How did units break into new vs replacement vs repeat?
№ 123 · Unit sales waterfall
Definition
MetricUnit mix by demand source
Unit% of units shipped in period, split new vs replacement vs repeat
Period-start units sold, plus new channel sales, minus returns and replacements, end units. Build by classifying each transaction. Returns drag is a quality signal as much as a finance one. Example: 42K units start, +8K new, -1.2K returns, 48.8K end.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| New below 20% of units, replacement above 70%, repeat below 5% | New 35% to 45%, replacement 45% to 55%, repeat 5% to 12% | New buyers above 55% of units, replacement 35% to 40%, repeat 5% to 10% |
Consumer connected devices (smartphones, wearables, action cameras), global, 2024 to 2026. Reconstructed from installed base growth, shipment growth and observed replacement cycle length, because vendors do not disclose the split directly. · No located publisher reports a new vs replacement vs repeat unit split as a published benchmark. The ranges above are reconstructed: with a category replacement cycle of roughly 3.8 to 4.1 years, an installed base growing at low single digits mechanically produces a replacement-dominated mix. Assurant measures the age of devices physically turned in, IDC and Counterpoint measure shipments against installed base, and neither reports the waterfall directly. Treat these as orientation, not as measured quartiles, and instrument the split yourself through registration data.
Category split omitted: No two independent Tier 1 or Tier 2 datasets publish a new vs replacement vs repeat unit split by device category, so any category table here would be invented rather than sourced.
When it looks bad
The new-buyer bar shrinks every period while the total stays level because the replacement and trade-in bars grow to cover it, so the waterfall is harvesting the installed base rather than expanding it.
New 18% of units, replacement 74%, repeat 8%, total units flat at 42,000 for the fourth consecutive quarter, with new down from 31% two years earlier.
What to do about it
- Instrument the split at point of sale through serial registration and trade-in capture rather than modelling it. Average device age at turn-in is 3.81 to 3.83 years (Assurant, 2025 and Q1 2026), so a unit sold today should be expected back in the funnel roughly four years out, and the cohort has to be addressable when it is.
- Fund the replacement bar with a trade-in credit rather than blanket price discounting. US trade-in programmes returned $6.4bn to consumers in 2025, up 42% year on year, and promotion-heavy quarters measurably pulled iPhone turn-in age forward to 3.76 years in Q4 2025 (Assurant).
- Split the marketing budget by bar with separate targets and separate creative: prospecting against new, lifecycle messaging against replacement, and bundle or multi-device offers against repeat. A single blended CAC target hides the fact that replacement units cost a fraction of new ones to win.
- If the repeat bar sits below 5% of units, test a second-device use case (second household member, second location, gifting) before increasing prospecting spend, since repeat units carry no acquisition cost and land at full device margin.
Sources
- Assurant US consumers received over $6.4bn through trade-in programmes in 2025, up 42% year on year. In Q4 2025 the average age of iPhones turned in fell slightly to 3.76 years while Android devices passed four years on average. assurant.com ↗
- Assurant Average age of traded-in devices held at 3.81 years against 3.83 years for full-year 2025. Q1 2026 was the most active quarter on record by devices traded and value returned. assurant.com ↗
- Assurant For the first time in any smartphone category, the average age of Android devices at turn-in passed four years, reaching 4.1 years in Q4 2025. assurant.com ↗
- IDC Global wearable shipments grew 6.1% in 2024 to 538 million units, with growth expected to slow to 3.9% in 2025 as the US and India approach maturity. Slowing shipment growth against a growing base shifts mix toward replacement. my.idc.com ↗
- IDC Total wearable shipments forecast to grow at a 2.6% compound rate from 2026 to 2030 to 693.2 million units, with hearables sustaining repeat-purchase cycles and smartwatches forecast to decline 2.8% in 2026. idc.com ↗
- US SEC (GoPro, Inc.) Camera units shipped fell to 1.826 million in 2025 from 2.431 million in 2024 and 2.984 million in 2023, a 25% and 19% decline respectively, while ASP rose 8% to $357. sec.gov ↗
- Axis Intelligence, citing CIRP and Apple SEC filings Apple reported an installed base above 2.5 billion active devices as of the quarter ended 27 December 2025. CIRP put iPhone platform loyalty at 87% in the March 2026 quarter, with 12% of new iPhone buyers arriving from Android. axis-intelligence.com ↗
- Counterpoint Research, reported via BigGo Finance The global average replacement cycle peaked at 43.4 months in 2023 and held near 43 months the following year, with 2025 estimated to remain in the 3.5 year range. finance.biggo.com ↗