Denis Dmitriev
All projects
Research/In progress/2026/Built for EvoSoft

Sales forecast across three products

Reconciling two mismatched deal registers into a single forecast that holds up under management questions.

  • One forecast instead of three independent lists in different shapes
  • A slice by probability tier shows where the forecast holds and where it is hope
  • A dedicated page for the defence: the numbers and the reasoning are shown together

The problem

Deal data lived in two spreadsheets: a working one and a CRM export. On the surface they describe the same thing, but three products sat as independent column blocks across the same rows — and the rows did not correspond. Any attempt to compute 'across the row' produced a plausible, wrong figure.

How it works

First, taking the sources apart and stating explicitly that the rows are independent. Then consolidation under the owner's methodology: a slice by probability tier rather than one bottom line, enriched with statuses and pipeline stages from the CRM. The result is presented as its own page where the number and its derivation sit side by side.

The hard part

The decisive move was refusing to compute weighted monthly revenue — the most obvious and best-loved metric in defences like this. A weighted sum hides structure: two utterly different portfolios yield the same number, and 'why that much?' has no answer. Probability tiers have one.

Evidence

  • The structural mismatch between sources was found before the calculation, not after a question in the room
  • The methodology is written down and reproducible on new exports
  • The forecast is presented so any figure can be unfolded down to the deal

Stack

Google SheetsBitrix24Pythonаналитика продаж