Role-Based AI Dashboards: What Sales, Finance & Ops Teams Should Track

Most BI builds in Indian mid-market businesses follow the same pattern: a consultant arrives, gathers requirements from every team, and ships one massive dashboard with 40 charts trying to cover everyone. Within 60 days, three things happen. The sales head stops opening it because their three numbers are buried. The CFO opens it once a month for the board meeting and reads everything else from email. The procurement head asks for "just a small change" that turns into a four-week consultant ticket. Within six months, the dashboard has become a wallpaper.

The honest problem is not the dashboard's design. It is the assumption that one view fits every decision. A sales head deciding which 5 customers to call this week needs pipeline velocity, recent deal slippage, and customer-wise margin. A CFO deciding which receivable to escalate needs DSO trend, top overdue, and the cash gap next 14 days. These are different decisions; they should be different views on the same underlying data.

Role-based AI dashboards solve this by inverting the model. One read layer underneath - reading Tally, CRM, ERP, and Excel once - and four role-tailored views on top, each surfacing the 3-5 KPIs that role acts on weekly. Less noise per role, faster decisions, no per-team rebuild.

Four roles cover almost every operational decision in an Indian mid-market business. Each card below names the role, the 4-5 KPIs that matter, the question each KPI answers, and the digest cadence that fits the rhythm of that role's week.

KPIs that matter: weekly pipeline value vs prior 4 weeks, top 20 customer revenue change, salesperson conversion rate, customer-wise realised margin, stalled-quote ageing. Where the data lives: CRM (Sell.do, LeadRat, custom), quotation email threads, Tally invoices and credit notes. The question this answers: "Which 5 customers should I personally call this week, and which salesperson needs a 1-on-1 with data?" Digest cadence: daily 8:30 pm summary plus a Monday morning deeper view with the top 3 actions for the week.

KPIs that matter: cash position this week vs commitments next 14 days, DSO trend, top 10 overdue receivables, GST input credit reconciliation gap, SKU-level margin drift. Where the data lives: Tally per company, bank statements, GST returns, scheme calendar in Excel. The question this answers: "What is the most important collection call to make today, and where is margin quietly drifting before month-close?" Digest cadence: 7:00 am cash digest, weekly margin drift review on Wednesday, monthly close summary.

KPIs that matter: top vendors by spend with committed vs realised lead time, PO-GRN-invoice mismatches this period, duplicate-invoice risk flags, raw material standard vs realised cost. Where the data lives: ERP / Tally for POs and invoices, WMS for GRNs, supplier rate cards in Excel, dispatch emails from vendors. The question this answers: "Which 3 vendors should I renegotiate with this quarter, and which duplicate-invoice flags need finance to review before payment?" Digest cadence: weekly Tuesday digest, plus instant alerts on duplicate-invoice flags and PO-GRN mismatches above a value threshold.

KPIs that matter: output vs plan per line per shift, dispatch risk for next 72 hours, dead-stock additions this week, inventory variance (Tally vs WMS) per godown. Where the data lives: ERP / MES, WMS, supervisor sheets, Tally godown stock, customer commitments from the CRM. The question this answers: "Which line needs intervention this shift, and which dispatch is at risk before the customer call lands?" Digest cadence: end-of-shift summary, daily 8:30 pm report on dispatch readiness and dead-stock additions.

Adoption is the only metric that matters for a dashboard. A view that no one opens may as well not exist. Three failure patterns show up reliably:

• Noise per role. A company dashboard with 40 charts shows every role 36 charts they do not need. The 4 that matter to them are scattered and easy to miss.
• Cadence mismatch. The sales head wants daily, the CFO wants weekly, the procurement head wants Tuesday morning, the plant head wants end-of-shift. One refresh schedule fits none of them.
• Cross-team politics. "Why is sales seeing finance's margin number? Why is finance seeing production's downtime?" Shared single dashboards create needless cross-team noise that role-based views avoid by design.
• Ad-hoc questions get blocked. When the dashboard cannot be tweaked per role, every new question becomes a consultant ticket. Adoption dies in the lag.

Role-based AI dashboards remove all four failure modes by sharing the read layer but separating the views, cadences, and query surfaces.

One AI Analytics Platform underneath. Four role-tailored views on top. No per-team consultant build, no duplicate data layer.

• Connect each source once. Tally per company (native connector), CRM via DB or API, ERP / MES, WMS, Excel from a shared folder. All four roles read from the same connections.
• Configure each role's KPI set. During the 14-day POC, each team picks the 3-5 KPIs that drive their weekly decisions. These become the digest body for that role.
• Pick the cadence per role. Sales gets 8:30 pm + Monday 7:00 am. CFO gets 7:00 am cash + Wednesday margin. Procurement gets Tuesday morning. Ops gets end-of-shift. Configurable, not fixed.
• Delivery in the channel that fits. Email digest for the office-bound roles. WhatsApp digest for the owner and field-bound roles. Web app for deep exploration. Native Android (iOS in App Store review).
• Plain-English query surface for all. When someone has a question their KPI set does not cover, they type it in English or Hindi. Same answer surface across every role.

Faster role-based visibility is not a dashboard. It is a different operating rhythm per team.

• Sales head stops asking the analyst. The 5 customers to call this week arrive in the Monday digest, ranked by margin opportunity and last-touch ageing.
• CFO stops chasing the accountant. The cash position vs commitments view lands at 7:00 am. The collection decision happens before 10:00 am.
• Procurement head walks into vendor reviews with data. Realised vs committed lead-time per vendor, ranked by drift. The conversation moves from "you have been late" to "you have been 7 days late on average this quarter, here is the data".
• Plant head reschedules during the shift, not after. Dispatch risk surfaces 24 hours before the customer call lands. The line gets rebalanced the same morning.
• The owner stops being the integration layer. Each team's view is self-contained. The owner sees a cross-team summary digest and steps in on the issues actually flagged red.

Worth being explicit about scope:

• Not a strategy tool. Each role gets the data on what is happening; the decision still requires the human's judgement. KolossusAI surfaces the gap, not the strategy.
• Not a replacement for the operational system. Sales keeps their CRM. Finance keeps Tally. Procurement keeps their PO workflow. Operations keeps their MES. We read these in place.
• Not for daily 1-on-1 performance management. Role dashboards show team-level performance and customer-level patterns. Individual employee surveillance is not the goal, and adoption dies if it becomes the perception.
• Custom KPIs need vocabulary tuning. If your business calls something different from the industry default (e.g., a custom margin formula), the POC week is when we align that. After tuning, every role reads it correctly.

One dashboard for the whole company is how most BI builds quietly fail. Role-based AI dashboards solve the noise problem by sharing the read layer and separating the views - sales sees what sales acts on, finance sees what finance acts on, procurement sees what procurement acts on, ops sees what ops acts on. Same data, four windows.

The cost is one connection per source, three weeks of vocabulary tuning, and configuring each team's 3-5 KPIs and digest cadence. The return is the four teams that stop asking each other for numbers and start having the right conversation backed by data. AI Analytics Platform - free 14-day POC on your real systems. The first role-based digest lands the evening you connect.