Corporate Training Vendor Scoring Rubric: Weights, Scale, Worked Scores & Decision Math (AHP, TOPSIS, QCBS)

Short answer: A fair corporate-training vendor scoring rubric carries 7–10 criteria with weights totaling 100%, an anchored 1–5 scale with descriptors (5 = strong, specific evidence), separated technical and commercial evaluation, and a technical pass threshold before price is opened. For strategic, scaled programs, deepen with AHP (Analytic Hierarchy Process — Thomas L. Saaty) for weights and TOPSIS (Hwang & Yoon) for ranking. For government/BUMN procurement, use QCBS (Quality- and Cost-Based Selection) under Presidential Reg. 16/2018 jo. Presidential Reg. 12/2021 and LKPP Reg. No. 12/2021 (amended by No. 4/2024) — technical weight 60–80%, price 20–40%, with the LKPP price-normalization formula. Score individually first, then team-calibrate, with per-score evidence preserved as audit trail.

Most "vendor scoring rubric" articles hand over an empty table without weight logic, without score anchors, and without a way to manage evaluator bias. The rubric looks tidy on paper while the final decision still falls to impression. This guide closes that gap: weight mathematics (intuitive vs AHP), an anchored scale with descriptors, a full QCBS worked example with LKPP price normalization, anti-bias techniques, multi-evaluator aggregation, TOPSIS integration, and ready templates per channel (private, BUMN, LKPP).

Intended readers: HR / HC / L&D / SDM, Procurement, and evaluation panels that score training-vendor proposals — in private companies, BUMN/BUMD, government agencies, institutions, associations, and non-profits.

Quick navigation

1. Why a weighted rubric is not a formality
2. Rubric anatomy: criteria, weights, scale, evidence
3. Three ways to set weights (intuitive, AHP, swing)
4. Anchored 1–5 scale: descriptors that keep scores honest
5. Four evaluation methods: QCBS, QBS, LCS, FBS
6. Full QCBS worked example (illustrative)
7. LKPP-style price normalization formulas
8. Intro to AHP & TOPSIS integration
9. Anti-bias evaluator techniques (seven)
10. Multi-evaluator score aggregation
11. Baseline template & per-channel variants (private/BUMN/LKPP)
12. Illustrative case: two finalist vendors
13. Common mistakes & how to avoid them
14. FAQ
15. Next step

Why a weighted rubric is not a formality

A training-vendor decision without a rubric is one of the largest hidden losses in L&D budgets. Without a rubric:

• Decisions fall to lowest price — 10% saved on price becomes 100% lost when training fails to change behavior.
• Decisions fall to impression — "Vendor A's presentation was great" beats a more competent Vendor B with less polish.
• Decisions cannot be defended — auditors/management ask "why was Vendor X chosen?" and "team consensus" is not enough.
• Evaluator bias compounds — halo effect, anchoring, recency stack up unchecked.
• Healthy vendors stop bidding — when the process appears subjective, the best vendors with many client options will serve clients with clearer processes.

A weighted rubric turns the decision from impression to weighted evidence. It is the one artifact that makes the entire RFP cycle defensible before finance, audit, and the board.

Rule of thumb: A rubric is the panel's contract with itself before reading any proposal. Without that contract, the cheapest price and the slickest deck win every time.

Rubric anatomy: criteria, weights, scale, evidence

Four mandatory components:

Component	What is needed	How to do it right
Criteria	7–10 decision dimensions	Tune from a baseline template; ensure measurement, tax, data are included
Weights	Total 100%	Intuitive/consensus for normal training; AHP for scaled-strategic
Scale	Anchored 1–5	Per-level descriptors reduce evaluator drift
Evidence	Written per score	Mandatory for 5 (strong) and 1–2 (weak) — to enable audit

Plus two process artifacts:

• Technical pass threshold (e.g. weighted ≥ 3.5 of 5, or 70 of 100) — vendors below it do not enter price evaluation.
• Per-evaluator score sheet + team consolidation — kept as audit trail.

