your Pricing Ceiling

Most SaaS founders I work with between $5M and $15M Annu­al Recur­ring Rev­enue (ARR) can recite their ARR, their churn, and their CAC pay­back from mem­o­ry — but go blank when I ask what their ARPU is. ARPU (Aver­age Rev­enue Per User) is the aver­age recur­ring rev­enue a sin­gle user gen­er­ates over a peri­od, and it is the qui­et lever that decides whether your next 1,000 users make you rich­er or just busier. A com­pa­ny can dou­ble its user count and watch rev­enue grow more slow­ly than head­count, sup­port load, and infra­struc­ture cost — and the only met­ric that expos­es that trap ear­ly is ARPU. If you are not watch­ing it, you are fly­ing a growth plan blind to whether each new user is actu­al­ly worth acquir­ing.

Here is the part most peo­ple get back­wards. They treat ARPU as a van­i­ty num­ber to report, when it is real­ly a con­straint that gov­erns your entire go-to-mar­ket. Your ARPU sets the ceil­ing on what you can spend to acquire a user, how much human touch you can afford to give them, and which pric­ing mod­el is even viable. A $40/month ARPU and a $400/month ARPU are not the same busi­ness with dif­fer­ent dec­i­mal points — they are two com­plete­ly dif­fer­ent com­pa­nies that hap­pen to both call them­selves SaaS.

This guide cov­ers what ARPU actu­al­ly mea­sures, the crit­i­cal dis­tinc­tion between ARPU and ARPA that trips up most B2B founders, the exact for­mu­la and the mis­takes that cor­rupt it, a worked exam­ple you can hold against your own dash­board, the bench­mark ranges that mat­ter, and — most impor­tant­ly — the spe­cif­ic levers that raise ARPU with­out you hav­ing to find a sin­gle new cus­tomer.

What ARPU Actually Measures

Aver­age Rev­enue Per User (ARPU) is the aver­age recur­ring rev­enue gen­er­at­ed by a sin­gle user over a defined peri­od — usu­al­ly a month. It answers one blunt ques­tion: on aver­age, how much is each user worth to you per month?

The word “recur­ring” is doing the heavy lift­ing, exact­ly as it does every­where else in SaaS. ARPU is built from your sub­scrip­tion rev­enue — the pre­dictable, repeat­able rev­enue you can count on next month. One-time fees, imple­men­ta­tion charges, and pro­fes­sion­al ser­vices do not belong in it. If you let a $20,000 imple­men­ta­tion fee leak into the peri­od’s rev­enue, you will report an ARPU that describes the first invoice rather than the ongo­ing rela­tion­ship, and you will bad­ly over­state how valu­able each user real­ly is.

The word “user” is doing even more work — and this is where the met­ric gets gen­uine­ly tricky for B2B com­pa­nies. A “user” is a sin­gle seat, login, or indi­vid­ual per­son inside a cus­tomer orga­ni­za­tion. That is a dif­fer­ent unit from an “account,” which is the whole pay­ing com­pa­ny. When you sell to a 40-per­son com­pa­ny on a per-seat plan, that is one account and forty users. The dis­tinc­tion sounds pedan­tic until you real­ize it changes the met­ric by a fac­tor of forty — and changes which deci­sions the num­ber is good for. We will pull that apart in the next sec­tion, because get­ting it wrong is the sin­gle most com­mon ARPU mis­take in B2B SaaS.

For now, hold the core idea: ARPU iso­lates the eco­nom­ic val­ue of one unit of con­sump­tion (a user) so you can ask whether that unit is grow­ing more valu­able over time, hold­ing steady, or qui­et­ly shrink­ing while your user count climbs.

ARPU vs. ARPA: The Distinction That Actually Matters

This is the most impor­tant sec­tion in the arti­cle, so I am going to be pre­cise about it. In our house def­i­n­i­tions, the pri­ma­ry account-lev­el rev­enue met­ric is ARPA (Aver­age Rev­enue Per Account) — the aver­age recur­ring rev­enue per pay­ing account. ARPU (Aver­age Rev­enue Per User) is the per-user vari­ant you reach for specif­i­cal­ly when your billing unit is the indi­vid­ual user, such as a per-seat plan. They are sib­lings, not syn­onyms, and they answer dif­fer­ent ques­tions.

