Four parts of one job, held by one team.

The daily work, performed in your systems, under your controls, on your reporting cadence. Set up from scratch where nothing exists yet, or taken on from wherever you have reached.

• Promotion and incentive campaigns configured, reviewed and seed-tested before release
• Segments and journeys built and maintained rather than rebuilt each cycle
• Payout and verification queues worked down daily, not counted and reported
• Consent, eligibility and exclusion flags applied at send time
• Campaign, promotion and landing pages published to schedule, with terms kept current per market
• The weekly management pack, reconciled to platform figures

The half a support provider cannot deliver and will not commit to. The integration that keeps failing is repaired by the people who observed it fail, rather than queued behind eleven tickets on another supplier's roadmap.

• The platform to CRM event feed, including the transaction and promotion events most exports omit
• Reconciliation across platform, payment providers and finance
• Duplicate identity and balance remediation following a migration
• One agreed set of figures for marketing and finance, in a single timezone
• Jira Service Management configured with request types, service levels and named owners
• Alerting on silent failure, so a stopped job is raised the same night

We automate the parts that are rules and leave judgement where it belongs. Under regulation every automated action must be explainable and reversible, so logging is built before the automation is enabled rather than after it is questioned.

• The recurring report rebuilt by hand each week
• First-line triage and classification in the service desk, with escalation rules documented
• Drafted responses approved by a person, never an agent resolving a complaint alone
• Pattern detection for abuse and duplicate accounts, referred for human decision
• An agreed evaluation set before release, so behaviour can be demonstrated
• Every automated action logged with its input, its action and its reversal path

Activity is easy to report and hard to defend. We measure what the work changed, against customers it did not touch, in your data and with your definitions.

• A holdout group, a small random group left out of the new work for an agreed period
• Control groups on individual campaigns, counted on everyone picked, not only those who opened
• Net revenue over thirty days, after rewards and discounts, with a check for spend that only moved earlier
• Payback time for each month of new customers, before and after
• A monthly proof report, with every figure traceable to a query your team can run again