Three ways to set weights (intuitive, AHP, swing)

(1) Structured consensus

The panel discusses priorities and sets weights. Healthy when structured: each member proposes individual weights, then a consensus discussion produces written justifications. Fits standard training.

Baseline template (typical behavior/leadership program):

Criterion	Weight
TNA fit & business goal	20%
Facilitator quality & transparency	15%
Methodology & measurement plan	15%
Customization capability	12%
Verifiable track record	10%
Legality & quality standards	8%
Tax & procurement compliance	8%
Data security & protection	7%
Pre/post support & value for money	5%
Total	100%

Tune: certification programs → legality up; BUMN/government → tax/procurement up; sensitive-data programs → data up.

(2) AHP (Analytic Hierarchy Process)

Developed by Thomas L. Saaty (1980). For scaled-strategic training (multi-module academies, retainers, transformation):

1. Hierarchy: goal (pick the best vendor) → main criteria → sub-criteria → alternatives (vendors).
2. Pairwise comparisons between criteria on the Saaty 1–9 scale: 1 = equal, 3 = slightly more important, 5 = more important, 7 = much more important, 9 = absolutely more important (2,4,6,8 intermediates).
3. Priority vector: compute the eigenvector of the comparison matrix; that is the weight set.
4. Consistency: compute the Consistency Index (CI) and Consistency Ratio (CR = CI/RI). CR < 0.1 = consistent; ≥ 0.1 = redo comparisons.
5. Apply weights to alternative scores per criterion.

AHP is the standard in MCDM research and widely used in strategic procurement across energy, manufacturing, and healthcare. It fits when stakes are high; for single-class training, it is too heavy.

(3) Swing weighting

The panel rates "how much impact moving this criterion from the worst to the best score has on the overall decision". The highest-impact criterion gets weight 100; others are normalized proportionally and converted to percentages. Sits between intuitive and AHP.

Anchored 1–5 scale: descriptors that keep scores honest

Bare numbers (1, 2, 3, 4, 5) without meaning produce drift: one evaluator is conservative and rarely gives a 5, another is generous and always gives a 4. Descriptive anchors fix this.

Score	General anchor	Anchor for "Facilitator quality"
5	Strong, specific evidence, exceeds expectations	Per-named CVs, ≥ 1,000 hours in client industry, video sample, relevant certifications, anti-substitution clause
4	Good evidence, meets expectations	Per-named CVs, solid hours, sample available
3	Adequate but generic	General team profile without names, hours unspecific
2	Weak / thin evidence	Names only without CV, no sample
1	Unverified claim / evasive	"Trainer pool" without names, sample refused

Write descriptors per criterion before the process starts. Attach to the rubric. Evaluators match evidence to descriptor — they do not translate impression into a number.

Rule of thumb: Scores of 5 and 1–2 must carry written evidence. Scores of 3–4 may carry a one-line note. This filters lazy scoring and becomes the audit trail.

Four evaluation methods: QCBS, QBS, LCS, FBS

Methods rooted in World Bank and LKPP practice. Pick one in the KAK/RFP before the process starts.

Method	How	Fits
QBS (Quality-Based)	Pick highest technical, negotiate price	Very specialized consulting, quality stakes dominant
QCBS (Quality- and Cost-Based)	Final = tech_weight × tech_score + price_weight × price_score	Default for corporate training (tech 60–80%, price 20–40%)
LCS (Least-Cost)	Pick cheapest among those who pass the technical threshold	Commodity training, very uniform scope
FBS (Fixed-Budget)	Budget locked; pick highest technical within ceiling	Fixed DIPA/RKAP with hard ceiling

For TNA-based corporate training, QCBS with technical 70% and price 30% is the healthy default: it values approach without ignoring price reasonableness.

Full QCBS worked example (illustrative)

Illustrative scenario (numbers for the method):

Three vendors bid for a 5-day leadership program, 24 participants, in-house Jakarta. The KAK sets QCBS with tech weight 70%, price 30%, technical pass threshold 70/100.

