In B2B, the journey rarely ends with a click on Add to cart. Project volumes, framework agreements, custom manufacturing and tenders all run through quotes - and that is exactly where the digital process breaks down in many shops. 67% of B2B buyers prefer a rep-free buying process (Gartner), yet the moment a request appears, it lands in an email inbox and a spreadsheet. An end-to-end RFQ workflow in your B2B shop - from the quote cart through internal calculation to one-click conversion into an order - closes that gap. This article describes how the process is built technically, what approval logic sits behind it and which metrics turn it into a revenue lever.
Why the quote process is the blind spot in B2B shops
Many B2B shops are excellent on the catalogue side: customer-specific assortments, well-maintained product data, a solid search. But as soon as a buyer requests a line item that is not sold from stock at list price, they leave the system. The request goes to inside sales via a form or a phone call, a spreadsheet calculation appears, a PDF is emailed - and from that moment there is no digital state any more, only an inbox. The shop that models the entire preliminary stage cleanly loses the transaction exactly where the contribution margin is created.
The consequence is measurable. 91% of B2B purchases stall at some point in the process (Forrester, The State of Business Buying), and the third-most-cited reason is not price but the buyer's own organisation's purchasing process (Forrester). At the end of a purchase that looks successful from the supplier's side, 81% of buyers express dissatisfaction with the provider they chose (Forrester). Anyone who fails to see this friction as a process problem keeps optimising product images while the deal fails on a delay.
Time is the hardest lever here. A study of 2,241 companies found that suppliers responding to an online enquiry within one hour were nearly seven times as likely to qualify a lead as those responding just one hour later - and more than 60 times as likely as those waiting 24 hours or longer (Harvard Business Review). The average response time was 42 hours, and 23% of companies did not respond at all (Harvard Business Review). If your quote process runs on email and spreadsheets, this is the order of magnitude you are working in.
A Request a quote button that triggers an email digitises nothing - it merely moves the work into a different inbox. An RFQ workflow is a stateful transaction inside the shop: it has line items, versions, owners, deadlines and a defined transition into an order.
The quote cart: a second path alongside the shopping cart
Technically, an RFQ workflow starts with a second collection point alongside the shopping cart: the quote cart. It behaves like a cart - line items, quantities, variants, notes - but leads into a request rather than to the checkout. Both carts exist in parallel, and the buyer decides per line item which route it takes. A stock item with a stored customer price goes into the shopping cart, a custom length of 800 metres into the quote cart, and both can happen in the same session.
This separation matters more than it sounds. If request and order share the same cart, the customer has to decide before submitting whether the entire transaction is an order or an enquiry - an artificial fork that rarely matches real procurement. A procurement transaction typically contains both. The rule of thirds describes exactly this coexistence: at any given stage of the buying journey, roughly one third of customers prefer in-person interaction, one third remote contact and one third digital self-service (McKinsey, B2B Pulse). The quote cart serves the middle third without excluding the other two.
Line item, not transaction
The order or request decision is made per row, not per session. Rules can automate it: quantities above a threshold, items without a customer price or configurations from the product configurator move into the quote cart automatically.
Context is preserved
Customer number, delivery address, cost centre, project reference and agreed terms stay attached to the transaction. Inside sales does not have to reconstruct anything from an email and starts calculating with complete master data.
Several carts at once
A buyer often handles multiple projects in parallel. Named quote carts - Hall 4, Tender North - can be saved, shared with colleagues and resumed later without line items getting mixed up.
The quote cart is also the natural entry point for bulk requests. If you already run a quick-order form with CSV upload, the same input mask can be routed towards the quote path: 200 rows from the customer's ERP export are validated, known items with a customer price go straight into the shopping cart, unknown or explanation-heavy items into the request. A tender list becomes a structured transaction within seconds instead of an email attachment.
What belongs in the request: quantity tiers, target prices, drawings
The quality of a quote is created in the request, not in the calculation. The more precisely a buyer can describe their requirement, the fewer follow-up questions the transaction costs - and follow-up questions are the most expensive part of cycle time. A good request form therefore asks not only how many, but for the reasoning behind it.
- Quantity tiers instead of a single quantity: The buyer states they need 500, 1,000 or 2,500 units - depending on price. Sales calculates all three tiers in one pass instead of writing three quotes in sequence.
- Target price and competitive situation: An optional field for the desired price is not a disadvantage, it is qualification. It shows whether the transaction sits in your corridor at all before inside sales invests hours.
- Requested date and partial delivery: Whether a customer needs delivery in four weeks or in eight months changes calculation and sourcing route entirely. Whether partial deliveries are acceptable also belongs in the request.
