Finance and management · Spend Analytics & Procurement

Where the money is going, at what price and where to save

The analysis layer on top of Purchasing. The expense of order, receipt, contract, advance, account payable and write-off enters a cube classified by specific purchasing category, in three stages that never add up: committed to the order, realized in the debt and paid in the cash outflow. From there comes supplier concentration, reference price per item, savings opportunity with evidence, savings measured month by month, double payment to be recovered, three-way verification, average payment term, supplier scorecard, positioning matrix by category, comparison with market inflation, sourcing and compliance scenario simulation that really blocks it. Nothing is typed: everything is derived from what the company has already purchased.

Talk to a specialistSee plans

The problem and the solution

What changes when Spend Analytics & Procurement works inside the platform

This is connected information about classes, graduates and events across 22 modules, replacing scattered spreadsheets and messages.

Without the platform

Information about Spend Analytics & Procurement lives in separate spreadsheets, and nobody knows which version is current.

With Partiu Formatura

22 modules share the same class, graduate and event records.

Without the platform

A request for a number takes time because someone has to combine data from several places.

With Partiu Formatura

Spending Cube and Own Taxonomy and Automatic Classification and Unclassified Queue update together, with figures ready for the next meeting.

Without the platform

The process depends on who is working that day, and its history disappears.

With Partiu Formatura

Everyone follows the same workflow, with recorded actions and role-based permissions.

Without the platform

Connecting with other departments means exporting spreadsheets and entering data again.

With Partiu Formatura

Spend Analytics & Procurement shares information directly with other areas of the platform.

How it works

How information moves through Spend Analytics & Procurement

Each step is a real module in this area. Its output moves to the next step without anyone entering the same information again.

What is available in Spend Analytics & Procurement

Spending Cube and Own Taxonomy

The company's expenses read by the tree that the buyer understands, and not by the accounting account.

  • Spending taxonomy at three levels, separate from the financial category: accounting talks about consumables, those who negotiate talk about graphics, invitations and special paper
  • Initial model with nine families, including non-addressable spend, and more than forty subcategories typical of a graduation company
  • Category marked as critical has greater weight in the risk and positioning matrix
  • Cube assembled of the purchase order, order item, receipt, advance, bill payable and bank write-off, without typing
  • Three separate stages at all times and never added together: committed to the order, fulfilled to the account payable and paid at the cash outflow
  • Total expenditure and addressable expenditure side by side: tax, payroll, loan, bank fees and transfer between accounts appear in expenses and are outside of what purchases are negotiated
  • Coverage, purchase outside the process and expenses under contract calculated on the addressable, because tax has never been requested and never will be
  • Prorated payable account is distributed by cost center, category, class and branch, and not as a whole in the header branch
  • Return to supplier and credit note deduct the expense instead of disappearing from the account
  • Advance to supplier in the cube for the outstanding balance, which is the real exposure with those who have not yet delivered
  • Purchase in foreign currency converted with the accrual rate frozen at launch: March report does not change because the dollar moved in August
  • Checking the load on one screen: the total of the panel against the total of documents that generated it, by stage
  • Crossing two dimensions on the same screen between category, supplier, cost center, class, buyer, item, type of purchase and competence
  • Pareto curve with accumulated, showing how many suppliers are responsible for most of the spending
  • Clicking on any line opens the entries that form the number, with document and supplier
  • Item purchased loose grouped by standard description, bringing together the various spellings of the same material
  • Reconstruction of the cube fully preserves the hand-made classification
  • Export of the entire cut to a spreadsheet, with all dimensions and order marking, quotation and contract
  • Featured ratings coverage: Without it, everything else on the dashboard would be a well-formatted guess

Automatic Classification and Unclassified Queue

The expense sorts itself out, and what's left becomes a short line instead of thousands of lines.

  • Rules by supplier, catalog product, NCM, financial category, cost center, purchase type or word in the description
  • Explicit priority between rules, with a tiebreaker based on specificity: CNPJ is worth more than a keyword
  • NCM rule accepts prefix, taking the entire code family at once
  • Word rule requires all words in the description, so coated paper does not capture any paper
  • Unclassified queue grouped by supplier and item: four thousand launches saw one hundred and twenty decisions
  • Entire group mass sorting in one click
  • Rule created based on the classification that was just made, so that next month it will be classified
  • Application counter per rule, showing which rules actually work
  • General reclassification that reapplies rules without touching what was manually classified
  • Exclusion rules that tag unaddressable spend by finance category, cost center, supplier or word, with seventeen exclusions ready for day one
  • Own family for non-shopping, with the financial cost of delay falling on them alone

Concentration, Long Tail and Outside Buying

Who the company depends on, how much of the expense escapes the process and where the base is too fragmented.

  • HHI concentration index by category and for the entire company, with ready reading of pulverized, concentrated or very concentrated
  • Participation of the largest supplier and the three largest by category
  • Long tail measured by the cut defined by the company, with how many suppliers and how much value
  • Category with single supplier identified and flagged, with emphasis on critical categories
  • Expense that reached accounts payable without going through an order, measured in value and percentage
  • Purchase made outside of a current contract from the same supplier
  • Concentration history saved by competence, to compare before and after negotiation
  • Month-by-month evolution of the number of active suppliers

Savings Opportunities with Evidence

Each opportunity is born from a deterministic rule, with a calculated value and launches that support it.

