Metric trees for marketing teams

Marketing has more data than any other function. And less trust.

The gap between what you can measure and what you can prove is where marketing credibility dies. Every QBR, every budget review, every conversation with the CFO.

You optimise what you can measure, not what moves revenue

Cost per click, cost per lead and cost per MQL are on every dashboard. Whether those leads convert, expand and retain is not. Without a structured model connecting channel activity to pipeline and revenue, the team keeps optimising the top of the funnel while the bottom stays flat, and every metric that looks good in isolation gets reported as a win.

Attribution debates that consume every QBR

Last touch says paid search. First touch says content. Multi-touch says everything contributed. Meanwhile the CFO just wants to know whether the marketing budget is working. Another attribution model will not settle it. What settles it is a measured driver relationship between each channel and the outcomes it feeds, carrying a confidence level and statistical significance, tested against your data over time.

Campaign retros that run on self-reported success

The campaign launched, the impressions came in, the retro deck declares it a win. But nothing links the campaign to the metric it was meant to move, so nobody can say whether pipeline shifted because of it or despite it. Six months later the same play gets funded again on the same anecdotes.

A metric tree from revenue down to channel activity

Start at revenue and decompose downward. Revenue breaks into new business and expansion, new business into pipeline and win rate, pipeline into MQLs by channel, and each channel into the spend and activity beneath it. Describe your funnel in plain English and AI drafts the tree, then your team corrects it and the data takes over as judge. Every relationship is a directed driver edge carrying a confidence level and statistical significance, tested against your data nightly, so the connection from channel activity to revenue is proven in the data, period after period, not asserted in a deck.

The whole funnel sits on one canvas, from channel spend to revenue, instead of scattered across tools.
Every edge from channel activity to pipeline and revenue carries a confidence level and statistical significance.
You see which channels hold a statistically proven connection to revenue and which only look busy.
Your channel dashboards keep their job. The tree is the layer that shows what all that activity actually drives.

The tools marketing runs on, in the tree without an engineering ticket

Your warehouse covers the core numbers, but marketing runs on a long tail of tools that never make it there. Connect any MCP-compatible server as a data source, point KPI Tree at tools like HubSpot or Stripe, and turn what they return into tracked metrics without writing code. A suggestion button proposes the mapping from a sample response and a live preview shows the exact rows before you save. These metrics behave like any other: they sit in your trees, carry RACI ownership, drive targets and refresh on the schedule you set.

You add a connection with a server URL and a token, and KPI Tree discovers the tools the server offers.
You map a response to a metric by pointing at the date and value fields, with a live preview of the exact rows before you save.
Metrics from HubSpot or Stripe sit in the same tree as warehouse metrics, with the same ownership and targets.
Warehouse and semantic layer metrics sync alongside them, so one tree spans every source your funnel touches.

When pipeline moves, the why and the who arrive together

When a metric moves, the change insights waterfall decomposes the movement into what each driver added or removed, colour-coded, built on driver edges that carry confidence levels and statistical significance. And nobody has to be watching. Outlier and threshold triggers fire when the metric itself moves, and the push routes to the named Accountable owner within minutes via Slack or email with the driver context attached, escalating up the org chart when nothing happens.

A waterfall view shows exactly which drivers contributed to a movement in pipeline and by how much, each carrying its confidence level and statistical significance.
Outlier and threshold triggers fire the moment the metric moves, and the notification goes to the named Accountable owner.
Every metric in the funnel carries RACI ownership, so a slipping conversion rate lands with a person, not a dashboard.
When nobody acts, the escalation walks the live org chart automatically.

Every campaign tracked against the metric it was meant to move

When a campaign launches, the action is linked to the metric it targets, so success is defined before the work starts. The impact is then verified against the metric's subsequent movement, so the verdict is read from the data rather than claimed in the retro. Verified outcomes accumulate into a track record of which levers actually move pipeline and revenue, so next quarter's plan draws on proof instead of anecdotes. Budgets and targets flow through the same pipeline too, so plan versus reality on any marketing number is one view.

Each campaign action links to the metric it is meant to move, with a named owner attached.
Impact is read from how the metric moved after launch, not from the campaign team's own scorecard.
The verified history shows which campaign types moved pipeline and which only moved impressions.
Budgets and targets sit alongside actuals in the same tree, so plan versus reality needs no reconciliation.

“Your dashboards report the funnel. Your attribution model argues about credit. Neither shows what drives revenue, who should act when a number slips, or whether the last campaign worked.”

Channel dashboards describe the funnel. This closes the loop on it.

People change how they work when they can see the system, see their place in it, and see what moves when they do. A wall of channel dashboards does not create that understanding. A connected, statistically tested, owned metric tree does.

Proof that travels with the claim

Every relationship in the tree is a directed driver edge with a confidence level and statistical significance, tested against your data nightly and pruned by the people who know the business. When you tell the CFO a channel drives pipeline, the confidence and the significance arrive with the claim.

Answers in Slack, grounded in the tree

Mention KPI Tree in any Slack channel and the answer arrives in the thread, chart included, grounded in the same tree, the same driver edges and the same ownership as every other surface. AI agents connect to the same model through Canopy, KPI Tree's business context layer, so no tool in the stack gets a second version of the truth.

Pushes that fire on the metric, not the calendar

A scheduled Monday digest tells a channel what already happened. KPI Tree's push fires when the metric itself moves, on an outlier or a threshold cross, and routes to the named Accountable owner with the driver context attached, escalating up the org chart when nobody responds.

Common questions

How does KPI Tree handle marketing attribution?

KPI Tree does not replace your attribution model. It makes the debate less important. Instead of arguing about which channel gets credit for a single conversion, you see a directed driver edge between each channel and the outcomes it feeds, carrying a confidence level and statistical significance, tested against your data over time. Channels that consistently move pipeline and revenue show it in the evidence, and channels that do not stop absorbing budget on assertion.

Can we connect tools like HubSpot without engineering help?

Yes. Any MCP-compatible server can be added as a data source with a server URL and a token. KPI Tree discovers the tools the server offers, proposes a mapping from a sample response, and shows a live preview of the exact rows before you save. No code is involved, and the resulting metrics sit in your trees, carry ownership and drive targets like any other. Data that already lands in your warehouse connects the way it always has.

How is this different from the funnel reports in our marketing automation tool?

Funnel reports show conversion rates between stages. KPI Tree shows the system around them: how funnel metrics drive pipeline and revenue, with a confidence level and statistical significance on each relationship, who owns each metric, what reaches that owner the moment one moves, and whether the actions taken actually worked. It is the difference between reporting a funnel and running one.

Can the team use it without leaving Slack?

Yes. Mention KPI Tree in any Slack channel to ask how a metric is doing, what is driving a change, or who owns it, and the answer arrives in the thread, with a chart when the question is best answered visually. One-tap follow-ups let you dig deeper without retyping, every reply stays threaded, and a daily spend cap per workspace keeps AI cost predictable.

How do we know whether a campaign actually worked?

Every campaign action is tracked against the metric it was meant to move, and the impact is verified against the number itself. Two weeks later you are not reading a retro deck, you are reading the metric. Over time that verified history builds a track record of which campaign types move pipeline, which is what next quarter's budget conversation should run on.

How long does it take to set up?

Most marketing teams have a working metric tree within a few days. You describe your funnel in plain English, AI drafts the tree, and your team corrects it while you connect your data from the warehouse or over MCP. From there every driver relationship is tested against your data nightly, with confidence levels and statistical significance on every edge, so the tree keeps earning its place as the funnel changes.

Related guides