- Attachments and drawings: Technical drawings, bills of quantities, data sheets or photos of a legacy component. Files belong on the transaction, with a version state and access protection - not in a mail attachment that gets lost when forwarded.
- Free-text field for the use case: The sentence replacement for the 2011 system, connection has to fit often saves more time than any mandatory field. The field should be generously sized and not hidden away as optional.
- Contacts and buying group: Who else decides? On average, 13 internal stakeholders and nine external participants influence a B2B buying decision (Forrester, The State of Business Buying 2026).
The last point is regularly underestimated. A quote that goes only to the requesting buyer is a quote that gets forwarded internally, printed and discussed in a meeting without you ever hearing about it. Procurement professionals are now decision-makers in 53% of buying cycles (Forrester), not merely administrators. If your quote exists as a shared link with defined read permissions instead of a PDF attachment, the transaction stays visible to everyone involved - and measurable for you.
Drawings and bills of quantities are often the most sensitive documents in a project. Uploads should be validated server-side for file type and size, stored outside the web root and delivered via signed, time-limited links. A quote portal in which customer A can reach customer B's drawing through an incrementing ID is a data protection incident waiting to happen.
Internal calculation with clear approval limits
Once the request is in the system, the part the customer does not see begins - and it decides margin and speed. The calculation belongs in the same transaction, not in a parallel spreadsheet. Purchase price, surcharges, freight and discount are stored per line item, the contribution margin is calculated live, and approval logic applies automatically. This is the same idea as order approval and authorisation workflows, only with the sign reversed: it is not the customer who needs an approval, it is your own sales team.
At the core sits a matrix of role, discount level and order value. Within defined limits, inside sales decides on the spot; beyond them, the transaction moves to the next stage - but as a task with a deadline in the system, not as a request by email. This is exactly where the 42 hours of average response time come from (Harvard Business Review): not from the calculating, but from waiting for a signature.
| Aspect | Email and spreadsheet | RFQ workflow in the shop |
|---|---|---|
| Calculation basis | Copied table, unclear version | Live prices and terms from the system |
| Contribution margin | Retrospective, often not at all | Visible per line item and transaction |
| Approval | Mail to the manager, open outcome | Rule applies automatically, task with deadline |
| Traceability | Inbox of the person involved | Complete log on the transaction |
| Version state | quote_final_v3_new.pdf | Versioned quotes with full history |
| Handover to ERP | Manual entry, risk of typing errors | Structured handover via interface |
| Measurability | Estimate | Cycle time and win rate per segment |
Approval limits should work in both directions, not only upwards. A discount that pushes the margin below a defined threshold requires justification just as much as an order value above a million. It is also sensible to couple them to customer history: a customer who has paid on time for years and takes a stable volume deserves a different corridor than a first-time request without a credit check. How to model pricing logic, tiers and customer hierarchies cleanly is covered in depth in our article on B2B pricing strategies.
The versioned quote and the one-click order
A quote is not a document, it is a state with validity. It has a version, an expiry date, an author, a set of recipients and a status. That simply cannot be modelled as a PDF - a PDF does not know whether it is still valid, who has read it or whether a newer version exists. In the shop, by contrast, the quote is a page with line items, tier prices, terms and a visible countdown to expiry.
Versioning is not a convenience here but a legal and commercial necessity. When the customer removes one line item and increases the quantity of another, version 2 is created - and version 1 remains traceable, complete with timestamp and editor. Both sides see the same version at any point in time. The same diligence applies to price presentation in a quote as in the shop itself; the pitfalls that price indication rules hold for price comparisons are described in our article on strikethrough prices and discount advertising.
A quote the customer cannot turn into an order with one click is a quote that sits in an inbox for another week.
XICTRON project experience from B2B shop projects
The most important moment of the entire workflow is the transition from an accepted quote to an order. It should be exactly one click: the customer opens the quote, selects the desired tier where quantity tiers exist, adds a purchase order number and delivery address, and confirms. The quote line items are transferred into a cart with the negotiated terms and run through the regular checkout - including purchase on account and credit limit checks. This connectivity is why quote and shop need the same data foundation: a separate quoting tool recreates a break in the chain at exactly this point.
An expired quote should not quietly remain orderable - but it should not vanish without comment either. A clear state has proven effective: quote expired, line items visible, prices marked as no longer valid, plus a Request an update button that restarts the transaction with full context. That is legally clean and creates a new point of contact at the same time.
Handover to ERP and CRM without a break in the chain
The RFQ workflow in the shop replaces neither the ERP nor the CRM - it feeds both. The only question is where the system boundary runs and in which direction data flows. A clear cut has proven itself: the shop owns the request and quote transaction including customer interaction, the ERP remains the truth for prices, stock, terms and the eventual order, and the CRM receives the transaction and activity as a record.