  • Nine opportunity families: price above reference, supplier consolidation, off-order purchasing, fractional purchasing, single supplier, off-contract purchasing, never-quoted item, long tail, and idle contract
  • Potential value always conservative: applies the best price the company itself has ever achieved, not an imagined discount
  • Noise does not become an opportunity: differences below three percent or of an irrelevant value are discarded at the source
  • Evidence attached to each opportunity, with the numbers that generated it
  • Releases that support the number open on the screen, ready for conversation with the supplier
  • Own cycle of open, under analysis, accepted, captured and discarded, with responsible person and deadline
  • Discarded opportunity requires reason and is never suggested again by scanning
  • Deduplication between scans: the routine runs every day without filling the panel with the same thing
  • Manual recording of what the buyer sees and the rule does not

Reference Price and Basket Inflation

How much the company usually pays for each item, and how much this has risen without anyone noticing.

  • Reference price per item with last price, best price, quantity-weighted average, median and contract price
  • Supplier and date of each reference, to know who the best price was obtained from
  • Price dispersion by item, pointing out where the same material was purchased at very different prices
  • Competency pricing history, separating market growth from supplier switching
  • Comparison by supplier of the same item, with the lowest, average and highest prices charged
  • One hundred base price index per item, per category and for the entire shopping basket
  • Basket calculated with the quantity of the base period, separating price increase from volume increase
  • Twelve-month variation of the basket, which is the real inflation of the company's purchases
  • Item purchased once is excluded from the index, so as not to distort the average

Savings with Baseline and Measured Realization

Announced savings are not realized savings. Here the two appear side by side.

  • Saving classified as hard, soft and avoided cost
  • Baseline pulled from item reference price, with seven declared methodologies
  • Value calculated from baseline, negotiated price and quantity, never entered in the total field
  • Annualized value calculated based on the informed validity
  • Configurable mandatory approval, with approver, date and reason for rejection recorded
  • Approved Savings cannot be edited: changing the number after approval is the end of the panel's credibility
  • Achievement launched month by month, with the total always recalculated from the sum of skills
  • Annual company target and category, with percentage of achievement on the panel
  • Opportunity that turns into saving changes state on its own and leaves the queue
  • Own numbering per year in the standard of the rest of the platform

Category Management and Sourcing Pipeline

Each category has an owner, position in the risk and value matrix, strategy and initiatives that deliver the goal.

  • Positioning matrix with supply risk and financial impact, and the four quadrants of non-critical, leverage, bottleneck and strategic
  • Position suggested by the system based on expenditure, concentration and criticality, and adjustable by the category owner
  • Seven strategies per quadrant: consolidate, compete, partner, ensure supply, replace, standardize and internalize
  • Generated category plan already filled with expenditure from the year, previous year, number of suppliers and concentration index
  • Category owner, diagnosis, objectives and risks recorded in the plan
  • Sourcing pipeline in seven steps, from identification to completion
  • Savings weighted by probability, which is the number that the board should look at, and not the gross sum of what was predicted
  • Initiative born from opportunity, which now has responsibility and deadline
  • Initiative with an expired deadline flagged, because its savings continue to count in the pipeline
  • Suppliers, opportunities and category savings gathered on the same plan screen

Scorecard and Supplier Risk

Note that comes from what happened, and not from opinion: delivery on time, item accepted, price and document up to date.

  • Five axes with configurable weights that add up to one hundred: quality, deadline, price, service and compliance
  • Deadline measured for deliveries against the date promised in the order, with a tolerance in days defined by the company
  • Quality measured in the quantity accepted over that received, with return and damage weighing
  • Price measured against the reference of the item itself, weighted by value
  • Service coming from manual assessment, the only subjective axis, on purpose
  • Compliance from expired document, approval and blocking
  • Axis without data leaves the calculation and the weights are redistributed: lack of information does not result in a zero score
  • Supplier with few deliveries marked as insufficient sample instead of receiving a made-up note
  • Classification from A to D and evolution of the grade by competence
  • Risk in five separate and visible factors: dependence, documentation, performance, concentration and contractual
  • Segmentation into strategic, preferred, approved, tail and to be discontinued, suggested by the system and overwritten by the buyer
  • Hand-defined segment stops being recalculated, with justification and development plan

Purchasing and Compliance Policy

The rule that the company agreed to follow, checked against what actually happened.

  • Nine types of policy: mandatory order, mandatory quote, minimum number of proposals, mandatory contract, approved supplier, valid document, anti-fractionation, ceiling price and single supplier
  • Seven default policies ready to go, with limits coming from module configuration
  • Scan that generates the finding with document, supplier, buyer, value and evidence
  • Fractional purchase detected by the month's set of orders from the same supplier, each just below the quotation threshold
  • Supplier document checked on the order date, not today's date
  • Price above the reference with the tolerance defined by the company
  • Finding treated as justified, corrected, accepted or false positive, always with justification recorded
  • Finding already recorded is never rewritten by the next scan
  • Finding whose cause has ceased to exist closes itself, with date and reason: the order that received the quote later stops counting against the buyer
  • What was justified by hand remains untouched by automatic closing
  • Adherence to the process in percentage of addressable expenditure with order, quote and contract
  • Ranking of deviations by buyer
  • Blocking policy prevents the order from being issued immediately, with the message telling you what to do instead of just refusing
  • Seven types of policy checked upon issuance: quotation, minimum proposals, contract, approval, valid document, fractional purchase and price above the ceiling
  • Off-line compliance does not stop the company: the request goes through and becomes a finding in the scan, because false blocking costs more than late reporting

Purchasing Budget, Forecast and Treadmill

The year's plan by category, the forecast for the coming months and the routine that keeps everything up to date.