Step 1 — Technical scores per criterion (1–5 scale)

Criterion	Weight	Vendor A	Vendor B	Vendor C
TNA fit	20%	5	4	3
Facilitator	15%	4	5	3
Methodology & measurement	15%	5	4	3
Customization	12%	4	4	4
Track record	10%	4	5	3
Legality	8%	4	4	5
Tax & procurement	8%	5	4	4
Data security	7%	4	4	3
Support & value	5%	4	3	4

Step 2 — Weighted technical score (out of 5)

• Vendor A: 5·0.20 + 4·0.15 + 5·0.15 + 4·0.12 + 4·0.10 + 4·0.08 + 5·0.08 + 4·0.07 + 4·0.05 = 4.43 of 5 → 88.6 of 100.
• Vendor B: 4·0.20 + 5·0.15 + 4·0.15 + 4·0.12 + 5·0.10 + 4·0.08 + 4·0.08 + 4·0.07 + 3·0.05 = 4.18 of 5 → 83.6 of 100.
• Vendor C: 3·0.20 + 3·0.15 + 3·0.15 + 4·0.12 + 3·0.10 + 5·0.08 + 4·0.08 + 3·0.07 + 4·0.05 = 3.33 of 5 → 66.6 of 100.

Technical pass 70: Vendor C does not pass and exits the final evaluation.

Step 3 — Price score (LKPP normalization)

Bid price (excl. VAT, illustrative): Vendor A = IDR 220 million; Vendor B = IDR 180 million.

Normalization formula: price_score = (lowest_price ÷ vendor_price) × 100

• Vendor A: (180 ÷ 220) × 100 = 81.8
• Vendor B: (180 ÷ 180) × 100 = 100

Step 4 — Final QCBS score

Final score = (70% × tech_score) + (30% × price_score)

• Vendor A: 0.70 × 88.6 + 0.30 × 81.8 = 62.02 + 24.54 = 86.56
• Vendor B: 0.70 × 83.6 + 0.30 × 100 = 58.52 + 30.00 = 88.52

Winner: Vendor B (88.52 vs 86.56).

Interpretation

Vendor A leads technically (88.6 vs 83.6), but Vendor B wins on price (IDR 40 million / ~18% cheaper) — enough to flip QCBS. Lift the technical weight to 80% (high-risk consulting frame) and Vendor A wins (87.52 vs 86.88). Weights determine the winner — set them before proposals come in.

LKPP-style price normalization formulas

Variants used in Indonesian government consulting procurement:

Variant	Formula	Note
Inverse linear (most common)	score = (lowest_price ÷ vendor_price) × 100	Cheapest gets 100; others fall proportionally
Linear	score = 100 − [(vendor_price − lowest_price) ÷ lowest_price × 100]	Sensitive to small differences
Range	score = (highest_price − vendor_price) ÷ (highest_price − lowest_price) × 100	Linear map to 0–100; cheapest 100, dearest 0

The chosen formula is set in the KAK. Inverse linear is the most common because it rewards cheaper vendors proportionally without penalizing technically-stronger vendors too drastically when they cost more.

Intro to AHP & TOPSIS integration

For scaled-strategic procurement (multi-module academies, multi-year transformation), the AHP-TOPSIS hybrid delivers a stronger decision. MCDM literature in pharmaceutical, manufacturing, and power industries treats this integration as standard.

AHP sets criterion weights through pairwise comparison with a consistency check (Saaty, 1980). TOPSIS (Hwang & Yoon, 1981) ranks alternatives by distance to the positive ideal (per-criterion maximum) and negative ideal (per-criterion minimum / highest cost).

TOPSIS steps (with AHP-derived weights):

1. Build the decision matrix (vendor × criteria).
2. Normalize the matrix (vector formula: rᵢⱼ = xᵢⱼ ÷ √Σxᵢⱼ²).
3. Apply AHP weights: vᵢⱼ = wⱼ × rᵢⱼ.
4. Identify positive ideal (V⁺) and negative ideal (V⁻) per criterion.
5. Compute Euclidean distance of each vendor to V⁺ (Sᵢ⁺) and to V⁻ (Sᵢ⁻).
6. Relative closeness: Cᵢ = Sᵢ⁻ ÷ (Sᵢ⁺ + Sᵢ⁻). Highest Cᵢ wins.

