Spend sits in one system, revenue in another, and churn in a third. Nobody in the business can say which channels and campaigns actually made money. We build that answer, and we put it in writing.
Blended CAC, blended ARPU, a single churn rate. Averaged together, the channel that funds the business and the channel that quietly drains it cancel each other out, and the report stops telling anyone what to do on Monday. The numbers are usually all present somewhere: in the billing platform, the acquisition reports, the carrier settlement files. They are just never joined on the same definition of a customer.
Illustrative example. Not a client figure.
See what we answerThese are commercial questions, not analytical ones. They get asked in board meetings and budget reviews, and they usually get answered with an educated guess because the underlying numbers live in systems that were never joined.
One customer base, one set of definitions, six questions asked of it. These are not separate products bolted together. They run on the same joined view of your customers, which is why the answers reconcile with each other instead of contradicting across decks.
Cost per acquisition is easy to report. Cost per profitable acquisition is the one that changes the media plan. We attach realised revenue back to the channel, campaign and placement that brought each customer in.
Waiting twelve months to find out whether a cohort was worth buying is not a strategy. We project the value of customers acquired this month from the behaviour they have already shown.
Every acquisition is a loan to yourself. Payback timing decides how fast you can scale spend, and how much runway a campaign really needs before you judge it.
Aggregator reports, carrier settlements and ad platforms each count a conversion differently. We reconcile them onto one definition, so channel performance can be compared instead of argued about.
Not every cancellation deserves a save offer. We rank why customers leave, which causes the evidence says you can move, and which cohorts carry enough value to be worth the intervention.
Every figure traces back to the behaviour behind it, in language that survives contact with a board. No scores your team is asked to trust without explanation.
Revenue forecasts are cheap to produce and almost never scored afterwards. So the first thing we hand over is not a dashboard. It is a number we can be held to.
We commit the forecast to a hashed, timestamped file before you act on any of it. Nobody edits it after that point, including us.
We state the accuracy we expect, in writing, before a single result comes in.
Thirty days later we open the seal against what actually happened. Live. No retrofitting, no quiet edits.
If it earns it, standing monthly reporting starts. If not, you keep the report.
Illustrative example. Actual accuracy is declared and verified against your own data.
They visualise what each system already holds, on that system's own definition of a customer, a conversion and a month. Quick to read, and silent on the question the budget actually turns on: which spend produced customers worth having.
We reconcile those systems onto one definition, answer the commercial question directly, and say plainly when your data will not carry the conclusion. You get a decision with its reasoning attached, not another tile to interpret.
Mobile VAS, carrier billing and digital content, from the inside rather than from a case study. We have owned the CRM, churn and lifetime-value models in a major operator’s value-added services business, chased revenue that the aggregator report and the settlement file disagreed about, and defended a payback number to a commercial director who had every reason to doubt it. The analytical methods are on everyone’s shelf. Knowing how a carrier-billed subscription actually behaves in the data, and where that data lies to you, is the part that is hard to hire.
About UsDesign Partners get the diagnostic run on their own data at first-mover terms, with the sealed forecast as the next step. Send an extract and know within a week which of these questions your data can already answer, and which ones it cannot yet.