  • Budget by expense category, annually or month by month, without replacing Budget Management that bars requests without funds
  • Budget generated from the previous year's expenditure with the informed correction
  • Plan against completed and against committed, with the balance considering the order already issued
  • Burst line signaled, with consumption in percentage
  • Forecast for the coming months adding current contract to the highest value between open orders and observed recurrence
  • Open order and recurrence never add up, because the order is already part of what the recurrence measures
  • Seasonal factor per learned month of twenty-four months: in a graduation operation, predicting December by the average of a period that includes July is wrong by construction
  • Confidence band instead of dry number, wide in the erratic category and narrow in the stable
  • Limited seasonal factor and only calculated with sufficient history, so that an atypical purchase does not contaminate that month forever
  • Forecast method open on the screen, without statistical black box, and the simple average kept as an option because it is easy to explain
  • Daily treadmill that rebuilds the cube and recalculates price, index, concentration, scorecard, opportunities, recovery, compliance, forecast and alerts in the correct order
  • Incremental load per watermark: only the skills that had a new or changed document, instead of rebuilding two years every morning
  • Complete reconstruction once a month and on demand via the screen, in case the old document was changed without a stamp
  • Duration of each step recorded in the load, so you can see which step degraded when the routine starts to take time
  • Panel reading aggregate materialized at the end of the conveyor, with automatic drop for launch when the aggregate does not cover the period
  • Execution on demand via the screen, with the result of each step
  • Load history with lines inserted, classified, duration and author
  • Outdated cube alerts, uncategorised spending, high priority opportunity stopped, serious open deviation, supplier dependency, delayed initiative and savings waiting for approval
  • Separate permissions to consult, create, approve, delete and configure, checked both on the screen and in the API

Expense Recovery

Money that the company has already paid and can ask for back. On a basis with a few years of bills to pay, it exists.

  • Five scans: same document posted twice, same amount to the same supplier in a short window, note entered by receipt and by hand, bill paid in addition to the amount itself and advance credit never used
  • Note number is unique per issuer: two releases with the same number from the same CNPJ is the strongest finding on the list, and comes with high confidence
  • Slip paid at the bank and downloaded by hand then appears for the excess between the value of the account and the sum of write-offs, discounting interest and fines
  • Value identical to the cent in a fifteen-day window enters with low confidence on purpose: rent and monthly payments repeat value for legitimate reason
  • Each find contains the documents that formed it, so open them both side by side before deciding anything.
  • Own cycle of open, under analysis, confirmed, recovered and discarded, with who analyzed and what concluded
  • What was discarded is never resurrected by the next scan: a queue that is reborn every week is a queue that no one opens
  • Record of the value actually recovered, which is the only number that proves the module's return
  • Thirty-six month window: duplicity from two years ago continues to be money that can be asked for back

Three-Way Conference

Order, receipt and bill count as the same purchase. When all three don't close, someone paid what they didn't receive.

  • Five separate discrepancies: quantity received different from the one ordered, bill price different from the order price, account value different from the order value, account without receipt verified and receipt that never became an account
  • Percentage of purchases that closed in all three ways without divergence, the indicator that every buyer understands instantly
  • Adherence measured by value and not just by count: a hundred small orders closing do not compensate for a large divergence
  • The ruler is the same as the check in the Purchasing module, and not a second tolerance hidden in the report
  • Applied tolerance displayed next to the indicator, with the screen where it was defined
  • An order still in transit does not count as a discrepancy: it only becomes a problem when there is already an account to pay that no one has checked
  • Filter that only shows what is divergent, which is the actual work list

Working Capital and Cost of Delay

How long between receiving the note and paying, how much of it became interest and how much the advance brought back.

  • Average payment period weighted by value, not a simple average of days: one hundred small bills in cash cannot compensate for one large one in ninety days
  • Interest and fine for delay measured separately from the price of the item: it is a real expense caused by the process itself, and was previously invisible
  • Advance discount obtained credited as gain, with the net financial result on both ends
  • Average time and cost of delay by supplier and by expense category, with month-to-month evolution
  • Punctuality in bands: in advance, on time, up to seven days, from eight to thirty and over thirty
  • Simulation of how much cash changes if the category's deadline goes up or down, with the calculation memory open on the screen
  • What connects the module with Treasury and financial planning without duplicating any entry

Contract Management in the Cycle

Balance, maturity, adjustment and price inside versus outside. This is what sustains renewal.