When to use: strategic decisions with many vendors (≥ 5) and many criteria (≥ 10), especially when stakes are high and decision documentation will face external audit. For typical corporate training (3–5 vendors, 9 criteria), plain QCBS is sufficient.

Anti-bias evaluator techniques (seven)

#	Technique	Bias addressed
1	Per-level descriptive anchors	Inter-evaluator drift
2	Individual scoring before team discussion	Groupthink
3	Calibration session on a "calibration vendor"	Inconsistent standards
4	Partial blinding of vendor names	Reputation halo
5	Mandatory written evidence for 5 and 1–2	Lazy scoring
6	Cross-review on > 1-level differences	Extreme outliers
7	Separate technical and price evaluators	Price anchoring on technical

Most common biases in corporate training:

• Halo effect — a slick deck makes every criterion score high.
• Anchoring — the first vendor's score becomes the reference; later vendors are compared to it.
• Recency — the last vendor presented is remembered more vividly.
• Confirmation bias — evaluators look for evidence supporting first impressions.
• Group polarization — team discussion shifts consensus to extremes.

The seven disciplines neutralize them. Without them, a rubric only camouflages bias.

Multi-evaluator score aggregation

For a 3–5-evaluator panel, three aggregation options:

Method	How	Fits
Arithmetic mean	Σ scores ÷ evaluator count	Default; assumes equal evaluators
Weighted mean	Σ (evaluator_weight × score) ÷ Σ weights	When expertise differs (technical × 1.5; admin × 1.0)
Median	Middle value	Robust against extreme outliers

Healthy practice: arithmetic mean + calibration session for outliers > 1 level. Log every individual score + mean + correction reason.

Calibration session:

1. Each evaluator scores independently.

2. 1-level differences are discussed: the extreme evaluator explains evidence.

3. If a misreading is found (e.g. proposal misread), the score is corrected with a note.
4. If the gap persists (different interpretations of the same evidence), both scores are recorded and the mean is taken.

Rule of thumb: Widely-divergent individual scores are a productive signal worth investigating. They can mean descriptors are not sharp enough, proposal evidence is ambiguous, or an evaluator needs more briefing. Resolve at calibration — do not force premature consensus.

Baseline template & per-channel variants (private/BUMN/LKPP)

Baseline template (typical behavior/leadership program — private)

Criterion	Weight	Score 1–5	Weighted score	Evidence
TNA fit & business goal	20%
Facilitator quality & transparency	15%
Methodology & measurement plan	15%
Customization capability	12%
Verifiable track record	10%
Legality & quality standards	8%
Tax & procurement compliance	8%
Data security & protection	7%
Pre/post support & value for money	5%
Total	100%		Σ

Technical pass threshold: ≥ 3.5 of 5 (or 70/100).

Variant for BUMN procurement

Tune weights:

• Tax & procurement: 8% → 12% (BUMN VAT-collector status, code-03 tax invoice, SPK per BUMN procurement guidelines).
• Legality & quality: 8% → 10% (supplier-document completeness, registration in BUMN procurement system).
• Customization: 12% → 10% (scope usually more textured in BUMN KAK).

Total still 100% with proportional reduction elsewhere.

Variant for government procurement via LKPP

Method mandated QCBS per the KAK (or LCS/FBS per work nature). Technical scale 0–100 with pass threshold (typically 70). Technical weight 60–80%, price 20–40%. Price normalization per LKPP (inverse linear most common). Decision documentation follows SPSE/INAPROC and LKPP Reg. No. 12/2021 (jo. No. 4/2024). See the LKPP e-catalog & e-procurement guide for the full flow.

