applovin playbook

Tierra's canonical knowledge on AppLovin Ads. This is the entry point: read it whenever AppLovin (or Axon) comes up, then open the one reference file that fits the question. Everything deeper lives in references/, so this file stays short and the detail loads only when it's needed.

Where to go for what

Read the page that matches the question. On the wiki these are the pages in this section; through the Tierra Data connector, call read_playbook with the page key shown in parentheses.

Algorithm mechanics (algorithm-mechanics) — how the algorithm spends, sequences, and gets stuck; read before budget changes, pause decisions, or restructures.
BizDev and partnership (bizdev) — AppLovin's advertiser offer, referral program, matched credit, partnership standing, and client fit.
Catalog production (catalog-production) — building and debugging the catalog end card: feed types, fields, providers, pixel events, image and tile rules.
Compliance framework (compliance-framework) — producing and debugging creative compliance (vertical-agnostic).
Creative (creative) — building, iterating, and getting creative approved: video, the Interactive, the end card, compliance.
Cutdown workflow (cutdown-workflow) — re-editing longer-form source video into AppLovin-native cuts of 60 seconds or less.
Dev: API and automation (dev) — the Reporting API, the Campaign Management API, and how the data is ingested.
Interactive production (interactive-production) — building the Interactive (step 2): the Builder workflow, format choice, relaunch math, hero template.
Media buying (media-buying) — budgets, scaling, pausing, recovery, reading performance, and spend diagnostics during a live incident.
Parallel-campaign reset (parallel-campaign-tactic) — the force-spend tactic when the main campaign is locked onto a dominant creative and starves new assets.
Pixel install cookbook (pixel-install-cookbook) — event-by-event pixel specs, install-method decision rules, and a failure-mode debug catalog.
Relaunch playbook (relaunch-playbook) — planning a monthly relaunch cycle and bringing dormant winners back.
Reporting and tracking (reporting-and-tracking) — reading performance and instrumenting tracking: KPIs, attribution windows, dashboard-vs-API reads.
Strategy (strategy) — campaign architecture, audience strategy, scaling, and the relaunch-over-recover doctrine.
UTM templates (utm-templates) — the verbatim UTM strings, hierarchy rules, supported macros, and cutover timing.

If a question spans more than one, start with the closest fit and open others as needed. Don't guess at what a page says without reading it.

What AppLovin Ads is

AppLovin is a mobile technology company (based in Palo Alto, founded in 2012, public since its 2021 IPO, and added to the S&P 500 in 2025). AppLovin Ads is its ecommerce advertising platform. It was called Axon until mid-2026, when it was renamed AppLovin Ads. Plenty of clients and older documents still call it Axon, so treat the two names as the same thing.

The platform shows ads inside mobile apps: games like Candy Crush and Subway Surfers, and lifestyle and utility apps. Ads reach people while they're actively using their phones. There's no manual audience targeting; the platform's AI decides who sees which ad based on what's converting.

Scale, for context and for pitches:

Over 150 million daily active users in the US, and over 1.5 billion worldwide, across roughly 140,000 apps.
The fastest-growing ad channel for direct-to-consumer brands, with its ecommerce business alone past a billion-dollar annualized run rate.

Brief history: the ecommerce program ran as an invite-only pilot for large advertisers (roughly $30M+ in annual revenue) starting in late 2024, opened to new advertisers through referral codes in October 2025, and became broadly available under the AppLovin Ads name in mid-2026.

How Tierra works with AppLovin

Tierra is AppLovin's top ecommerce agency and is positioned to be the single largest AppLovin advertiser in the world. AppLovin is a fit for Tierra because it brings clients genuinely new customers rather than re-capturing traffic they already get from Meta or Google, and agency competition is still thin, so Tierra got in early.

The relationship itself is one of Tierra's biggest advantages, so protect it. The AppLovin team is deliberately lean: they teach Tierra best practices and trust Tierra to run the accounts, rather than staffing up to do that work themselves. Growth Strategists are the only people who contact the AppLovin reps directly. Everyone else (account managers, project managers, creative strategists) routes questions through Growth Strategy, which keeps the reps' time focused and the relationship healthy. Full detail on the offer and partnership is in bizdev.

