Playbooks

How the campaigns are actually built

Most case-study pages are a number and a logo. These are the working structures we use per category — what usually goes wrong, what we do about it, and the metric we agree to be judged on.

Overview

What these playbooks are

A playbook is the pre-agreed structure of a campaign: the billable event, the traffic mix, the guardrails and the success metric, written before launch so nobody argues about definitions after the first invoice. Six categories are documented below — apps and gaming, lending, e-commerce and COD, travel, edtech and subscriptions.

Each one exists because the category has a known failure mode — retention collapse, unreachable leads, RTO losses, refund-window churn — and the structure is designed against that failure, not against a generic funnel.

At a glance

Format: Situation → approach → committed metric
Categories: Apps, lending, e-commerce, travel, edtech, OTT
Named results: Shared under NDA with reference calls
Test phase: 2 weeks, capped, 300–500 conversions
Common rules: Downstream payouts, sub-IDs, caps, reconciliation
Start: Send your category for the applicable playbook

A note on numbers. Named advertiser results and reference calls are shared on request under NDA, not published as decoration. What follows is the method — which is the part you can actually evaluate before signing anything.

Apps & gamingCPI → CPA

Paying for installs, keeping the users

Situation. An install-priced campaign delivers volume, then day-7 retention collapses and the real cost per active user is triple the plan.

Approach. Move the billable event past the install. We price on registration or a first meaningful action, cap each sub-publisher, and read retention cohorts at day 1, 7 and 30 inside the advertiser’s MMP rather than our own log.

What we hold ourselves to. Retention of network-sourced users tracked against the advertiser’s blended baseline, reviewed weekly, with underperforming sub-publishers cut before the next payout cycle.

MMP integrationSub-ID capsCohort reads

Lending & cardsCPL → CPA

Leads the sales floor will actually call

Situation. Lead volume looks healthy, the contact centre reports half the numbers are unreachable, and nobody can agree whose fault it is.

Approach. Validation before delivery: syntax and carrier checks, network-wide de-duplication, geo and income-band pre-qualification in the form itself, and consent text stored with each record. Where the advertiser allows it, we move the payout to a downstream event such as application started or KYC completed.

What we hold ourselves to. Contact rate and downstream approval rate, reconciled monthly against the advertiser’s CRM, with rejected leads credited rather than argued over.

Consent trailLive validationCRM reconciliation

E-commerce & D2CCPS

Cash-on-delivery without the return shock

Situation. COD orders convert well and then 30–40% never get delivered, so a healthy-looking CPS is quietly unprofitable.

Approach. Payouts structured on delivered orders rather than placed orders, with a reversal window matched to the advertiser’s logistics cycle. Publisher mix skewed toward content and comparison traffic where intent is higher, away from placements that inflate order counts.

What we hold ourselves to. Delivered-order rate by publisher, return-adjusted CPS, and a monthly reconciliation against the advertiser’s order database.

Delivered-order payoutReturn clawbackPublisher mix

TravelCPS

Booking windows longer than the attribution window

Situation. Travel research happens weeks before the booking, so short cookie windows credit the last touch and starve the partners who actually created demand.

Approach. Longer agreed attribution windows with click ID passthrough rather than cookie reliance, and separate payouts for research-stage content partners versus closing partners, so both keep promoting.

What we hold ourselves to. Cancellation-adjusted revenue per partner, with cancellations reversed inside the stated window and nothing clawed back after it.

Long windowsClick IDCancellation handling

EdtechCPL → CPA

Enquiries versus counselling calls completed

Situation. An enquiry is cheap and nearly meaningless. Counsellor time is the scarce resource, and it gets burned on leads that were never going to enrol.

Approach. Pre-qualification inside the lead form — course intent, city, budget band — plus a payout that sits on counselling-call-completed rather than form-submitted. Publishers are briefed on the student profile, not just the offer.

What we hold ourselves to. Call-connect rate and cost per completed counselling session, reported weekly against the advertiser’s own dialler data.

Pre-qualificationDownstream payoutWeekly reads

Subscriptions & OTTCPA

Trial starts that survive the refund window

Situation. Trial-start payouts reward volume; a share of those trials cancel inside the refund window and the acquisition cost lands on the wrong side of LTV.

Approach. Billing on the paid conversion or on trial-day-N survival, whichever the advertiser can measure cleanly, with the refund window written into the reversal terms up front.

What we hold ourselves to. Trial-to-paid conversion rate by publisher, and cost per retained subscriber at day 30 rather than cost per trial.

Refund windowPaid-conversion payoutLTV alignment

Common thread

Four things every playbook has in common

The payout sits downstream

As close to the event that makes the advertiser money as their measurement allows. Everything upstream of that is a proxy.

Sub-publishers are visible

Quality is a publisher-level property. Without sub-IDs you can only turn the whole campaign off, which is why so many get turned off.

Caps come before scale

Every campaign starts capped. Raising a cap is a decision made on data; removing one is almost never a good idea.

Reconciliation is scheduled

Not triggered by a complaint. Monthly, against the advertiser’s system of record, with credits applied automatically.

Playbook questions

The working structure we apply to a category: the billable event we recommend, the traffic mix, the guardrails that prevent the category’s known failure modes, and the metric we agree to be judged on. It is the repeatable part of a campaign — the part you can evaluate before spending anything.

Because our advertiser agreements treat campaign data as confidential, and because logo walls prove nothing about what a network will do for you. Named results and reference calls are shared under NDA with qualified prospects instead — evidence for buyers, not decoration for visitors.

Ask through the contact page with your category. We connect you with an advertiser running something comparable, with their consent, usually within a few days.

Two weeks, a capped budget sized to 300–500 conversions, a limited publisher set, and a quality read at 48 hours, 96 hours and day 14. The decision at the end is scale, restructure or stop — made on your data, not our summary.

Most campaigns reach a steady state inside four to six weeks: two weeks of testing, then two to four weeks of scaling the publisher mix that survived the quality read. Categories with long approval funnels, such as lending, take longer to read and we say so up front.

Match the metric that makes you money: retained users → app growth; contactable buyers → lead generation; delivered revenue → e-commerce & sales; reach against a CPA → programmatic; audience trust → creators; partner scale → affiliate programme.

The structures travel; the parameters change. COD guardrails matter most in India and the Gulf, consent handling dominates in the UK and US, and payout levels shift with media prices per market. Every playbook is re-parameterised for the geo before launch.

Four things: the payout sits as far downstream as your measurement allows, sub-publisher IDs are mandatory, every campaign starts capped, and reconciliation is scheduled monthly rather than triggered by complaints.

If we can hold live supply and satisfy the category’s regulator, yes — we scope it in the proposal so you can see the structure before committing. If we cannot, we decline rather than experiment with your budget.

On the metric named in the playbook and written into the insertion order — delivered-order rate, contact rate, day-7 retention, cost per retained subscriber — reconciled against your system of record, not our dashboard. If the metric is not moving, the campaign changes or stops.

Next step

Want the version with names and numbers?

Reference calls and named campaign results are available under NDA. Tell us your category and we will put you in touch with an advertiser running something comparable.

Request references Browse solutions