Illustrative case: two finalist vendors

Illustrative scenario.

Ministry X runs a procurement for leadership training, 30 middle managers, 6 days, in-house Jakarta. Method QCBS tech 70% price 30%, technical pass 70/100. Two finalists pass from 4 proposals.

	Vendor M	Vendor N
Technical weighted (of 100)	88	82
Bid price (excl. VAT)	IDR 280 million	IDR 230 million
Price score (inverse linear)	(230÷280)×100 = 82.1	(230÷230)×100 = 100
Final QCBS	0.70×88 + 0.30×82.1 = 61.6 + 24.6 = 86.2	0.70×82 + 0.30×100 = 57.4 + 30.0 = 87.4

Vendor N wins by a narrow margin (1.2 pt) because its price advantage (~22%) offsets Vendor M's technical advantage (~7%).

Panel notes (audit trail):

• The price gap has a technical explanation: Vendor M includes 2 senior facilitators + 1 supporting facilitator; Vendor N includes 1 senior facilitator. Vendor N risk: if the single facilitator falls ill, sessions are disrupted. Mitigation: facilitator-backup clause in the contract.
• Vendor M leads technically on "methodology & measurement" (Kirkpatrick L1–L4 plan with detailed instruments) and "track record" (three other ministries reachable as references).
• The QCBS decision is followed because it is mandated in the KAK. With QBS (no price), Vendor M wins. With FBS at IDR 250m ceiling, only Vendor N qualifies.

The case shows: method and weights determine the winner as much as scores themselves. Lock them in the KAK/RFP before the process starts; do not change midstream.

Common mistakes & how to avoid them

Key takeaways:

• Rubric without descriptive anchors → evaluator drift. Write per-level descriptors.
• Weights set after seeing proposals → bias. Set them before proposals come in; document.
• Opening price together with technical → price hijacks. Separate envelopes; technical first to threshold.
• Team scoring straight to consensus without individuals → groupthink. Score individually, then calibrate.
• Scores without written evidence → unauditable. Mandate it for 5 and 1–2.
• Too many criteria (>10) → attention split. Consolidate to 7–10 with descriptor sub-criteria.
• Choosing purely on price → cost of a failed training is far larger. Use QCBS; honor the technical threshold.
• No method set in the KAK/RFP → decision indefensible. Lock QCBS/QBS/LCS/FBS upfront.
• AHP for a single-class training → overkill. Structured consensus is enough.

FAQ

What is a vendor scoring rubric and why is it mandatory for training procurement?

A vendor scoring rubric is a decision matrix containing evaluation criteria, per-criterion weights (totaling 100%), a standard rating scale (typically 1–5), and a per-vendor weighted-score area. It is mandatory because decisions without a rubric fall to impression, lowest price, or relationship — biases that misdirect training and get billing rejected at audit. A rubric makes decisions comparable (apples-to-apples), defensible (written reasoning per score), and auditable (structured trail). For BUMN/government procurement via LKPP, rubric-based decision documentation is a precondition of BPK/BPKP audit.

How do you set fair criteria weights?

Three methods. (1) Intuitive/consensus: the panel discusses and sets weights — fast but exposed to voice-dominance and cognitive bias. (2) AHP (Analytic Hierarchy Process) by Thomas L. Saaty: pairwise comparisons between criteria on a 1–9 scale, with computed priority vector + Consistency Ratio (CR < 0.1) — the healthiest for strategic decisions, auditable, bias-reducing. (3) Swing weighting: the panel rates 'how important is the change from worst to best on each criterion'. For typical corporate training, AHP is overkill; structured consensus with a baseline weight template (TNA 20% / facilitator 15% / methodology 15% / customization 12% / track record 10% / legality 8% / tax 8% / data 7% / support 5%) tuned per case is adequate.

1–5 or 1–10 scoring scale — which is better?

