The report you actually open on Monday.

Four things, in order: automated feeds from every source, agreed definitions for every metric, a small set of views built for real decisions, and the habits that keep them in use. The visualisation is the least difficult part. What takes the work is making cost, behaviour and closed revenue agree well enough that a single chart can carry them. A good engagement leaves you with pipelines you do not maintain by hand, a written metric dictionary, three or four reports people open without being asked, and a monthly review that changes what you do next.

Platform reports answer one question well: how did this platform perform, judged by its own attribution. They cannot tell you what the whole programme produced, because each platform counts conversions it believes it influenced, so the totals overlap. They also stop at the click and know nothing about what closed in your CRM. A dashboard built across sources gives you blended efficiency, contribution against plan and cost per outcome that finance recognises — while the platform views stay exactly where they are useful, for in-flight optimisation by whoever is running the campaigns.

Often Looker Studio with well-built connectors is genuinely enough, and we will say so rather than sell a warehouse. You need one when reports are slow because the visualisation layer is doing arithmetic across five live sources, when you want more than a year of history, when refunds and cancellations materially change the picture, or when revenue truth lives in a billing or CRM system that has to be joined properly. In those cases BigQuery or Snowflake with a modest transformation layer is the cheaper answer, because reconciling exports by hand costs more every month than the warehouse does.

Because they are built as artefacts rather than habits. BARC's research with 1,000 business intelligence users found the benefits are real — 97% report faster reporting and 94% better decisions — while average employee use sits at 25%, and IBM measures active use at roughly 29% of employees. Our answer is structural: fewer views, each with a named owner and a standing meeting; the first few of those meetings run from the dashboard with us in the room; walkthroughs recorded for new joiners; and anything nobody opened after a month pruned rather than maintained. Reporting that is part of a routine survives; reporting that is a link in an email does not.

Yes, and for most B2B and service businesses that is the whole point. Call tracking, booked appointments, CRM stages, quotes, signed contracts and invoiced revenue can all be joined back to the campaign that started them, which lets you report cost per qualified opportunity instead of cost per form fill. Where the cycle runs for months we report a leading indicator and the closed number side by side, with the cohort dated to when the spend happened, so nobody has to wait a quarter to make a decision or misread a slow month as a bad one.

Daily is standard, hourly where a decision depends on it, and every page carries a freshness stamp so nobody argues with a chart that has not updated. Platforms restate their own numbers and rename their own fields regularly, so the pipelines are built to backfill rather than overwrite, and we monitor for schema changes instead of discovering them in a meeting. When something does break, the alerting tells us before it tells you, and the fix is documented in the same place as the model.

Less, in our experience — the days spent every month pulling and reconciling exports are the first thing to disappear. We also build fewer reports than most clients expect, because three trusted views beat fifteen nobody checks, and each extra chart has to be maintained forever. What we do ask for is about an hour a week from someone who owns the numbers during the build, and their attendance at the first few meetings we run from the dashboard. That involvement is what makes the reporting stick after we hand it over.

Yes, and it is often the best split. We tend to own the marketing measurement layer — tags, events, platform APIs, marketing definitions — while your data team owns the warehouse and governance, with the transformation layer shared and reviewed like any other code. Everything we write is documented and version-controlled so it survives handover, and we keep the stack conventional rather than clever for exactly that reason. Where there is no in-house data capability, we run the layer and train whoever will inherit it.

The build is a fixed-scope project quoted up front after the reporting review, so you approve a number rather than an open-ended retainer: connections, metric dictionary, the core views, alerting, documentation and training. Ongoing support is a modest monthly fee scoped to the reviews, alert monitoring and changes you want, and it is optional — plenty of clients take the build, keep it running themselves, and come back when they add a channel. We never price reporting as a percentage of media spend; the two are unrelated.

We can, and many clients prefer it because the loop closes faster when the team reading the numbers is the team changing the campaigns. Our paid search, paid social and SEO teams work from the same warehouse and the same definitions. It is equally fine to buy reporting alone while other agencies buy your media: we report on their performance straight, share the methodology with them, and let the numbers do the arguing.