- The request is created in the shop: The transaction gets an ID, the customer sees it in their account, the clock for cycle time starts.
- Master data is pulled from the ERP: Customer terms, tiers, availability and replenishment times come live or from a short-interval cache - details in our article on ERP integration between Shopware and SAP.
- Calculation and approval run in the shop: The transaction stays in the shop until acceptance, because this is where the customer interacts and the version is created.
- The accepted quote becomes an order: Only with the order does the sales order appear in the ERP - structured, with a quote reference, without manual re-entry.
- The CRM receives the transaction: Request, quote value, version, status and follow-up activities land on the customer record so field sales sees the same state as inside sales.
- The document closes the loop: The invoice is created from the order in the required format - see our overview of e-invoicing requirements for online shops.
The same rule applies to the connection as to any interface: every field has exactly one leading system. If the discount can be maintained both in the shop and in SAP Business One or Microsoft Dynamics, discrepancies arise that become visible in the quote - and a wrong price in a quote is considerably more expensive than a wrong price on a catalogue page. How transaction and contact data come together cleanly is also described in our article on CRM integration in online shops.
Follow-up automation: the revenue sits in the follow-up
The most common reason a good quote does not become an order is neither price nor competition - it is simply that nobody asked again. When the transaction lives in the shop, the system can take on that job without anyone maintaining a reminder list. Three days before expiry a reminder goes to the customer, a task is created for the responsible employee at the same time, and when the quote is opened by a previously unknown recipient, sales is notified.
45% of buyers used AI tools during a recent purchase (Gartner), and 94% use AI at some point in the buying process (Forrester). Even so, 69% of buyers turn to a sales rep to validate AI-generated insights (Gartner). Buyers report that sales reps make them feel confident in their purchase decision 32 percentage points more often than generative AI (Gartner) - and those who reach that decision confidence report a high-quality deal twice as often (Gartner). Automation handles the reminding and the logging, people handle the judgement.
This combination explains why the RFQ workflow and the desire for self-service are not a contradiction. 70% of buyers prefer a completely digital, self-service buying experience (Gartner), and the preference for a rep-free process rose from 61% to 67% within a single year (Gartner). The quote transaction in the shop serves both: the customer handles request, review and order independently and on their own schedule, while sales steps in exactly where judgement is needed - visible in the same transaction instead of a parallel mail chain. How far this idea can be carried is shown in our article on the B2B self-service portal.
One frequently overlooked side effect: follow-up automation produces structured rejection reasons. When a customer can close a quote in two clicks as too expensive, project postponed or awarded elsewhere, you build up a data basis over months that no sales statistic delivers. That is the foundation for the metrics in the next section.
Metrics: cycle time, quote-to-order rate, win rate
A process you do not measure is an opinion. As soon as the quote transaction lives in the system, it delivers the metrics without extra effort - not as a monthly report, but in real time per segment, sales team and product group. Three measures carry the process.
Quote cycle time
Time from request to sent quote, split into calculation and approval. Comparing it with the 42 hours of average response time to online enquiries (Harvard Business Review) quickly shows where your process actually stands.
Quote-to-order rate
The share of quotes that become orders - by count and by value. The gap between the two numbers is revealing: a high count rate with a low value rate suggests the large transactions are being lost.
Win rate per segment
The same rate, broken down by industry, customer size, product group and sales team. Only this breakdown turns a metric into a decision about where capacity is deployed sensibly.
It is important to cut cycle time honestly. The clock starts when the request is submitted, not when inside sales opens it, and it stops when the quote is sent, not when the calculation is finished. It is also sensible to look at approval waiting time separately, because that share can often be halved in a single afternoon with a rule change, while the calculation itself has a technical floor.
The quote-to-order rate needs a proper expiry date, otherwise it measures itself into meaninglessness: quotes without expiry stay permanently open and flatter the statistics. Every evaluation therefore needs a defined period and an enforced end state - accepted, rejected or expired. It is also worth looking at the stage before: if buyers cannot find the item they want to request in the catalogue at all, the request does not happen at all - a topic we cover in our article on search relevance and OpenSearch tuning.
A win rate of 30% is neither good nor bad - it depends on industry, sales model and request quality. It only becomes meaningful over time and in comparison between segments. Incidentally, anyone wanting to raise the rate has two routes: win more, or calculate fewer hopeless requests. The second route is often the more profitable one.
The quote process as a revenue lever
German B2B internet trade through online shops and marketplaces reached 509 billion euros, up seven percent year on year, with a further 6.3 percent growth expected for the following year (IFH Cologne). Through these channels, wholesalers and manufacturers realise 12.1 percent of their total revenue (IFH Cologne) - measured against total B2B online trade in Germany of roughly 1.67 trillion euros (Statista/IFH Cologne), that leaves considerable headroom. The share that does not run through the shop today consists largely of exactly the transactions that need a quote.