An anchored 1–5 scale (5 = strong, specific evidence; 4 = good; 3 = adequate but generic; 2 = weak; 1 = unverified / evasive) is healthier for corporate training. Reasons: (a) 1–10 granularity is illusory precision — evaluators do not really distinguish 7 from 8; (b) 1–5 forces clearer qualitative discussion; (c) per-level descriptive anchors reduce inter-evaluator drift. 1–10 fits when many vendors (>15) demand fine differentiation, or when used alongside TOPSIS. For Indonesian government QCBS, a technical 0–100 scale with a pass threshold is common.

QCBS vs QBS vs LCS vs FBS — when does each apply?

Four evaluation methods rooted in World Bank / LKPP practice. QBS (Quality-Based Selection): pick highest technical without price consideration — for very specialized consulting. QCBS (Quality- and Cost-Based Selection): final score = (tech weight × tech score) + (price weight × price score); healthy default for corporate training with tech 60–80% and price 20–40%. LCS (Least-Cost Selection): among all who pass the technical threshold, pick the cheapest — for commodity training. FBS (Fixed-Budget Selection): budget locked; pick highest technical quality within the ceiling — fits a fixed DIPA/RKAP. For BUMN/government procurement, the method is chosen per the nature of the work and set in the KAK.

How do I compute the per-vendor weighted score (QCBS example)?

Three steps. (1) Technical weighted score per vendor = Σ (criterion weight × criterion score 0–100). Example Vendor A: TNA 5/5 × 20% + facilitator 4/5 × 15% + ... = weighted 4.1 of 5 or 82 of 100. (2) Price weighted score: common LKPP normalization → vendor_price_score = (lowest price ÷ vendor price) × 100. The cheapest vendor gets 100; others fall proportionally. (3) Final QCBS score = (70% × tech score 0–100) + (30% × price score 0–100). Highest final score wins. Only vendors who pass the technical threshold (e.g. ≥ 70) enter the final calculation.

How do you reduce evaluator bias?

Seven techniques: (1) descriptive anchors per score level — not bare numbers; (2) score individually before team discussion to avoid groupthink; (3) calibration session on a 'calibration vendor' scored together; (4) partial blinding — hide vendor names while scoring certain sections; (5) require written evidence for every 5 and every 1 or 2; (6) cross-evaluator review on differences > 1 level; (7) separate evaluators for technical and price. The most common biases: halo effect (a good impression on one criterion spreads to all), anchoring (the first score sets the reference), and recency (the last vendor is remembered more sharply). Process discipline neutralizes them.

How do you aggregate scores across multiple evaluators?

Three options. (a) Arithmetic mean per criterion per vendor — simplest, assumes equal evaluator weight. (b) Weighted mean per evaluator — when expertise differs (e.g. technical evaluator × 1.5; administrative × 1.0). (c) Median — robust against outliers, fits when one evaluator is consistently extreme. For a 3–5-evaluator panel, arithmetic mean + a calibration session that corrects outliers > 1 level is the healthiest and most defensible. Log every individual score + mean + correction reason as an audit trail.

What is AHP and when do you use it for training procurement?

Analytic Hierarchy Process (Thomas L. Saaty, 1980) is a multi-criteria decision-making method. Steps: (1) structure the problem as a hierarchy (goal → criteria → sub-criteria → alternatives); (2) pairwise comparisons between criteria on a 1–9 Saaty scale (1 = equally important, 9 = absolutely more important); (3) compute the priority vector from the matrix; (4) check consistency (Consistency Ratio < 0.1 — else, redo comparisons); (5) apply weights to alternative scores. AHP fits large-scale strategic training (multi-module academies, multi-year retainers, transformation programs) where the investment warrants structured justification. For a single-class training, AHP is overkill — structured consensus suffices.

What is TOPSIS and how does it complement AHP?

TOPSIS (Technique for Order Preference by Similarity to Ideal Solution, Hwang & Yoon 1981) ranks alternatives by distance to a positive ideal and a negative ideal solution. Steps: (1) normalize the decision matrix; (2) apply weights (often from AHP); (3) define positive ideal (per-criterion maximum for 'higher is better' criteria, minimum for cost) and negative ideal (the inverse); (4) compute Euclidean distance from each alternative to both ideals; (5) compute relative closeness Cᵢ = negative_distance ÷ (positive_distance + negative_distance). Highest Cᵢ wins. AHP-TOPSIS hybrids are standard in MCDM supplier-selection literature because AHP is strong on weights and TOPSIS is strong on ranking.

