Pruning the Base

Pruning the Base

GoDaddy's customer count has fallen about 3% since its 2023 peak, to 20.4 million, even as revenue grew 16%. The evidence — retention near 85%, ARPU up from $203 to $242, bookings intact, and a high-spend cohort growing toward 10% of the base — points to a decline that is substantially deliberate: deep discounts cut, hosting divested, low-value products retired. The unfinished part is the net customer growth management promised for 2025, which arrived a year late.

The number that most moves the ten-year revenue case is not ARPU, which the moat chapter already traced (Moat and Durability); it is why the base underneath that ARPU is shrinking. A customer count that is falling because low-value accounts are being cleared out is a different investment than one falling because the franchise is eroding — the first strengthens the durability leg of the thesis, the second undermines it. This chapter diagnoses which one GoDaddy is.

What actually left

The base peaked at 21.0 million customers at the end of 2023 and has declined since: to 20.5 million in 2024 and 20.4 million in 2025, before ticking up 13,000 in the first quarter of 2026 [1] [2]. Over the same span ARPU rose without pause, from $203 to $242 and $246 on a trailing basis by the first quarter of 2026 [3] [4].

Source: total customers at period end, FY2023 and FY2025 Annual Reports (Form 10-K) and Q1 FY2026 Form 10-Q; on GoDaddy's current customer definition [5] [6] [7].

Two things matter about who left. Management's account is that the churn concentrated in the lowest-value tier — "primarily among one product customers, often low-cost domains" moving through their renewal cycle after product end-of-life migrations [8]. At the same time, the top of the base grew: the cohort of customers spending more than $500 a year rose to nearly 9% of the total by mid-2025 and about 10% by the first quarter of 2026, at what the company calls near-perfect retention [9] [10]. A base that loses single-product, low-dollar domain holders while adding $500-plus multi-product customers is trading down in headcount and up in value.

The case that the shrink is deliberate

The decline traces to a sequence of choices management began making in 2023 and has narrated consistently since. In 2023 it "took deliberate steps… to rationalize" a hosting portfolio that was not accretive to the model [11]. By the end of 2024 the CFO put the whole reset in one line: "This focus on quality over quantity has increased our model's resilience, profitability, and cash flow. By eliminating deep discounts, completing targeted divestitures, and migrating certain offerings, our customer base declined to 20.5 million" [12]. The work continued into 2026: the company retired another "lower-value product offering" in the first quarter, which "did not materially impact bookings" [13].

Sources: earnings-call commentary, Q4 FY2023, Q4 FY2024, Q2 FY2025 and Q1 FY2026 [14] [15] [16] [17].

Management's word is a lead, not evidence, so the account has to be tested against harder facts. Four independent lines corroborate it.

First, the framing predates the decline and sits in a document written for a regulator, not investors. In November 2022 — before the base turned down — GoDaddy told the SEC why it discloses customer count only annually: "While customer count is certainly a factor in our growth trends, it is not always a primary driver; go-to market and experimentation strategies we utilize at different periods in time may cause short-term changes in customer count that will not necessarily translate into sustained revenue growth." It added that "customers with lower-dollar contributions may be concentrated in a given quarter based on the timing of go-to-market promotions" [18]. A company that explicitly separated low-dollar accounts from durable revenue a year before shedding them is doing something it had already reasoned about, not reacting to attrition it did not foresee.

Second, the metrics that would deteriorate under genuine erosion did not. Blended retention held at roughly 85% through the decline, dipping only to about 84% in 2024 — the year of the heaviest divestitures and end-of-life work — and the customers on the integrated platform retained above 85% [19]. ARPU compounded throughout. Bookings, the leading edge of cash flow, were not materially dented by the product retirements [20]. Erosion of a franchise shows up as falling retention and softening cash collection; neither appeared.

Third, GoDaddy demonstrably retains the ability to add customers when it chooses to. Promotional offers run in late 2025 and early 2026 "generated significant gross customer adds — over 100,000 new customers from that set of promotions," and drove new domain registrations up 6% [21] [22]. The company then chose to "retire an older, lower-value product" in the same quarter, partially offsetting those adds — the CEO's words were that "these are deliberate portfolio choices" and the CFO's that the retired product had "only a slight impact on churn within that customer group" [23]. A firm that can turn a six-figure customer tap on and off is managing the count, not losing it.

Fourth, the domain base is confirming the turn. Domains under management, which fell from 83.6 million in 2023 to 80.8 million at the end of 2025, rose to 81.4 million in the first quarter of 2026 — an increase of about 600,000 in a single quarter, consistent with the promotional adds rather than with disintermediation [24] [25].

Sources: retention and cohort commentary, Q4 FY2024, Q2 FY2025 and Q1 FY2026 calls; domains from the Q1 FY2026 Form 10-Q [26] [27] [28] [29]; "over $500" cohort shown at its disclosed ~10% level.

What the deliberate read still owes

Calling the shrink deliberate does not make it costless, and two facts keep the diagnosis honest.

The first is a promise that slipped. Management guided a return to customer growth for 2025 — first in the third quarter of 2024, "with the previously mentioned divestiture and migration efforts behind us, we expect to return to customer growth in 2025" [30], repeated at year-end 2024, and again in mid-2025 as growth "later this year" [31]. The base still ended 2025 below where it ended 2024 (20.4 million against 20.5 million); the first net addition did not land until the first quarter of 2026, and it was 13,000 [32] [33]. A return to growth guided for 2025 that arrives, barely, in 2026 fits the pattern the credibility chapter documented — cash and margin targets met, growth targets set high and delivered late or low (Promise and Delivery).

The second is that some of the churn is genuine, even if concentrated. In mid-2025 the company acknowledged it was still seeing "residual pressure from migrations as they move through their initial renewal cycles," and that this fell "primarily among one product customers, often low-cost domains" [34]. These are real customers not renewing. The defensible claim is that they were low-value and that GoDaddy chose the migrations that surfaced them — not that no one is leaving.

There is also a structural limit worth naming. Pruning is a one-time reset, not a repeatable engine: a base can be cleaned of its lowest tier once, after which the arithmetic of growth reverts to either net customer additions or continued per-user gains. GoDaddy has spent three years trading unit growth for customer quality; from here, either the count resumes rising or the case is most sensitive to the ARPU lever the moat chapter flagged.

The read

The weight of the evidence is that the customer decline is substantially deliberate — a mix of eliminated discounts, hosting divestitures and low-value product retirements that cleared the lowest-dollar tier while retention held near 85%, ARPU compounded, bookings stayed intact, and the high-spend cohort grew. The SEC correspondence, written before the decline and to a regulator, is the strongest single piece of corroboration that this was a considered strategy rather than a rationalization of erosion. The strongest fact against the benign read is the return-to-growth guidance that management set for 2025 and did not meet until 2026, which shows the "clean base flips to net growth" step is real but running late.

This recasts the durability question the report has been circling. The residual doubt about whether revenue is higher a decade out (What the Price Implies) is not, on this evidence, that customers are fleeing a disintermediated product — the base is being managed, and both the count and the domain total turned up in early 2026. It is the narrower question of whether pricing power can keep carrying a base the company has chosen to keep lean. What would change the read: sustained net customer additions through 2026 alongside continued ARPU gains would confirm the deliberate account and the return to growth; a base that resumes falling while ARPU flattens would say the pruning story was covering for erosion after all.