You can run an agency relationship on a dashboard. Most do. The dashboard puts every signal in front of the client. The client scans the dashboard on Monday morning. The dashboard tells them what changed. The dashboard does not tell them what to do.
That is the gap. The work happens between the touchpoints. The visibility does not follow.
Next Best Action is three surfaces. One job each. None of them are a dashboard.
The dashboard problem
A dashboard is a snapshot of metrics on a refresh interval. It tells you Meta CPC went up 18% last week. It does not tell you whether to pause the audience, swap the creative, or accept the inflation because the LTV is up too.
Decision making is the job. Display is the side effect. Most marketing dashboards stop one step short of the answer.
We saw the gap inside our own client book first. Every Monday a strategist would log into HubSpot, GA4, Meta Ads, Google Ads, and the warehouse. The strategist would assemble the picture. The strategist would surface the items the marketing leader needed to decide on. The strategist would put those items in front of the leader, usually in a Slack thread or a status doc.
The strategist was doing three different jobs. Pulling the picture together. Surfacing the decisions. Presenting them in a way the leader could act on quickly. We named those three jobs and built three surfaces around them.
Status
The first surface tells you what is happening. What shipped this week. What is in flight. What is blocked. What is waiting on you. What we owe you next.
It is not a dashboard. The fact that Meta CPC is up 18% lives in here, but so does the next-step recommendation. The work scroll updates in real time. The Watch Tower component flags signals before you have to ask: a paid campaign is overspending, the lifecycle email rebuild shipped yesterday, the Q3 brief is blocked on your sign-off.
Open it Monday morning. The week so far is right there. The "send me an update" email goes to zero.
Ask
The second surface is for questions you would have asked the strategist on a phone call. Type or speak. The agent runs against your connected data: HubSpot, Salesforce, GA4, Google Ads, Meta, your warehouse, plus our playbook layer.
Show me ROI on May. Which campaign is bleeding budget right now. What is the LTV trend on the SaaS segment. The agent answers with the chart, the numbers, and the citations. Every claim has a source you can click. Click through to the query, the row count, the timestamp.
The strategist reviews high-stakes answers before they reach you. Routine ones go directly. The bar between "routine" and "high-stakes" is a setting in the workspace and it is the strategist's call, not the agent's.
Approve
The third surface is where decisions happen. Incoming work proposals from the agency team show up as cards. Each card carries what we recommend, what it costs, what we expect to come back, and a one-tap path to accept or decline.
Decline routes a comment back to the strategist. Accept routes the work into our queue. Read the memo, reply, schedule the next call. Decisions live where the work lives.
Three primitives. Each one replaces a slow analog ritual.
| Surface | What the strategist used to do | What the surface does now |
|---|---|---|
| Status | Compile the weekly status doc on Friday afternoon | Real-time work scroll, updated as events ship |
| Ask | Answer "what's going on with X" by hand on a Tuesday call | Direct query, sourced answer, strategist review on high-stakes |
| Approve | Send work proposals over email and chase the reply | One-tap card, decision logged, audit trail kept |
Why three surfaces and not one
The temptation when building a workflow product is to consolidate. One surface. Everything in one place.
Consolidation was the wrong choice for us. The three jobs require three different mental modes. Status is scanning. Ask is querying. Approve is deciding. A single surface that tries to be all three ends up doing none of them well.
The marketing leader who opens Status on Monday is not in deciding mode. They are in orienting mode. The leader who opens Ask in the middle of a budget conversation needs the query interface, not the work scroll. The leader who opens Approve on Thursday afternoon to clear the queue needs the decision cards, not the data layer.
Three surfaces. Three mental modes. One workflow.
What this looks like as a buyer
A typical marketing leader running NBA opens Status three times a day, runs an Ask query once or twice, and clears Approve twice a week. That cadence took six weeks to stabilize across the pilot deployments. The leader who opens Status the first week opens it twenty times. By week three it is three or four. By week six it is a habit.
The agency strategist on the other side of the surface is doing fundamentally different work. Less compiling. More judgment on what gets routed to Ask review. More time on the Approve queue itself. The strategist becomes the curator and the agent becomes the runner. The leverage stays in the strategist's hands.
That is the part we did not anticipate when we first sketched the three surfaces on a whiteboard in February. We thought we were building a client-facing product. We were building an agency-side reskill at the same time.
The CFO sits in the workspace at no charge
One detail that matters. Every approved card lives in an audit trail the CFO can read. Every Ask answer has source citations the CFO can click through to verify. Every dollar of agency spend is traceable to a card that was accepted, not just an invoice that was approved.
Procurement gets a clean record of what was authorized, when, by whom, and against which campaign. The financial layer becomes a viewing window over the work layer rather than a separate audit project at the end of the quarter.
The CFO seat is free in every workspace. Not because we are generous. Because the day the CFO finds out the spend is traceable per-decision, the renewal conversation becomes a different conversation.
What we will not build
We will not build a fourth surface. We will not add a Plan tab, a Goals tab, an OKR tab, or a Strategy tab. Those things either compose from Status, Ask, and Approve or they are the agency's job to deliver outside the product.
We will not build a dashboard. The Status surface refuses to become one. The moment we add a tile grid with rollup numbers, we have crossed a line and we know it.
Three surfaces. Refused additions. That is the product.
If you want to see it running on real client data, the SprintRay pilot is the place to start. Twenty minutes. The Innovative Group team. Your procurement questions answered.
Request a demo