App marketing has an unusually seductive vanity metric. Installs can be bought cheaply and in volume, they look like progress on any chart, and a large share of them are gone within a week. The only number that decides whether the spending worked is what it costs to acquire a user who is still there and still paying after the period your economics actually depend on, and agreeing how that is measured is the whole of the negotiation.
Agree the metric before the budget
Cost per install is the number most proposals lead with because it is the number that looks best. Insist instead on cost per retained user at a horizon that matches your product, day seven for a habit app, day thirty for a subscription, and on the revenue that user produces. An agency comfortable being measured that way will say so immediately. One that steers the conversation back to install volume is telling you what its media buying is good at.
The store listing is conversion, and it is measurable
Everything spent driving traffic passes through a store page, and the icon, the first two screenshots, the opening line of the description and the rating are what decide whether a visit becomes an install. Both major stores support testing variants of that page. Improving the listing multiplies the return on every other channel at once, which makes it the cheapest work available and the thing most often left until after the media budget is spent.
Attribution on mobile is a design decision now
Platform privacy changes mean you cannot assume per-user tracking across apps, and the measurement framework, the consent prompt and the aggregated reporting you rely on all have to be chosen deliberately. Ask a prospective agency what they use, what its known blind spots are, and how they would run a holdout or geographic test when attribution is ambiguous. Any answer that implies clean deterministic attribution is either out of date or not being honest.
Organic and paid are not separate teams
Paid installs feed the ranking signals that produce organic installs, and a well-ranked listing lowers the effective cost of the paid ones. Treating store optimisation and media buying as separate purchases from separate suppliers is how each ends up optimising something the other undoes. If you buy them separately, at least put both on the same retention metric so they are pulling the same way.
Cheap installs are usually the expensive option
Incentivised traffic, misleading placements and low-quality networks deliver on any install target you set and produce users who never open the app again. The damage is not only the wasted spend: it corrupts the data every subsequent decision is made from, and it can distort the store signals you were trying to build. Ask explicitly which networks and placements will be used, and require the right to exclude any of them.
Questions people ask about app marketing services
What does app marketing cost?
The agency fee is the small part; media is the large one, and what it buys depends entirely on your category's competition and what a user is worth to you. Start from what you can afford to pay for a retained paying user and work backwards, rather than accepting a budget figure and hoping.
Is app store optimisation enough on its own?
For a product with existing demand and an established category, sometimes. For a new product nobody is searching for, it improves the conversion of whatever traffic you create but cannot create the traffic, so it works as a multiplier rather than a source.
How long before we can judge a campaign?
Long enough for a cohort to reach your retention horizon plus enough volume to be meaningful, which for most products is six to twelve weeks. Judging at two weeks measures installs, which is the number you specifically decided not to be judged on.
Should the agency have access to our analytics?
Yes, and to the revenue data, or they are optimising blind towards installs by default. Set the accounts up under your ownership and add the agency, so the history stays with you when the relationship ends.