Metric	Denominator	The Question It Answers	Best For
ARPA (Average Revenue Per Account)	Total active accounts (paying companies)	How much is a typical customer worth per month?	B2B planning, sales economics, account-level CAC
ARPU (Average Revenue Per User)	Total active users (seats/individuals)	How much is a typical seat worth per month?	Per-seat pricing, B2C, expansion-by-user analysis

The rea­son the dis­tinc­tion is not aca­d­e­m­ic: in B2B, one account can con­tain many users, so the two met­rics can dif­fer by an order of mag­ni­tude or more. Con­sid­er a com­pa­ny billing 200 accounts that togeth­er hold 4,000 seats. If those accounts gen­er­ate $600,000 in Month­ly Recur­ring Rev­enue (MRR), the account-lev­el view (ARPA) is $600,000 ÷ 200 = $3,000 per account per month, while the user-lev­el view (ARPU) is $600,000 ÷ 4,000 = $150 per user per month. Same com­pa­ny, same rev­enue, same month — two num­bers that are 20× apart. Quote the wrong one in a board meet­ing and every down­stream esti­mate built on it (CAC bud­get, pay­back, expan­sion tar­gets) inher­its the error.

So which should you anchor on? Here is my rule:

• If you bill by the account (a flat plat­form fee, a tier the whole com­pa­ny shares), ARPA is your pri­ma­ry plan­ning met­ric. The account is the unit you acquire, the unit you sell to, and the unit that churns. Your CAC is spent per account, so the rev­enue you com­pare it against should also be per account.
• If you bill by the seat (price scales with the num­ber of users), ARPU becomes the sharp­er instru­ment, because each addi­tion­al user is a dis­crete unit of rev­enue and the path to growth runs through get­ting more seats live inside each account.
• Most B2B SaaS com­pa­nies between $5M and $15M ARR should watch both — ARPA to under­stand the eco­nom­ics of land­ing a cus­tomer, and ARPU to under­stand whether you are suc­cess­ful­ly expand­ing inside the accounts you already have. When ARPA is ris­ing but ARPU is flat, you are land­ing big­ger logos but not deep­en­ing seat pen­e­tra­tion. When ARPU is ris­ing but ARPA is flat, indi­vid­ual seats are get­ting more valu­able but your accounts are shrink­ing in head­count. Those are dif­fer­ent prob­lems with dif­fer­ent fix­es, and only watch­ing both met­rics tells them apart.

If you take one thing from this arti­cle, take this: ARPU and ARPA are not inter­change­able, and the choice of which to lead with is dic­tat­ed by your billing unit, not by which num­ber sounds big­ger. (For broad­er B2C-style prod­ucts billed straight to indi­vid­u­als, ARPU is sim­ply the nat­ur­al met­ric — there is no sep­a­rate “account” lay­er to wor­ry about.)

The ARPU Formula

There is one for­mu­la, and it is mer­ci­ful­ly sim­ple — which is exact­ly why peo­ple get slop­py with the inputs.

ARPU = Total Recur­ring Rev­enue in the Peri­od ÷ Total Active Users in the Peri­od

If you work in month­ly terms (the default), that becomes:

Month­ly ARPU = MRR ÷ Total Active Users

And the account-lev­el sib­ling, for con­trast, is the same shape with a dif­fer­ent denom­i­na­tor:

Month­ly ARPA = MRR ÷ Total Active Accounts

A few pre­ci­sion points that sep­a­rate a clean num­ber from a mis­lead­ing one:

1. Use recur­ring rev­enue in the numer­a­tor, not total rev­enue. Strip out one-time imple­men­ta­tion fees, set­up charges, and pro­fes­sion­al ser­vices. Those are real mon­ey, but they are not recur­ring, and fold­ing them in inflates ARPU and makes every ratio built on it — CAC pay­back, LTV/CAC — look health­i­er than it is.
2. Match the peri­od on both sides. Month­ly MRR over month­ly active users gives month­ly ARPU. If you want annu­al ARPU, use ARR in the numer­a­tor (ARR = MRR × 12) and keep the user count con­sis­tent. Do not divide a month­ly rev­enue fig­ure by an annu­al user count or you will pro­duce a num­ber that means noth­ing.
3. Define “active user” once and hold it. Is a deac­ti­vat­ed-but-still-paid-for seat a user? Is a free guest seat a user? There is no uni­ver­sal right answer, but there is a uni­ver­sal wrong move: chang­ing the def­i­n­i­tion between peri­ods so the trend line lies to you. Pick a def­i­n­i­tion, write it down, and lock it.
4. Decide paid-only vs. all-users delib­er­ate­ly. Some teams com­pute ARPU over pay­ing users only; oth­ers over all users includ­ing free-tier accounts. Pay­ing-user ARPU tells you how much you extract from a mon­e­tized user; all-user ARPU (some­times called ARPU blend­ed with free) tells you how well your free-to-paid fun­nel con­verts. Both are legit­i­mate — just label which one you are show­ing, because they can dif­fer wild­ly for a freemi­um prod­uct.