  • Contracted, consumed and to be consumed by contract, with a projection of how many months the balance will end at the current rate
  • Registration consumption and calculated expenditure consumption shown side by side, because the divergence between the two is also information
  • Expiration alert in sixty days, distinguishing the contract that renews alone and requires prior notice
  • Contract expired and still marked as in force explicitly indicated
  • Adjustment window notified in advance, with the contract index alongside the real inflation of your basket
  • Percentage of what the company purchased from the supplier that went outside the current contract
  • Item-by-item comparison of the price inside and outside the contract, on what was purchased both ways in the period
  • Separate loss and gain: one goes to renegotiation, the other goes to the question of why we are buying out

Buyer Panel

The whole module looks at supplier and category. Here the question is who buys it.

  • Portfolio per buyer with expenditure under management, suppliers, categories and orders issued
  • Coverage of classification, spend on order, quoted and under contract by buyer, always on the addressable
  • No one is responsible for the tax that went through their wallet: the denominator is only what that person could negotiate
  • Average time between creating and issuing the order, the indicator that the buyer recognizes as theirs
  • Conformity findings opened by buyer, with serious ones highlighted
  • Personal queue on a screen: what to point out, what to classify and which portfolio suppliers are at risk
  • Appointment of those with relevant volume, so that the buyer with two orders in the year does not pollute the ranking with one hundred percent
  • The resource that takes the module out of the monthly report and puts it into the weekly routine

Classification Assisted by Artificial Intelligence

AI suggests the category with justification and confidence. Nothing enters the cube without someone accepting it.

  • Category suggestion for the unclassified queue, always with the highest value groups first
  • Justification in one sentence citing what in the description or the supplier led to the choice
  • Degree of confidence by suggestion, visible before any decision
  • AI chooses only from the categories that exist: it never invents names and never creates parallel taxonomy
  • Nothing is recorded without human acceptance, and the category can be changed at the time of acceptance
  • When several similar lines fall into the same suggestion, the system proposes the rule instead of the entry, because the rule also resolves the month that has not yet happened
  • Rule is only proposed with high confidence and repeated group: bad guess becomes permanent and silent misclassification
  • What the AI classified is marked as such, distinct from manual and rule
  • Acceptance rate monitored on the screen, which is what tells you whether it is worth continuing to use AI there
  • Bulk rollback of everything the AI classified, without touching what was done by hand or by rule
  • Uses artificial intelligence already configured by the company, respecting the monthly usage limit

Benchmark against the Market

Your basket rose 18%. How much did the market rise? Without that second half, the first number says nothing.

  • Comparison of the internal index with IPCA, IGP-M, INCC and INPC
  • Comparison with commodities and exchange rates (soybeans, corn, wheat, coffee, sugar, cotton, cattle, oil, copper and dollar) from the series the company registers
  • Link by category: food follows the chosen commodity and the others the right index, instead of everything against full IPCA
  • Reference basket with more than one index, weighted by the weight that the company defines
  • Both sides accumulated from the same base competence, which is the classic error in this comparison
  • Competition without a launched index does not enter as zero variation: the accumulated reference stops growing instead of retreating
  • Difference translated into reais on the period's expenses, with the sentence ready for presentation
  • Items that pushed the index the most, ordered by the effect in money and not by percentage change

Sourcing Scenario Simulator

Before opening the RFQ: what happens to price, concentration and risk if I change the base?

  • Consolidation: what happens to the concentration index, price and risk when keeping few suppliers
  • Price savings derived from history, and not from premise: the item purchased from several is now purchased at the best price that the company itself has ever paid among those that remain
  • Volume discount declared as your premise on the screen itself, because there is no way to derive a discount from history for a volume that no one received
  • Simulating with zero discount only shows what comes out of history, without any built-in assumptions
  • Consolidating concentrates, and the simulator says this: the calculated risk appears alongside the economy
  • Removing a supplier by measuring exposure based on the item that only the supplier delivers, not its total value
  • Price impact of switching with the correct sign: removing cheap supplier costs money, and this appears as a negative number
  • Scenario saved, comparable later, and promoted to category plan initiative with one click
  • Initiative born from scenario enters with a fifty percent probability: it was simulated, not negotiated

Group Consolidated

Each branch trades alone. The group's real purchasing power appears here.

  • Total expenditure of the headquarters and branches, with the participation of each unit
  • Suppliers that serve more than one unit, identified by the document and not by registration, which is per tenant
  • Same item purchased at different prices between units, crossed by the standard description because the product code changes from branch to branch
  • Savings calculated as what the group would pay at the lowest price it has ever achieved: it is not a premise, it is the price practiced
  • View restricted to those in the parent company: branch does not see its sister changing a parameter on the screen

ESG and Base Diversity

Percentage spent with a local, small, certified and diverse supplier, with coverage on the side.

  • File by supplier with size, location, ownership diversity, certifications and evidence link
  • Environmental and social risk per supplier, with exposed expenditure added
  • Percentage of spending on local, small, certified and diverse-based suppliers
  • Geographic concentration measured by the same index used for suppliers: how much of spending depends on a single place
  • Base coverage always comes first on the screen: ESG indicator calculated on twelve percent of the base is fiction, and whoever reads the slide will not look for the footnote
  • Queue of the largest suppliers still without a record, which is the fastest way for coverage to increase

Active Alerts and Data Output

An alert that requires remembering to open the screen is not an alert. And the cube spreadsheet is not data output.

