Low CPI looks good in the dashboard, but it often says nothing about user acquisition performance. The only installs that matter are the ones that retain, convert and pay back your spend.
If you optimise only for cheap installs, you usually buy low-intent users in low-value placements. That can drag down retention, ROAS and long-term growth, even while CPI trends down and volume trends up.
What does "good" user acquisition performance actually mean?
User acquisition performance is not cost per install. It is the relationship between what you pay and the value those users generate over time.
For most mobile apps, real performance is a mix of:
- Payback time (how fast spend recoups via revenue)
- ROAS at key timeframes (D7, D30, D90, etc.)
- Retention and engagement (D1, D7, D30 retention, sessions per user)
- Monetisation behavior (ARPDAU, conversion to payers, ad ARPU)
CPI is a useful control metric, but it is a leading indicator, not the goal. Growth teams that treat CPI as the main KPI usually end up scaling campaigns that look efficient but destroy unit economics.
Why does low CPI often mean low intent users?
When you push hard to lower CPI, algorithms and networks respond by seeking cheaper supply. That usually means:
- Lower quality placements (cheap banners, remnant inventory)
- Lower-intent contexts (rewarded walls where the reward matters more than your app)
- Lower-income or misaligned geos and devices
- Users who tap by mistake or out of boredom
These users install because it costs them almost nothing in time or attention, not because they want your product.
Typical patterns when CPI drops aggressively:
- D1 retention falls below 25% for non-game apps or below 30% for many casual games
- Purchase conversion shrinks to a fraction of what you see from higher-intent channels
- Session length and frequency collapse
The campaign looks healthy on a cost chart, but you need more and more volume to hit the same revenue.
How should you compare CPI with downstream metrics?
CPI matters, but only in context. At minimum, read CPI alongside:
- D1, D7, D30 retention
- Conversion to payer or subscription
- Ad ARPU or IAP ARPU by D7 / D30
- ROAS curves (D3, D7, D14, D30, etc.)
A sanity check many UA teams use:
- If a new source is 30% cheaper CPI but D7 retention is down 40% vs your baseline, the trade is probably bad.
- If purchase rate or ARPU drops faster than CPI, you are losing.
You want sources where value per user declines more slowly than cost per user, or grows while cost remains stable.
Can a higher CPI campaign outperform on ROAS?
Yes. It happens all the time.
Example pattern:
- Campaign A: CPI £1, D30 ROAS 40%
- Campaign B: CPI £3, D30 ROAS 80%
Even though Campaign B costs 3x per install, it recovers double the ad spend in the same time window. If your goal is profitable scale, you would rather buy from B until marginal users stop paying back.
In practice, high-CPI campaigns often:
- Target higher-income or more competitive geos
- Run on premium video and CTV inventory
- Use refined audiences or better creatives
These users cost more, but they stay longer and spend more. For subscription and high LTV apps, this is usually where the real growth sits.
How do channels differ for user quality vs CPI?
Different channels naturally trade off CPI vs user value. You should plan for that instead of expecting every line item to hit the same CPI.
Here is a simplified view for mobile apps:
| Channel / inventory type | Typical CPI tendency | User intent / quality tendency | Common use case |
|---|---|---|---|
| Social self-attrib (Meta, etc.) | Low–medium, scalable | Mixed, can be very strong with signals | Always-on UA, prospecting and retargeting |
| Rewarded video in-app | Low–medium | Often low purchase intent, good ad revenue users | Volume for ad-monetised titles, top-of-funnel tests |
| OEM / preloads | Low–medium on paper | Very mixed, retention can be weak | Long-horizon install bases, utility apps |
| Programmatic display | Low–medium | Highly variable, needs strict controls | Incremental scale once measurement is solid |
| CTV / premium video | Higher CPI | Fewer installs, but strong engagement and spend | High-LTV apps, cross-device growth |
Treat this table as directional, not a rulebook. The point is that you should expect a CTV or premium video campaign to carry a higher CPI, but also to outperform on ARPU and retention if the targeting and creative are right.