How the platform works

Targeting is minimal by design. The only manual lever is geography. You can target by country (international is supported now, not US-only), by US state, and by US city (city-level is a newer addition and carries some IP-based imprecision). Budgets can be set globally or per country. The AI finds the audience by watching which creatives perform.
You can share your existing customer list (a file upload, the Shopify connection, or order-history data) to steer the model toward finding new customers who resemble your buyers.
The AI serves each shopper a personalized run of ads over time. A typical shopper sees roughly 8 to 10 of our ads before a first purchase, and sometimes many more.
Don't pause a weak-looking creative to kill it. Removing a creative that's part of an active run can break the whole sequence it belonged to, which usually costs more than it saves. When you need to move spend around, launch a parallel campaign to redirect it instead. (Details in media buying.)
If one creative set is taking more than half of a campaign's spend, read that as a structural signal (the model has settled into a rut or is running out of fresh audience), not necessarily proof the creative is a winner.
AppLovin and Meta share some patterns but aren't interchangeable. A creative that won on Meta (Facebook or Instagram) is worth trying here, but it won't automatically win. Test it, don't assume it. Some structures carry over; audience targeting, ad-set-level budgets, and fast pausing do not.

The three steps of an ad

Every ad plays three steps in a single impression:

Video: the opening video.
Interactive: a built page the shopper can tap through after the video (a quiz, carousel, image, testimonial, and so on). This has been called the "end card" in the past, which isn't strictly accurate; the end card is the next step.
End card: the closing screen. It shows the catalog (product recommendations) when a catalog is set up, or branding when there isn't one.

Building and testing interactives, end cards, and catalogs is covered in creative.

Terms to keep straight

Campaign: the container that holds creative sets, and where the main settings live: budget, the default landing-page URL and tracking parameters (both can be overridden on an individual creative set), the catalog, the optimization goal (ROAS, CPP, or the newer CPE, cost per event) and its target, the audience strategy (Universal, Prospecting, or Discovery), and the attribution day target (D0 or D7). How to choose among these is in strategy and reporting and tracking. Each campaign has its own ID.
Creative set: a container holding several videos and several interactives that all point to one destination URL. A set with 5 videos and 3 interactives gives the AI 5 × 3 = 15 combinations to work with. People loosely call a creative set an "ad." It is not the same as an "asset."
Asset: any single piece of media that goes into a creative set, meaning one video, one interactive, or one image. On an individual asset you can see impressions, clicks, click-through rate, and spend, and nothing else. Those numbers are totaled at the asset level, not broken out per creative set.
Node: Tierra's term for the single combination the AI actually spends against, one video paired with one interactive pointing at the set's URL. That set of 15 combinations above is 15 nodes. The end card is the shared closing step, not part of the combination. AppLovin doesn't show nodes as objects; they come from the math of the set.
Sequence: the personalized run of nodes the AI serves one shopper on the way to a purchase.
ROAS, CPP, and CPE: the optimization goals. ROAS is return on ad spend. CPP is cost per purchase. CPE is cost per event, a newer third goal. ROAS and CPP are the everyday choices.
Attribution window (D0, D1, D3, D7, D28): how many days after a click a conversion is counted. Exact day math is in reporting and tracking.

Brand Settings

Brand Settings, found in Account Settings, hold the logo, brand name, description, and call-to-action button. They fill step 3 (the end card) when no catalog is set up. The exact fields and limits are in creative.

The API and automation

AppLovin's API and its automation tools (including the MCP connection) are the same system as the campaign-management screen in the browser, just reached programmatically instead of by clicking. Developer detail is in dev.

Referral program

Tierra refers new advertisers to AppLovin and both sides get credit. New accounts get a matched credit (spend $5k, get $5k). Always use the current signup link so the referral is tracked:

Signup link: https://ads.applovin.com/auth/signup?referralCode=TIERRA
Referral code: TIERRA

Full offer details are in bizdev.

The AppLovin reps

Adam Foroughi: CEO and founder. Only Alex (Tierra's CEO) contacts him, and only for rare, high-stakes matters.
Cathy Sun: VP of Ecommerce. Runs the ecommerce division and press.
Gareth Domingo: Ecommerce Director and Tierra's agency lead. Bring him high-level strategy, new-business questions, and escalations, not day-to-day issues.
Julian Reiche: the current day-to-day rep. He handles the tactical account questions that come up during the work, and Tierra meets with him regularly.
Dan Sebanc: the previous day-to-day rep through October 2025.