The for­mu­la is triv­ial. The dis­ci­pline around the inputs is what makes the met­ric trust­wor­thy.

A Worked Example: ARPU at an $8M ARR Company

Num­bers make this con­crete. Take a B2B SaaS com­pa­ny at $8M ARR — square­ly in the range where these deci­sions bite. That is about $666,667 in MRR ($8,000,000 ÷ 12). Sup­pose the com­pa­ny bills on a per-seat mod­el and cur­rent­ly has 250 pay­ing accounts hold­ing 5,500 active paid seats between them.

The two head­line num­bers fall straight out of the for­mu­las:

Month­ly ARPU = $666,667 ÷ 5,500 = $121.21 per user

Month­ly ARPA = $666,667 ÷ 250 = $2,666.67 per account

So the aver­age account pays about $2,667/month, and the aver­age seat inside those accounts is worth about $121/month. The ratio between them — $2,666.67 ÷ $121.21 = 22.0 — is just the aver­age num­ber of seats per account (5,500 ÷ 250 = 22.0). That ratio is itself a use­ful num­ber: it tells you, on aver­age, how deep you have pen­e­trat­ed each cus­tomer’s orga­ni­za­tion.

Now watch what the blend­ed ARPU hides. Sup­pose the book actu­al­ly breaks down like this:

Segment	Accounts	Paid Seats	MRR	ARPU (per seat)
SMB	180	1,440	$144,000	$100.00
Mid-market	60	2,400	$312,000	$130.00
Enterprise	10	1,660	$210,667	$126.91
Total	250	5,500	$666,667	$121.21

The blend­ed ARPU of $121.21 does not describe a sin­gle real user. SMB seats come in at $100, mid-mar­ket seats at $130, enter­prise seats at about $127. The vari­ance is not huge in this par­tic­u­lar book — but the seat counts per account are wild­ly dif­fer­ent: SMB aver­ages 8 seats per account (1,440 ÷ 180), mid-mar­ket 40 (2,400 ÷ 60), enter­prise 166 (1,660 ÷ 10). That means a 5% lift in seat-lev­el ARPU is worth far more in the enter­prise seg­ment, where each account car­ries 166 seats, than in SMB, where it car­ries 8. If you are going to invest in rais­ing ARPU, the seg­ment­ed view tells you where a point of improve­ment actu­al­ly moves the most mon­ey.

This is the habit that mat­ters most with ARPU: seg­ment it. Cal­cu­late it sep­a­rate­ly by deal size, plan tier, ver­ti­cal, and acqui­si­tion chan­nel. In my expe­ri­ence there are always mean­ing­ful vari­ances — and the blend­ed aver­age is the num­ber most like­ly to lead you astray pre­cise­ly because it looks so clean on a slide.

One more lay­er, to show how a mis­take prop­a­gates. Sup­pose the enter­prise seg­ment charges a one-time onboard­ing fee equal to 25% of its annu­al recur­ring val­ue, and a care­less ana­lyst lets that fee land in the month it was billed. Enter­prise MRR for that month would appear as $210,667 × 1.25 = $263,333, push­ing report­ed enter­prise ARPU to $263,333 ÷ 1,660 = $158.63 instead of the cor­rect $126.91 — a 25% over­state­ment that would tell you enter­prise seats are far more valu­able than they are, and tempt you to over­spend acquir­ing them.

ARPU Benchmarks — and Why the Range Is So Wide

Founders always want a sin­gle “good” ARPU num­ber. There isn’t one, and any­one who hands you a uni­ver­sal tar­get is sell­ing some­thing. ARPU is almost entire­ly a func­tion of who you sell to and how you price. A self-serve prod­uct billed to indi­vid­u­als and an enter­prise plat­form billed per seat can both be excel­lent busi­ness­es with ARPUs two orders of mag­ni­tude apart.