Which post-install metrics matter most for UA decisions?
If you want to judge user acquisition performance instead of cheapness, track and optimise toward:
- D7 and D30 ROAS for IAP / hybrid games
- Subscription trial start rate and trial-to-paid conversion for subscription apps
- Ad ARPU and retained ad impressions for IAA-heavy games
- D7 and D30 retention for early-stage products without stable monetisation
You can run CPI-optimised campaigns while still steering toward these metrics if:
- Your MMP or analytics passes back purchase, revenue or retention signals to the network
- You use value-based bidding where available
- You build aggregated SKAN / AdAttributionKit conversion value schemas that reflect revenue bands or retention brackets
User acquisition performance improves when algorithms optimise for value, not installs.
How do you analyse cohorts to compare CPI vs value?
Cohort analysis is the only honest way to compare campaigns with very different CPIs.
Practical process:
- Group users by campaign, channel or creative and install date.
- Track revenue, sessions, retention and any key events against those cohorts over time.
- Compute LTV or ROAS curves per cohort (D0, D3, D7, D14, D30...).
- Divide LTV by CPI to see value per pound of spend.
You may find:
- A cheap Android OEM campaign that looks efficient on day 0 never climbs above 20% ROAS by D60.
- A more expensive iOS video campaign recovers 50–60% by D30 and keeps growing.
Once you see cohorts side by side, it becomes much harder to justify scaling only on CPI.
How should you allocate budget when cheap traffic underperforms?
If low-CPI traffic is dragging down performance, reallocate methodically.
- Rank all campaigns by D30 ROAS or LTV / CPI, not by CPI.
- Put more budget into the top quartile, even if their CPI is the highest.
- Cut or cap the bottom performers, even if their CPI is the lowest.
- Keep a small testing budget (for example, 10–20%) for new sources and creatives.
Many teams end up with:
- A "spine" of stable, high-ROAS campaigns with high or mid-range CPI
- A limited number of cheap CPI campaigns that are kept only if they deliver ad ARPU or strategic reach
- Test campaigns that are evaluated on D7 retention and very early ROAS before real scale
The rule: every pound should go to the highest expected value per install, not the lowest cost per install.
How do you pick bidding strategies that reflect user value?
Networks and DSPs will optimise to the goal you set. If you set CPI, expect them to chase CPI.
Better options when you care about user acquisition performance:
- Value-based bidding where possible (for example, target ROAS or value optimised campaigns)
- CPA on a qualified event (tutorial complete, registration, first order, day-2 open)
- Smart SKAN / AdAttributionKit schemas that encode value or retention, not just "install/no install"
On programmatic and OEM inventory, push for:
- Post-install data sharing with strict privacy controls
- Bidding strategies that use engagement or revenue signals when available, rather than pure CPI buying
You still monitor CPI to stay within guardrails, but your primary target shifts to value.
FAQ
What is user acquisition performance in mobile marketing?
User acquisition performance is how efficiently your paid campaigns turn budget into valuable users. It combines acquisition cost metrics like CPI with downstream outcomes such as retention, revenue and ROAS.
Why is CPI a weak primary KPI for UA?
CPI ignores everything that happens after the install. A campaign can hit a very low CPI while driving users who churn in a day and never pay or watch ads, which makes it a poor use of budget.
When is a higher CPI acceptable in user acquisition?
A higher CPI is acceptable when those users deliver better ROAS, LTV or retention. If a channel costs 2–3 times more per install but recovers spend faster and keeps monetising, it is usually the better buy.
Which post-install metrics should I track alongside CPI?
Track D1, D7 and D30 retention, conversion to payer or subscriber, ARPU at set timeframes, and ROAS curves. For ad-monetised apps, also track impressions per user and ad ARPU by cohort.
How often should I reevaluate channels with very low CPI?
Review low-CPI channels at least weekly on a cohort basis. If D7 or D30 ROAS and retention stay well below your baseline after a few cohorts, reduce spend or reframe their role to testing or reach instead of core performance growth.