  • Eight assignable events: opportunity above floor, serious finding, contract expiring, category over budget, supplier falling to high risk, approved savings not realized, material recovery, and load failure
  • Weekly digest as standard, not one message per event: the ruler that triggers with each finding is silenced in the second week and never turned on again
  • Sending via notice in the panel or by email, using the channels the company already has
  • Subscription value floor, so the digest does not arrive with forty lines of one thousand reais hiding the one for two hundred thousand
  • Subscription restricted to a category, so the owner receives only what is his
  • What has already been warned is not warned again, and sending failure returns the item to the next digest instead of losing it
  • Excel with summary, evolution, category, supplier and detail tabs, with the definitions of each indicator in the first tab
  • Monthly closing package with variation against the previous month and against the same month last year, and the definitions traveling along with it
  • Own reading key for the company's BI, with scope by data set, validity, usage counter and revocation
  • The BI key is never a person's credentials: it does not die when someone leaves the company nor does it become eternal administrative access

Audit of Number and Parameters

Every indicator is a sum that someone will dispute. The path to the document must exist.

  • Definition of each indicator on the panel in a sentence, with the stage used and the percentage base, visible next to the number
  • Path from each indicator to the releases that formed it, already filtered and ready to export
  • Conference that compares the total of the panel with the total of source documents, by stage, with declared tolerance
  • Track of who changed which parameter, when, from what value to which, and what that parameter changes
  • Changing the default stage changes all module numbers, and the screen notifies you of this at the time of the change rather than leaving discovery until the following week
  • Change in scorecard weight recorded, because it changes the score for every supplier at once

Checklist

Everything included in Spend Analytics & Procurement

All 216 features in this area, grouped by module so you can compare systems item by item.