That said, rough bands are use­ful for san­i­ty-check­ing where you sit. Here is a direc­tion­al map for month­ly per-user ARPU by seg­ment:

Segment	Typical Monthly ARPU Range	What Drives It
Self-serve / individual (B2C-style)	~$20–$50	Low-touch, credit-card signup, single-user plans
SMB B2B	~$50–$200	Per-seat plans, small teams, light onboarding
Mid-market B2B	~$200–$2,000	Larger seat counts, tiered features, some sales touch
Enterprise B2B	~$2,000–$10,000+	High-value seats, premium tiers, full sales motion

A point on these num­bers: spe­cif­ic ARPU bands and bench­marks shift with mar­ket con­di­tions, pric­ing-mod­el trends, and how each source defines a “user.” The fig­ures here are illus­tra­tive and meant to show rel­a­tive dif­fer­ences between seg­ments — not fixed thresh­olds. Ver­i­fy cur­rent bench­marks against a recent source before you set tar­gets. The right ques­tion is nev­er “is my ARPU high?” in the abstract — it is “is my ARPU high enough to sup­port the way I acquire and serve users?” A $40 ARPU is out­stand­ing for a self-serve prod­uct with near-zero acqui­si­tion cost and dis­as­trous for a busi­ness that puts a sales­per­son on every deal.

Exter­nal bench­mark pub­li­ca­tions can help you locate your seg­ment. Pad­dle’s ARPU resource is a sol­id primer on the met­ric and its vari­ants, and Besse­mer Ven­ture Part­ners’ scal­ing guid­ance for cloud busi­ness­es puts per-user eco­nom­ics in the broad­er con­text of build­ing toward a nine-fig­ure out­come.

How ARPU Connects to Unit Economics and Valuation

ARPU is not a stand­alone van­i­ty stat. It feeds direct­ly into the unit eco­nom­ics that deter­mine whether you can scale and what your com­pa­ny is worth at exit. Trace the con­nec­tions and you will nev­er treat it as a report­ing after­thought again.

Start with Cus­tomer Acqui­si­tion Cost (CAC) pay­back. The ful­ly loaded cost to acquire a user has to be recov­ered out of the gross prof­it that user gen­er­ates — and ARPU is the top of that gross-prof­it stack. A high­er ARPU means each user throws off more month­ly gross prof­it, which short­ens pay­back and lets you spend more to acquire the next one. Con­crete­ly, with our $121.21 ARPU and a 75% gross mar­gin, each user con­tributes $121.21 × 0.75 = $90.91 of gross prof­it per month. If it costs $545 in ful­ly loaded sales and mar­ket­ing to acquire that user, the CAC pay­back peri­od is $545 ÷ $90.91 = 6.0 months — fast enough to recy­cle cap­i­tal aggres­sive­ly. Raise ARPU to $150 at the same mar­gin and the same $545 CAC now pays back in $545 ÷ ($150 × 0.75) = 4.8 months. You did not touch acqui­si­tion spend; you sim­ply made each user worth more, and the whole engine sped up.

It also feeds Life­time Val­ue (LTV). In its deci­sion-use­ful form, LTV = ARPU × Gross Mar­gin % × Aver­age Cus­tomer Lifes­pan, where aver­age lifes­pan is 1 ÷ month­ly churn. Take the same $121.21 ARPU, 75% mar­gin, and a 2% month­ly churn rate. Aver­age lifes­pan is 1 ÷ 0.02 = 50 months, so LTV per user is $121.21 × 0.75 × 50 = $4,545. Against a $545 CAC, that is an LTV/CAC ratio of $4,545 ÷ $545 = 8.3× — well above the 3.0× bench­mark for healthy unit eco­nom­ics, which sig­nals you can like­ly afford to invest more aggres­sive­ly in growth. (Note that 2% month­ly churn does not mean 24% annu­al churn; churn com­pounds, so annu­al churn is 1 − (1 − 0.02)¹² = 21.5%. Always com­pound it — nev­er mul­ti­ply by twelve.)