How many criteria are ideal in a rubric?

7–10. Fewer than 7 misses important dimensions (tax, data, post-support); more than 10 reduces evaluation quality per criterion (split attention) and per-criterion weight becomes too small to differentiate vendors. A healthy baseline of nine: TNA fit, facilitator quality, methodology & measurement, customization, track record, legality & quality standards, tax & procurement compliance, data security, support & value for money. If sub-criteria are needed (e.g. facilitator = experience + certification + sample), hierarchize them in AHP or use per-score descriptors — do not split into 20 shallow rows.

Can the same rubric serve private RFPs and government/BUMN procurement?

The rubric core (criteria, weights, scale, worked scores) is the same. What differs: (a) document form — government attaches to a formal KAK and SPSE; (b) selection method — government commonly QCBS with technical 0–100 scale and LKPP price-normalization formulas; (c) weights — tax/procurement compliance and legality go up for government/BUMN; (d) outputs — government requires selection-result minutes, SPK contracts per Presidential Reg. 16/2018 jo. 12/2021, e-purchasing via Electronic Catalog V6 (Head of LKPP Circular No. 9/2024). Private corporate rubrics are shorter and more flexible. Use the same baseline template and tune weights + document form per channel.

How do you handle vendors whose technical scores are equal but prices differ widely?

First check whether the price gap has a technical explanation: different scope (days, participants, customization), different facilitator seniority (senior vs junior), or different tax structure (incl./excl. VAT, PPh 23 readiness). When scope is equal, QCBS naturally leans toward the cheaper because of the price-normalization formula (lowest price gets the maximum). Avoid choosing purely by price when the gap < 10% and qualitative differences exist (track record, facilitator) — the cost of a failed training far exceeds 5–10% savings. For government procurement, follow the method set in the KAK (QCBS/LCS/FBS) — decisions outside the formula = audit findings.

Next step

You now have a complete rubric framework: criteria, weights, an anchored scale, the QCBS method with a worked example, evaluator anti-bias, team aggregation, and per-channel variants. The sensible next step is to build an RFP/KAK with this rubric as a mandatory annex — before inviting any vendor.

Neksus operates in this lane in a way that makes scoring honest: every proposal arrives with a module-to-measurable-objective map, per-named facilitator profiles with sample sessions, a Kirkpatrick L1–L4 plan with instruments, a transparent budget with tax positioning, and procurement-document readiness (NPWP/PKP, SPSE/INAPROC profile where relevant). A healthy vendor scores high because the evidence is real; deck polish carries minimal weight. Discuss your team's need and request an initial TNA via the Neksus contact page — no obligation.

Also see the companion guides:

• Corporate Training RFP: Template & Evaluation Criteria
• How to Choose a Corporate Training Vendor
• Training Needs Analysis (TNA)
• LKPP E-Catalog & E-Procurement for Training Vendors
• Training PO, VAT & Tax Invoice Procedure
• Kirkpatrick 4-Level Deep
• Phillips ROI Level 5
• See the full training catalog →

---

Last updated: 18 May 2026. The frameworks cited (AHP — Saaty 1980; TOPSIS — Hwang & Yoon 1981; QCBS/QBS/LCS/FBS — World Bank & LKPP framework; Presidential Reg. 16/2018 jo. Presidential Reg. 12/2021; LKPP Reg. No. 12/2021 jo. No. 4/2024; Head of LKPP Circular No. 9/2024) are attributed to their official sources. Worked calculations are illustrative for method demonstration; the numbers exist to show the math. Procurement mechanics are confirmed per contract under current regulations — validate with your procurement and legal teams. Neksus does not publish client names or success statistics.