  • Spending taxonomy at three levels, separate from the financial category: accounting talks about consumables, those who negotiate talk about graphics, invitations and special paper
  • Initial model with nine families, including non-addressable spend, and more than forty subcategories typical of a graduation company
  • Category marked as critical has greater weight in the risk and positioning matrix
  • Cube assembled of the purchase order, order item, receipt, advance, bill payable and bank write-off, without typing
  • Three separate stages at all times and never added together: committed to the order, fulfilled to the account payable and paid at the cash outflow
  • Total expenditure and addressable expenditure side by side: tax, payroll, loan, bank fees and transfer between accounts appear in expenses and are outside of what purchases are negotiated
  • Coverage, purchase outside the process and expenses under contract calculated on the addressable, because tax has never been requested and never will be
  • Prorated payable account is distributed by cost center, category, class and branch, and not as a whole in the header branch
  • Return to supplier and credit note deduct the expense instead of disappearing from the account
  • Advance to supplier in the cube for the outstanding balance, which is the real exposure with those who have not yet delivered
  • Purchase in foreign currency converted with the accrual rate frozen at launch: March report does not change because the dollar moved in August
  • Checking the load on one screen: the total of the panel against the total of documents that generated it, by stage
  • Crossing two dimensions on the same screen between category, supplier, cost center, class, buyer, item, type of purchase and competence
  • Pareto curve with accumulated, showing how many suppliers are responsible for most of the spending
  • Clicking on any line opens the entries that form the number, with document and supplier
  • Item purchased loose grouped by standard description, bringing together the various spellings of the same material
  • Reconstruction of the cube fully preserves the hand-made classification
  • Export of the entire cut to a spreadsheet, with all dimensions and order marking, quotation and contract
  • Featured ratings coverage: Without it, everything else on the dashboard would be a well-formatted guess
  • Rules by supplier, catalog product, NCM, financial category, cost center, purchase type or word in the description
  • Explicit priority between rules, with a tiebreaker based on specificity: CNPJ is worth more than a keyword
  • NCM rule accepts prefix, taking the entire code family at once
  • Word rule requires all words in the description, so coated paper does not capture any paper
  • Unclassified queue grouped by supplier and item: four thousand launches saw one hundred and twenty decisions
  • Entire group mass sorting in one click
  • Rule created based on the classification that was just made, so that next month it will be classified
  • Application counter per rule, showing which rules actually work
  • General reclassification that reapplies rules without touching what was manually classified
  • Exclusion rules that tag unaddressable spend by finance category, cost center, supplier or word, with seventeen exclusions ready for day one
  • Own family for non-shopping, with the financial cost of delay falling on them alone
  • HHI concentration index by category and for the entire company, with ready reading of pulverized, concentrated or very concentrated
  • Participation of the largest supplier and the three largest by category
  • Long tail measured by the cut defined by the company, with how many suppliers and how much value
  • Category with single supplier identified and flagged, with emphasis on critical categories
  • Expense that reached accounts payable without going through an order, measured in value and percentage
  • Purchase made outside of a current contract from the same supplier
  • Concentration history saved by competence, to compare before and after negotiation
  • Month-by-month evolution of the number of active suppliers
  • Nine opportunity families: price above reference, supplier consolidation, off-order purchasing, fractional purchasing, single supplier, off-contract purchasing, never-quoted item, long tail, and idle contract
  • Potential value always conservative: applies the best price the company itself has ever achieved, not an imagined discount
  • Noise does not become an opportunity: differences below three percent or of an irrelevant value are discarded at the source
  • Evidence attached to each opportunity, with the numbers that generated it
  • Releases that support the number open on the screen, ready for conversation with the supplier
  • Own cycle of open, under analysis, accepted, captured and discarded, with responsible person and deadline
  • Discarded opportunity requires reason and is never suggested again by scanning
  • Deduplication between scans: the routine runs every day without filling the panel with the same thing
  • Manual recording of what the buyer sees and the rule does not
  • Reference price per item with last price, best price, quantity-weighted average, median and contract price
  • Supplier and date of each reference, to know who the best price was obtained from
  • Price dispersion by item, pointing out where the same material was purchased at very different prices
  • Competency pricing history, separating market growth from supplier switching
  • Comparison by supplier of the same item, with the lowest, average and highest prices charged
  • One hundred base price index per item, per category and for the entire shopping basket
  • Basket calculated with the quantity of the base period, separating price increase from volume increase
  • Twelve-month variation of the basket, which is the real inflation of the company's purchases
  • Item purchased once is excluded from the index, so as not to distort the average
  • Saving classified as hard, soft and avoided cost
  • Baseline pulled from item reference price, with seven declared methodologies
  • Value calculated from baseline, negotiated price and quantity, never entered in the total field
  • Annualized value calculated based on the informed validity
  • Configurable mandatory approval, with approver, date and reason for rejection recorded
  • Approved Savings cannot be edited: changing the number after approval is the end of the panel's credibility
  • Achievement launched month by month, with the total always recalculated from the sum of skills
  • Annual company target and category, with percentage of achievement on the panel
  • Opportunity that turns into saving changes state on its own and leaves the queue
  • Own numbering per year in the standard of the rest of the platform
  • Positioning matrix with supply risk and financial impact, and the four quadrants of non-critical, leverage, bottleneck and strategic
  • Position suggested by the system based on expenditure, concentration and criticality, and adjustable by the category owner
  • Seven strategies per quadrant: consolidate, compete, partner, ensure supply, replace, standardize and internalize
  • Generated category plan already filled with expenditure from the year, previous year, number of suppliers and concentration index
  • Category owner, diagnosis, objectives and risks recorded in the plan
  • Sourcing pipeline in seven steps, from identification to completion
  • Savings weighted by probability, which is the number that the board should look at, and not the gross sum of what was predicted
  • Initiative born from opportunity, which now has responsibility and deadline
  • Initiative with an expired deadline flagged, because its savings continue to count in the pipeline
  • Suppliers, opportunities and category savings gathered on the same plan screen
  • Five axes with configurable weights that add up to one hundred: quality, deadline, price, service and compliance
  • Deadline measured for deliveries against the date promised in the order, with a tolerance in days defined by the company
  • Quality measured in the quantity accepted over that received, with return and damage weighing
  • Price measured against the reference of the item itself, weighted by value
  • Service coming from manual assessment, the only subjective axis, on purpose