And at exit, ARPU shapes the con­ver­sa­tion in two ways. First, a ris­ing ARPU is direct evi­dence you have pric­ing pow­er and an expand­ing rela­tion­ship with each user — both of which read as durable, high-qual­i­ty recur­ring rev­enue, and qual­i­ty of rev­enue is one of the largest levers on the val­u­a­tion mul­ti­ple. Sec­ond, grow­ing ARPU lets you grow rev­enue faster than user count, which means faster growth with­out a pro­por­tion­al increase in sup­port and infra­struc­ture load — exact­ly the mar­gin-expan­sion sto­ry acquir­ers pay a pre­mi­um for. For the full pic­ture of how recur­ring rev­enue trans­lates into a sale price, see the guide on SaaS com­pa­ny val­u­a­tion.

The Five Mistakes That Distort ARPU

Each of these is a real mon­ey mis­take, not a round­ing error — and each one I have watched qui­et­ly cor­rupt a real com­pa­ny’s growth deci­sions.

1. Let­ting one-time fees leak into the numer­a­tor. Imple­men­ta­tion, set­up, and ser­vices fees inflate ARPU and make every down­stream ratio look health­i­er than it is. Use recur­ring rev­enue only; keep one-time fees out.
2. Con­fus­ing users with accounts. Report­ing ARPA and call­ing it ARPU (or vice ver­sa) is the car­di­nal B2B sin. The two can dif­fer by 20× or more. Decide which unit your billing mod­el makes pri­ma­ry, label it cor­rect­ly, and nev­er let the two num­bers get swapped in a deck.
3. Using one blend­ed ARPU for a mul­ti-seg­ment busi­ness. A com­pa­ny-wide aver­age hides the truth when your book spans self-serve and enter­prise. The blend­ed num­ber is use­less for decid­ing where to invest. Seg­ment by tier, deal size, ver­ti­cal, and chan­nel.
4. Mix­ing free and paid users with­out say­ing so. All-user ARPU and pay­ing-user ARPU answer dif­fer­ent ques­tions and can dif­fer enor­mous­ly for a freemi­um prod­uct. Pick one for the deci­sion at hand and label it; do not let the denom­i­na­tor silent­ly shift between reports.
5. Mis­match­ing the peri­od. Divid­ing month­ly rev­enue by an annu­al user count (or annu­al rev­enue by a month­ly count) pro­duces a mean­ing­less num­ber that nonethe­less looks author­i­ta­tive. Match the peri­od on both sides of the for­mu­la, every time.

How to Raise ARPU Without Finding New Customers

This is where ARPU stops being a met­ric and becomes a growth strat­e­gy. Rais­ing ARPU is one of the high­est-lever­age moves avail­able to a SaaS com­pa­ny between $5M and $15M ARR, because it grows rev­enue with­out the cost and risk of acquir­ing a sin­gle new logo. Every dol­lar of ARPU lift drops onto the user base you already have. Here are the levers, rough­ly in order of how reli­ably they work.

Raise prices on a base that won’t leave. The bluntest lever is also the most under-used. Most founders price below what the mar­ket will bear because they have nev­er test­ed oth­er­wise. If you can raise prices and keep your users, you have pric­ing pow­er, and a price increase flows almost entire­ly to ARPU and gross prof­it. Test it on new cus­tomers first, watch con­ver­sion, then roll it to renewals. Even a 10% increase that holds is a 10% ARPU lift you did not have to earn through acqui­si­tion.

Build a tiered struc­ture that pulls users up. A sin­gle flat price leaves mon­ey on the table from the users who would hap­pi­ly pay more for more. A good-bet­ter-best tier lad­der lets users self-select into high­er-val­ue plans as their needs grow, mechan­i­cal­ly rais­ing ARPU over time as your base matures into the upper tiers. The tiers should be built around the val­ue met­ric that scales with how much the cus­tomer gets out of the prod­uct — seats, usage, advanced fea­tures — so that grow­ing cus­tomers nat­u­ral­ly migrate upward.

Dri­ve seat expan­sion inside exist­ing accounts. On a per-seat mod­el, ARPU and account depth grow when you get more of each cus­tomer’s team actu­al­ly using the prod­uct. This is the lever where ARPU and net rev­enue reten­tion meet: a cus­tomer-suc­cess motion that dri­ves adop­tion across an orga­ni­za­tion adds seats, lifts ARPU, and push­es NRR above 100% — the thresh­old where your exist­ing base grows on its own with­out new acqui­si­tion. Watch ARPU and NRR togeth­er; ris­ing ARPU dri­ven by gen­uine expan­sion is one of the strongest sig­nals a SaaS busi­ness can show.

