You can automate a technical quotation process without replacing the Excel workbook that calculates it. The workbook stays the calculation engine, and the automation takes over the repetitive work around it: capturing the RFQ, filling the inputs, checking them, routing approvals and preparing the proposal. Engineers still own the assumptions, the price and the sign-off. This guide explains where that work sits, what a workable workflow looks like, and the cases where CPQ software is the better choice.
Why technical sales teams keep using Excel
It is easy to dismiss Excel as a legacy habit. For technical quotations it often has good reasons behind it.
- The logic belongs to the engineers. The workbook encodes how the company really estimates: costing rules, safety factors, installation allowances, lead times. The people who understand that logic can read and change it without a ticket.
- It handles non-standard work. When every project differs, a rigid product model fits badly, while a spreadsheet can be extended for the unusual case.
- It is transparent. An engineer reviewing a price can follow the formulas. That matters when a figure becomes a commitment.
- It is already trusted. Years of past quotations were calculated in it, and changing the calculation tool means rebuilding that trust.
So the question is rarely "how do we get rid of Excel". It is "where does the time go around Excel, and how do we reduce it without breaking what works".
Where the repetitive work sits around the workbook
In most teams the calculation itself is the smaller part. The time goes on the steps before and after it:
- reading the RFQ email and attachments and copying requirements into input cells;
- finding the current version of the workbook, or the right copy for this product line;
- re-typing prices from supplier quotes, price lists or the ERP;
- checking the inputs are complete and consistent before calculating, often after a first wrong result;
- copying results from the workbook into the proposal document and its pricing table;
- chasing colleagues by email for missing inputs and approvals;
- saving the files and updating the CRM or ERP afterwards.
Each is small. Together they decide how long a quotation takes, and they are where re-keying mistakes enter. If you have not measured it, Where the Time Goes in an Industrial Quote shows how, and the Technical Quotation Cost Calculator turns your own figures into annual hours and internal cost.
A workflow that connects around the workbook
A tailored workflow can leave the workbook in the middle and automate the hand-offs on either side. This is a generic picture, not a fixed product: which steps you need depends on your process.
- RFQ arrivesEmail, portal, PDF or drawing attachments
- Structured intakeRequirements captured as fields, missing information listed for follow-up
- Inputs mapped to the workbookA controlled copy of the current version, with only the input cells filled
- Checks before calculatingCompleteness, ranges, units and rule checks, with exceptions sent to a person
- Engineering reviewThe engineer reviews inputs, assumptions and the calculated result
- ApprovalPrices, margins or discounts above your thresholds go to the named approver
- Proposal draftBuilt in your template from the approved outputs only
- Sign-off and filingAn engineer approves, then files go to the CRM or document store
The calculation itself is still your workbook. What changes is who types, who checks and who copies.
Prepare the workbook first
Automation exposes the habits of a workbook, so the unglamorous preparation decides whether it works. A workbook that only its author can drive safely is not ready.
- Inputs, calculations and outputs are on separate, clearly named areas, with the input cells identified (for example as named ranges).
- Formulas are protected, so a changed cell is a deliberate act, not an accident.
- Prices come from one maintained table with an owner and a date, not from figures typed into formulas.
- The workbook carries a version number and date that are recorded with every quotation.
- Assumptions and exclusions are written down, not only implied by the formulas.
- A small test set exists: three to five past quotations with known inputs and approved results, which the workbook must reproduce after any change.
The test set is the most valuable item. It lets you change the workbook, or the automation around it, and see at once whether results moved.
What "validated calculations" should mean
Validation here is not an AI saying the number looks right. It is checks you can read and defend:
- Input checks: required fields present, values within allowed ranges, units consistent.
- Rule checks: combinations your rules forbid are flagged before calculating.
- Integrity checks: formulas unchanged from the approved version; totals reconcile with their components.
- Reasonableness checks: the result compared with similar past quotations, with large gaps flagged for a person.
The calculation itself stays deterministic: the same inputs and the same workbook version give the same price, because the numbers come from your rules and not from a model's guess. Treat the checks as a safety net, not a guarantee. Decades of research on spreadsheet risk, much of it published through EuSpRIG, has found errors to be common; Ray Panko's 2015 paper What We Don't Know About Spreadsheet Errors Today notes that in large spreadsheets at least one incorrect bottom-line value is very likely to be present. That is a reason for checks and for human review, and it is why Keeping Automated Quotes Accurate and Always Reviewed insists on both.
Keep a record of every quotation
When a customer asks how a price was reached six months later, you need more than the final PDF. For each quotation, a controlled workflow can store the workbook version, the input values, the results of the checks, who approved what and when, and which proposal version was sent. That gives engineers something to reproduce and audit. It also means the record holds commercial data, so access and retention need deciding for your environment.
Which decisions must remain with engineering
Automation should prepare and check. These stay with people:
- technical feasibility and design choices;
- the assumptions and exclusions the quotation rests on;
- risk allowances and special-case pricing;
- margin and discount approval;
- the decision to bid at all;
- the final sign-off and what the customer is told.
If a language model is used to read RFQs, its output is a suggestion for a person to confirm, never a value that flows straight into a price. The numbers should come from your workbook's rules.
When CPQ is the better answer
Keeping Excel is not always right. Consider CPQ or another quoting system when:
- your products can be described as options and rules, and the range keeps growing;
- many people edit the same workbook and several versions are already in circulation;
- the workbook depends on fragile macros that nobody except one person can maintain;
- you need role-based approvals and a full audit trail as a requirement, not a nice-to-have;
- pricing must come live from the ERP, or the rules change so often that copying them is a risk.
Automation around Excel can be a useful bridge, but do not invest heavily in a bridge to a place you will leave. CPQ vs Custom Quotation Automation sets out how to decide, and the two can also work together.
Practical limitations and risks
- Where the workbook lives matters. A file on SharePoint or OneDrive, a shared drive and a desktop workbook with macros can each be connected differently, and some set firm limits on what can be automated reliably. Your IT policy decides what is allowed.
- Unstructured RFQs need a person. Reading a customer's documents is the least predictable step. Expect missing or ambiguous items and design for them to be flagged.
- Concurrent use. Several quotations processed at once cannot all write into one shared file. Each needs its own controlled copy.
- Sensitive data. Cost and margin sit in these workbooks. Decide who and what may read them, and what leaves the file.
- Drift. If someone edits the formulas or the price table without telling anyone, automated results can be wrong quietly. Version control and the test set exist for this.
- Key-person risk. If one person understands the workbook, document it before connecting anything to it.
- No guaranteed saving. Time saved depends on your process and on adoption. Measure the baseline first and treat any estimate as a hypothesis.
A sensible first step
- Choose one workbook and one product line, not the whole portfolio.
- Run three past RFQs through the current process and note each manual step and how long it takes.
- Write the test set and fix the obvious workbook problems from the checklist above.
- Pick one boundary to automate first: usually intake to input cells, or approved outputs to the proposal draft.
- Run the new step in parallel with your normal process on real RFQs before relying on it.
For a wider view of the stages involved, read How to Automate Your RFQ Process Without Replacing Existing Tools.
If you would like help with this, Cifral Custom scopes tailored automation around the workbooks and tools you already use. Each implementation is built around your own workbook, environment and approval rules, with your engineers approving the result. It starts with a 30-minute scoping call to look at your process, and some processes will be better served by another approach.