Commercial Engine

ICP Scorecard

Enter what you know about an account and see how well it fits XDARK. Leave a field on “Unknown” when you do not know it. Unknown is never counted as zero: it only lowers how much of the score is based on facts.

Account

Example inputs

Account name

Commercial Engine

Rank a list

Upload a CSV or Excel list of accounts. Every account is scored with the same scorecard and ranked from best fit down. Column names from the XDARK scorecard workbook are picked up automatically; otherwise map them below.

Drop one or more .xlsx, .xls or .csv files here

No file loaded. Showing example rows.

Column mapping

File Sheet

Infer Regulation and Timing from sector when the file has no such columns (Likely for banks and non-bank finance, None for other sectors)

#	Account	Fit	Tier	Match (known)	Coverage	Score mix	Flags

Click a row to see how each factor was read and scored. Fit is the rank score: known points plus a share of the unknown weight. Match is known points divided by the weight we know. Coverage is the share of the 100 points that rest on real data.

Methodology

Scorecard & weights

Eight factors, 100 points. Regulation, budget fit and track record with us carry the most. Every number below can be edited and the Score and Rank pages recalculate at once. Edits stay in this browser. Unknown values are shown as “?” and are never read as zero.

Eight factors, 100 points

Factor	Max points	Comes from	How it is scored

Max points set how much weight a factor has when it is known. Points for each answer are in the tables below. Keep the max equal to the highest points in that factor's table.

Size, budget and ranking settings

Fit, tier and unknowns

FitKnown points plus 30% of the weight that is unknown. This is the number used for the tier and the ranking.

MatchKnown points divided by known weight. It says how good the account looks on the facts we have.

CoverageKnown weight out of 100. Below 50, treat the score as a guess and fill gaps before deciding.

Tiers

Big names

Evidence and limits

Source dataFactWhiteGuard backlog and pipeline (Nov 2023 to May 2026), HubSpot, the Control and Collective sheets, ICP Rec, Market Research and Channel EGY, merged into one Egypt-only account master. Accounts outside Egypt are excluded.

Time testFactScored on 2024 and 2025 deals only (74 accounts, 17 wins), without Track record and Persona: AUC 0.61, 90% interval 0.47 to 0.74. The top third won 32% against 18% for the rest. Weak, but better than random.

In-sampleInferenceThe weights were set after looking at the same deals, so tiers flatter the model. Treat the tiers as a priority order, not a probability of winning.

Lost dealsInferenceEvery backlog row is read as a proposal issued. A proposal with no recorded outcome is treated as probably lost. Lost reasons are estimated at 70% price and 30% missing NTRA certification, so a loss scores 10 and not 2.

RegulationUnknown“Likely” means the sector or tag points to CBE, FRA or NTRA and the licence was not seen. Only web-checked accounts are “Confirmed”. Payments and e-money fall under CBE; non-bank lending, microfinance, leasing, insurance and brokerage under FRA; telecom under NTRA.

Thin factorsUnknownOffer fit, Timing and Persona rest on thin or inferred evidence. Persona titles showed no lift in the data and are kept at 5 points. Employee counts are third-party estimates.