  • Compliance from expired document, approval and blocking
  • Axis without data leaves the calculation and the weights are redistributed: lack of information does not result in a zero score
  • Supplier with few deliveries marked as insufficient sample instead of receiving a made-up note
  • Classification from A to D and evolution of the grade by competence
  • Risk in five separate and visible factors: dependence, documentation, performance, concentration and contractual
  • Segmentation into strategic, preferred, approved, tail and to be discontinued, suggested by the system and overwritten by the buyer
  • Hand-defined segment stops being recalculated, with justification and development plan
  • Nine types of policy: mandatory order, mandatory quote, minimum number of proposals, mandatory contract, approved supplier, valid document, anti-fractionation, ceiling price and single supplier
  • Seven default policies ready to go, with limits coming from module configuration
  • Scan that generates the finding with document, supplier, buyer, value and evidence
  • Fractional purchase detected by the month's set of orders from the same supplier, each just below the quotation threshold
  • Supplier document checked on the order date, not today's date
  • Price above the reference with the tolerance defined by the company
  • Finding treated as justified, corrected, accepted or false positive, always with justification recorded
  • Finding already recorded is never rewritten by the next scan
  • Finding whose cause has ceased to exist closes itself, with date and reason: the order that received the quote later stops counting against the buyer
  • What was justified by hand remains untouched by automatic closing
  • Adherence to the process in percentage of addressable expenditure with order, quote and contract
  • Ranking of deviations by buyer
  • Blocking policy prevents the order from being issued immediately, with the message telling you what to do instead of just refusing
  • Seven types of policy checked upon issuance: quotation, minimum proposals, contract, approval, valid document, fractional purchase and price above the ceiling
  • Off-line compliance does not stop the company: the request goes through and becomes a finding in the scan, because false blocking costs more than late reporting
  • Budget by expense category, annually or month by month, without replacing Budget Management that bars requests without funds
  • Budget generated from the previous year's expenditure with the informed correction
  • Plan against completed and against committed, with the balance considering the order already issued
  • Burst line signaled, with consumption in percentage
  • Forecast for the coming months adding current contract to the highest value between open orders and observed recurrence
  • Open order and recurrence never add up, because the order is already part of what the recurrence measures
  • Seasonal factor per learned month of twenty-four months: in a graduation operation, predicting December by the average of a period that includes July is wrong by construction
  • Confidence band instead of dry number, wide in the erratic category and narrow in the stable
  • Limited seasonal factor and only calculated with sufficient history, so that an atypical purchase does not contaminate that month forever
  • Forecast method open on the screen, without statistical black box, and the simple average kept as an option because it is easy to explain
  • Daily treadmill that rebuilds the cube and recalculates price, index, concentration, scorecard, opportunities, recovery, compliance, forecast and alerts in the correct order
  • Incremental load per watermark: only the skills that had a new or changed document, instead of rebuilding two years every morning
  • Complete reconstruction once a month and on demand via the screen, in case the old document was changed without a stamp
  • Duration of each step recorded in the load, so you can see which step degraded when the routine starts to take time
  • Panel reading aggregate materialized at the end of the conveyor, with automatic drop for launch when the aggregate does not cover the period
  • Execution on demand via the screen, with the result of each step
  • Load history with lines inserted, classified, duration and author
  • Outdated cube alerts, uncategorised spending, high priority opportunity stopped, serious open deviation, supplier dependency, delayed initiative and savings waiting for approval
  • Separate permissions to consult, create, approve, delete and configure, checked both on the screen and in the API
  • Five scans: same document posted twice, same amount to the same supplier in a short window, note entered by receipt and by hand, bill paid in addition to the amount itself and advance credit never used
  • Note number is unique per issuer: two releases with the same number from the same CNPJ is the strongest finding on the list, and comes with high confidence
  • Slip paid at the bank and downloaded by hand then appears for the excess between the value of the account and the sum of write-offs, discounting interest and fines
  • Value identical to the cent in a fifteen-day window enters with low confidence on purpose: rent and monthly payments repeat value for legitimate reason
  • Each find contains the documents that formed it, so open them both side by side before deciding anything.
  • Own cycle of open, under analysis, confirmed, recovered and discarded, with who analyzed and what concluded
  • What was discarded is never resurrected by the next scan: a queue that is reborn every week is a queue that no one opens
  • Record of the value actually recovered, which is the only number that proves the module's return
  • Thirty-six month window: duplicity from two years ago continues to be money that can be asked for back
  • Five separate discrepancies: quantity received different from the one ordered, bill price different from the order price, account value different from the order value, account without receipt verified and receipt that never became an account
  • Percentage of purchases that closed in all three ways without divergence, the indicator that every buyer understands instantly
  • Adherence measured by value and not just by count: a hundred small orders closing do not compensate for a large divergence
  • The ruler is the same as the check in the Purchasing module, and not a second tolerance hidden in the report
  • Applied tolerance displayed next to the indicator, with the screen where it was defined
  • An order still in transit does not count as a discrepancy: it only becomes a problem when there is already an account to pay that no one has checked
  • Filter that only shows what is divergent, which is the actual work list
  • Average payment period weighted by value, not a simple average of days: one hundred small bills in cash cannot compensate for one large one in ninety days
  • Interest and fine for delay measured separately from the price of the item: it is a real expense caused by the process itself, and was previously invisible
  • Advance discount obtained credited as gain, with the net financial result on both ends
  • Average time and cost of delay by supplier and by expense category, with month-to-month evolution
  • Punctuality in bands: in advance, on time, up to seven days, from eight to thirty and over thirty
  • Simulation of how much cash changes if the category's deadline goes up or down, with the calculation memory open on the screen
  • What connects the module with Treasury and financial planning without duplicating any entry
  • Contracted, consumed and to be consumed by contract, with a projection of how many months the balance will end at the current rate
  • Registration consumption and calculated expenditure consumption shown side by side, because the divergence between the two is also information
  • Expiration alert in sixty days, distinguishing the contract that renews alone and requires prior notice
  • Contract expired and still marked as in force explicitly indicated
  • Adjustment window notified in advance, with the contract index alongside the real inflation of your basket
  • Percentage of what the company purchased from the supplier that went outside the current contract
  • Item-by-item comparison of the price inside and outside the contract, on what was purchased both ways in the period