Add usage-based or con­sump­tion com­po­nents. Lay­er­ing a usage-based ele­ment on top of a sub­scrip­tion lets rev­enue grow as cus­tomers get more val­ue, with­out a rene­go­ti­a­tion every time. Done well, it ties your ARPU direct­ly to cus­tomer suc­cess: the more they use, the more they pay, the more com­mit­ted they become. Be pre­cise about which part of that rev­enue is gen­uine­ly recur­ring ver­sus vari­able when you report it.

Cross-sell adja­cent prod­ucts and pre­mi­um fea­tures. Once a cus­tomer trusts you, sell­ing them an addi­tion­al mod­ule or a pre­mi­um capa­bil­i­ty is far cheap­er than acquir­ing a new cus­tomer for that rev­enue. Each suc­cess­ful cross-sell rais­es the rev­enue per user with­out chang­ing the user count — a direct ARPU lift that also deep­ens switch­ing costs and makes you hard­er to dis­place.

The through-line across all five levers: ARPU growth is expan­sion-dri­ven growth, and expan­sion-dri­ven growth is the most cap­i­tal-effi­cient growth there is. Fix your reten­tion first — pour­ing effort into ARPU while users churn out the bot­tom is wast­ed motion — but once churn is under con­trol, rais­ing ARPU is very often the high­est-return project on the table.

Frequently Asked Questions About ARPU

What is ARPU in SaaS?

ARPU (Aver­age Rev­enue Per User) is the aver­age recur­ring rev­enue a sin­gle user gen­er­ates over a peri­od, usu­al­ly a month. You cal­cu­late it by divid­ing recur­ring rev­enue (MRR) by the num­ber of active users in that peri­od. It tells you how much eco­nom­ic val­ue each indi­vid­ual seat or user con­tributes, which in turn gov­erns how much you can afford to spend acquir­ing and serv­ing them.

What is the difference between ARPU and ARPA?

ARPU divides rev­enue by users (indi­vid­ual seats or peo­ple); ARPA (Aver­age Rev­enue Per Account) divides rev­enue by accounts (pay­ing com­pa­nies). In B2B, one account often con­tains many users, so the two num­bers can dif­fer by 20× or more. Use ARPA as your pri­ma­ry plan­ning met­ric when you bill per account, and ARPU when you bill per seat. Many B2B com­pa­nies watch both — ARPA for the eco­nom­ics of land­ing a cus­tomer, ARPU for expan­sion inside accounts.

How do you calculate ARPU?

Divide your total recur­ring rev­enue for the peri­od by the num­ber of active users in that peri­od: ARPU = MRR ÷ Total Active Users for a month­ly fig­ure. Use recur­ring rev­enue only (exclude one-time imple­men­ta­tion and ser­vices fees), match the peri­od on both sides of the for­mu­la, and hold your def­i­n­i­tion of “active user” con­stant between peri­ods so the trend is com­pa­ra­ble.

Should ARPU include free users?

It depends on the ques­tion you are answer­ing. Pay­ing-user ARPU (pay­ing users only in the denom­i­na­tor) tells you how much you extract from a mon­e­tized user. All-user ARPU (includ­ing free-tier users) tells you how well your free-to-paid fun­nel con­verts. Both are legit­i­mate — just label which one you are pre­sent­ing, because for a freemi­um prod­uct they can dif­fer dra­mat­i­cal­ly.

What is a good ARPU for a SaaS company?

There is no uni­ver­sal “good” ARPU — it is almost entire­ly a func­tion of who you sell to and how you price. Self-serve prod­ucts can thrive around $20–$50/month per user; enter­prise per-seat mod­els often run into the thou­sands. The right ques­tion is not whether your ARPU is high, but whether it is high enough to sup­port your acqui­si­tion and ser­vice mod­el. A num­ber that is excel­lent for a low-touch self-serve prod­uct would be ruinous for a sales-led enter­prise motion.

Does ARPU include one-time fees?

No. A clean ARPU cal­cu­la­tion uses recur­ring rev­enue only and excludes one-time fees — imple­men­ta­tion, set­up, onboard­ing, and pro­fes­sion­al ser­vices. Those are non-recur­ring, and fold­ing them into the numer­a­tor over­states how valu­able each user actu­al­ly is on an ongo­ing basis, which then dis­torts every ratio (CAC pay­back, LTV/CAC) built on top of ARPU.