The willingness is there, as the transaction-size figures show: 73% of B2B buyers are willing to spend more than USD 50,000 per order online - up from 59% two years earlier - 39% would handle more than USD 500,000 and 20% more than one million US dollars digitally (McKinsey, B2B Pulse). At the same time, buyers now use an average of ten interaction channels, up from five in 2016 (McKinsey), and 71% of surveyed companies already sell through at least one form of B2B e-commerce (McKinsey). The question is not whether large transactions run digitally, but whether they run digitally with you.
The RFQ workflow is therefore not an add-on feature but the extension of the shop into the area where money is actually earned in B2B. It shortens the time to quote, makes margin and approvals transparent, keeps the transaction visible to everyone involved and delivers the metrics that let you steer sales. If you would like to explore how a quote process fits into your existing e-commerce landscape and your systems, we are happy to discuss your project - from process analysis through custom development to the connection to your ERP system.
This article draws on data from: Gartner (Sales Survey published in March 2026; survey of around 650 B2B buyers from August to September 2025 - 67% rep-free preference, 70% self-service preference, 45% AI usage, 61% in the year-on-year comparison; plus a Gartner survey from May 2026 on sales reps and AI validation), Forrester (The State of Business Buying 2026, Buyers' Journey Survey of nearly 18,000 global business buyers - 13 internal and nine external participants, 94% AI usage, 53% procurement decision-makers; plus The State of Business Buying 2024 with more than 16,000 respondents - 91% stall rate, 81% dissatisfaction), McKinsey (B2B Pulse Survey, 3,942 decision-makers across the United States and twelve other countries - rule of thirds, ten channels, 71% e-commerce, 73/39/20 percent transaction willingness), IFH Cologne (B2B market monitor - 509 billion euros of B2B internet trade, plus seven percent, 12.1 percent revenue share, forecast plus 6.3 percent), Statista/IFH Cologne (B2B e-commerce in Germany - roughly 1.67 trillion euros total volume) and Harvard Business Review (The Short Life of Online Sales Leads, an audit of 2,241 companies - 42 hours average response time, factor 7 and factor 60, 23% with no response). The figures cited refer to different survey periods and populations and may vary depending on the point in time and market segment.
In our experience, once a meaningful share of requests cannot be served from stock at list price - that is, with project volumes, custom manufacturing, framework agreements or configurations that need explanation. As a rough orientation: as soon as inside sales regularly builds quotes in a spreadsheet and sends them by email, there is typically enough friction to justify a structured transaction. The specific threshold depends on request volume, transaction value and staff effort, and can usually be derived from your own figures within a few weeks.
As a rule, no. A clear division of responsibilities works well: the shop owns the request and quote transaction including customer interaction and versioning, the ERP remains leading for prices, stock, terms and the eventual order, and the CRM receives the transaction and activities on the customer record. What matters is that every field has exactly one leading system - where responsibilities overlap, discrepancies typically arise that become visible in the quote. Which interface fits depends on the system in use.
The sooner the better - though there is no universally valid target time, because calculation effort and complexity vary considerably. A study of 2,241 companies provides the order of magnitude: those responding to an online enquiry within one hour qualified the lead nearly seven times as often as with a one-hour delay, and more than 60 times as often as with 24 hours or more (Harvard Business Review). In practice, a two-stage approach works well: an immediate automatic acknowledgement with a transaction number and a realistic time estimate, followed by the actual quote.
Typically not, provided the transaction takes place in the shop rather than in an inbox. 67% of B2B buyers prefer a rep-free buying process (Gartner), while at the same time 69% turn to a sales rep for validation when it comes to AI-generated insights (Gartner). A good RFQ workflow models both: the customer submits the request, reviews the quote and orders independently on their own schedule, while sales can step into the same transaction where judgement is needed.
Uploads should be validated server-side for file type and size, stored outside the publicly reachable directory and delivered via signed, time-limited links. Access is checked per transaction and permission, not through a guessable ID. A log of who retrieved which file and when is also advisable, along with a defined retention period. This allows a high level of protection to be achieved; a blanket statement about absolute security would be misleading, because the protection level also depends on overall operations.
In our experience, quote cycle time, split into calculation and approval time - it is the fastest to obtain and usually reveals the largest immediate lever, because approval waiting times can often be shortened considerably with a rule change. The second step is the quote-to-order rate by count and by value, followed by the breakdown per segment. An enforced end state per quote is important - accepted, rejected or expired - otherwise permanently open transactions distort the evaluation.