  • Separate loss and gain: one goes to renegotiation, the other goes to the question of why we are buying out
  • Portfolio per buyer with expenditure under management, suppliers, categories and orders issued
  • Coverage of classification, spend on order, quoted and under contract by buyer, always on the addressable
  • No one is responsible for the tax that went through their wallet: the denominator is only what that person could negotiate
  • Average time between creating and issuing the order, the indicator that the buyer recognizes as theirs
  • Conformity findings opened by buyer, with serious ones highlighted
  • Personal queue on a screen: what to point out, what to classify and which portfolio suppliers are at risk
  • Appointment of those with relevant volume, so that the buyer with two orders in the year does not pollute the ranking with one hundred percent
  • The resource that takes the module out of the monthly report and puts it into the weekly routine
  • Category suggestion for the unclassified queue, always with the highest value groups first
  • Justification in one sentence citing what in the description or the supplier led to the choice
  • Degree of confidence by suggestion, visible before any decision
  • AI chooses only from the categories that exist: it never invents names and never creates parallel taxonomy
  • Nothing is recorded without human acceptance, and the category can be changed at the time of acceptance
  • When several similar lines fall into the same suggestion, the system proposes the rule instead of the entry, because the rule also resolves the month that has not yet happened
  • Rule is only proposed with high confidence and repeated group: bad guess becomes permanent and silent misclassification
  • What the AI classified is marked as such, distinct from manual and rule
  • Acceptance rate monitored on the screen, which is what tells you whether it is worth continuing to use AI there
  • Bulk rollback of everything the AI classified, without touching what was done by hand or by rule
  • Uses artificial intelligence already configured by the company, respecting the monthly usage limit
  • Comparison of the internal index with IPCA, IGP-M, INCC and INPC
  • Comparison with commodities and exchange rates (soybeans, corn, wheat, coffee, sugar, cotton, cattle, oil, copper and dollar) from the series the company registers
  • Link by category: food follows the chosen commodity and the others the right index, instead of everything against full IPCA
  • Reference basket with more than one index, weighted by the weight that the company defines
  • Both sides accumulated from the same base competence, which is the classic error in this comparison
  • Competition without a launched index does not enter as zero variation: the accumulated reference stops growing instead of retreating
  • Difference translated into reais on the period's expenses, with the sentence ready for presentation
  • Items that pushed the index the most, ordered by the effect in money and not by percentage change
  • Consolidation: what happens to the concentration index, price and risk when keeping few suppliers
  • Price savings derived from history, and not from premise: the item purchased from several is now purchased at the best price that the company itself has ever paid among those that remain
  • Volume discount declared as your premise on the screen itself, because there is no way to derive a discount from history for a volume that no one received
  • Simulating with zero discount only shows what comes out of history, without any built-in assumptions
  • Consolidating concentrates, and the simulator says this: the calculated risk appears alongside the economy
  • Removing a supplier by measuring exposure based on the item that only the supplier delivers, not its total value
  • Price impact of switching with the correct sign: removing cheap supplier costs money, and this appears as a negative number
  • Scenario saved, comparable later, and promoted to category plan initiative with one click
  • Initiative born from scenario enters with a fifty percent probability: it was simulated, not negotiated
  • Total expenditure of the headquarters and branches, with the participation of each unit
  • Suppliers that serve more than one unit, identified by the document and not by registration, which is per tenant
  • Same item purchased at different prices between units, crossed by the standard description because the product code changes from branch to branch
  • Savings calculated as what the group would pay at the lowest price it has ever achieved: it is not a premise, it is the price practiced
  • View restricted to those in the parent company: branch does not see its sister changing a parameter on the screen
  • File by supplier with size, location, ownership diversity, certifications and evidence link
  • Environmental and social risk per supplier, with exposed expenditure added
  • Percentage of spending on local, small, certified and diverse-based suppliers
  • Geographic concentration measured by the same index used for suppliers: how much of spending depends on a single place
  • Base coverage always comes first on the screen: ESG indicator calculated on twelve percent of the base is fiction, and whoever reads the slide will not look for the footnote
  • Queue of the largest suppliers still without a record, which is the fastest way for coverage to increase
  • Eight assignable events: opportunity above floor, serious finding, contract expiring, category over budget, supplier falling to high risk, approved savings not realized, material recovery, and load failure
  • Weekly digest as standard, not one message per event: the ruler that triggers with each finding is silenced in the second week and never turned on again
  • Sending via notice in the panel or by email, using the channels the company already has
  • Subscription value floor, so the digest does not arrive with forty lines of one thousand reais hiding the one for two hundred thousand
  • Subscription restricted to a category, so the owner receives only what is his
  • What has already been warned is not warned again, and sending failure returns the item to the next digest instead of losing it
  • Excel with summary, evolution, category, supplier and detail tabs, with the definitions of each indicator in the first tab
  • Monthly closing package with variation against the previous month and against the same month last year, and the definitions traveling along with it
  • Own reading key for the company's BI, with scope by data set, validity, usage counter and revocation
  • The BI key is never a person's credentials: it does not die when someone leaves the company nor does it become eternal administrative access
  • Definition of each indicator on the panel in a sentence, with the stage used and the percentage base, visible next to the number
  • Path from each indicator to the releases that formed it, already filtered and ready to export
  • Conference that compares the total of the panel with the total of source documents, by stage, with declared tolerance
  • Track of who changed which parameter, when, from what value to which, and what that parameter changes
  • Changing the default stage changes all module numbers, and the screen notifies you of this at the time of the change rather than leaving discovery until the following week
  • Change in scorecard weight recorded, because it changes the score for every supplier at once
← Previous areaPurchasing & SuppliesNext area →Production Planning & Manufacturing

How Spend Analytics & Procurement connects to the rest of the platform

These areas share the same records. Information entered here is available there without exporting spreadsheets or entering it again.

Finance, Treasury & Banks

Collections, accounts payable, cash position, bank integrations, remittances, returns and reconciliation in the same flow.

Budget Planning, Forecasting & Scenarios

Versioned budget, rolling forecast, income statement, cash flow and balance sheet projected in the same model.

Budget Management

From annual planning to the end of the month: budget for the company, the class, the event and the committee's cash flow in a single cycle.

Want to see Spend Analytics & Procurement in your operation?

See a demonstration using your own numbers, with no obligation. A conversation will help you decide whether the module solves your needs.

See plansExplore other areas