What to measure in B2B eCommerce when the order isn't closed online

Your customer visits your website, compares two items, downloads the technical data sheet, and builds a cart. Then, they call their usual sales rep to finalise the order over the phone.

That order doesn't show up in your analytics. And in your ERP—the system where your orders and invoicing live—it appears without a source.

Both tools are telling the truth. On their own, neither one is useful.

This isn't a channel failure. It’s how the industry buys: negotiated pricing, framework agreements, technical validation, and internal approval workflows. If you measure this channel as if it were a retail store, your report will say it isn't working.

In this article, you will see why conversion rate is useless in industrial B2B, which four metrics actually explain your business, and which systems you need to connect to calculate them.

Why conversion rate lies in industrial B2B

Conversion rate divides orders by sessions. In an industrial B2B eCommerce, both terms work differently.

The numerator leaves out all the orders that started online and finished via email. The denominator includes traffic that was never going to buy: competitors, students, job applicants, and the end customers of your distributors.

You are dividing an incomplete number by an inflated one. The result measures nothing.

What that number is actually measuring

It measures what part of your business fits into an online store checkout.

For a manufacturer with negotiated prices and framework agreements, that part is small by design. Not because of a usability problem.

The real risk comes later. If you optimise to increase that figure, you push your website toward a retail store model. And that is the opposite of what a technical buyer needs.

The four metrics that actually explain your B2B eCommerce

These four assume you can identify the customer. Without a private area or login, none of them work. It is the first implementation decision that dictates everything else.

1. Active accounts, not unique visitors

Measure what percentage of your portfolio has logged in over the last 90 days, then segment it by account age, size, and whether they have an assigned sales rep.

There is an extra signal: an account that stops logging in is a churn warning that arrives before sales even detects it.

2. Technical inquiry depth per account

Measure downloaded data sheets, comparisons, part number searches, and time spent on product pages.

In the industrial sector, technical inquiry isn't just a step before the purchase. It is the purchase. When a customer downloads three data sheets from the same range in one week, they have a project on their desk.

This metric depends on the quality of your B2B product catalogue. If the data sheets are incomplete, you aren't measuring interest: you're measuring frustration.

3. Days between first visit and order

Measure the actual cycle, not the session.

If the cycle shortens, the platform is working, even if conversion remains flat. This is the metric that best translates the channel into management language: shorter cycle days mean more cash flow.

4. Assisted orders: how much offline business started online

The most important and the most difficult.

Calculate what percentage of orders coming in via phone or email originate from an account with online activity in the preceding days.

That percentage is the real value of your eCommerce. And it is usually the number that changes the conversation in the boardroom.

You cannot measure what isn't connected

This is the real reason why almost no manufacturer has these four metrics. It is not an analytics problem. It is an architectural problem.

Business data doesn't live on the web. It lives scattered across five systems that often don't talk to each other.

What each system contributes

• PIM. The system that centralises product information. It provides the complete data sheet and clean references. Without it, technical query metrics mean nothing. Here we explain how to choose a PIM based on your catalogue.
• ERP. It provides the order, the amount, and the customer code. It is the source of truth for sales.
• CRM. It provides the account, the assigned sales representative, and the open opportunity. Without this, you cannot calculate assisted orders.
• CPQ. The quote configurator. It records the configuration and the offer, which is where industrial purchasing decisions are made. We cover this in industrial quoting architecture.
• DAM. It stores blueprints, certificates, and technical documentation. Every download is a signal of purchase intent that is lost if files are just sitting in a loose folder.

This is the difference between implementing an eCommerce site and connecting the entire ecosystem. Without the latter, you have a storefront you cannot evaluate.

The customer ID is the only key.

Cookies expire in weeks and don't survive device changes. Don't use them for this.

What does survive is the customer ID, the same one your ERP uses. That code connects all five systems.

What you can measure without integrating anything yet

If you're far from that architecture, start anyway. There are two questions you can answer this month without touching any integrations.

1. The first: which customers have shown online purchase intent? Your website answers that on its own. Measure quote requests and account-based downloads: these are commercial intent events that don't depend on anyone else. That gives you the part of the assisted order process that happens before the call.
2. The second: how many orders this month came from those customers? Sales has that answer. Ask them. It will be an estimate, since it comes from a salesperson's memory rather than a system. It still works: it gives you an order of magnitude and, in the process, makes sales a part of the measurement.

Put both halves together in a spreadsheet and you have your first data point.

That's enough for the first few months, and enough to know if it's worth connecting the ecosystem. Don't start by building a data warehouse just to find out.

Klinger case study: when the salesperson steps in mid-order

Klinger manufactures sealing, fluid control, and monitoring solutions, and operates in over 40 countries. Their challenge wasn't to sell more online. It was to gain efficiency in order management without losing the personal sales touch.

We implemented a self-service B2B portal integrated with their Microsoft Dynamics. Customers buy and request quotes on their own, and the sales team can intervene in real time when the order requires it.

The ERP integration wasn't just a technical requirement. It was the condition for the portal to function as a sales channel rather than just a catalogue with a contact form.

That design decision breaks conversion rate as a metric. An order the customer completes alone and an order the salesperson closes after the customer configures it are both the same success. The latter doesn't appear in any standard analytics.

Take a look at how we evaluated the project with Klinger:

• Platform adoption by their customers
• Improved efficiency in order management
• Time saved for the sales team
• Continuous platform evolution

None of these four are conversion rates. The success criteria were operational, not transactional. Agreeing on that before launching is what prevents a healthy channel from looking like a failure.

Which decision each metric unlocks

A dashboard that doesn't change any decisions is just a pretty report. These four metrics actually drive the eCommerce strategy.

• Low active accounts in a specific segment. The problem isn't the platform. It's the onboarding for that segment. The investment goes into portfolio activation, not rebuilding the website.
• High technical inquiries and low orders. The customer finds the product and gets stuck on the price or the offer. That's where a CPQ, no more traffic.
• Long cycle despite high inquiry volume. Lack of information to decide without calling. This is a sign of incomplete product data, and points to PIM and DAM.
• High assisted order volume. The channel is already working. The next investment expands self-service so that those orders close automatically, freeing up sales reps for new accounts.

That interpretation is what turns four numbers into a digital roadmap.

The most common mistakes we see in industrial projects

None of these mistakes are technical. They are decisions made in the first few weeks of the project that take their toll months later, when correcting them becomes expensive.

All five have the same effect: they leave the channel with no way to prove it works.

1. Evaluating the channel at six months using B2C metrics: The industrial cycle is longer than the first results review.
2. Launching without a private area: Without a login, there is no customer code, and without a customer code, you cannot measure any of the above. Recovering it later is costly.
3. Implement the platform and leave integrations for phase two: Phase two is postponed. And without an ERP connected, eCommerce is blind.
4. Measuring eCommerce against the sales team: If the dashboard pits the digital channel against sales, sales won't adopt it. And without sales, assisted order data never arrives.
5. Not agreeing on what an assisted order is before starting: That definition is negotiated with sales, not decided by marketing.

Let's talk about your B2B eCommerce

At Novicell we implement B2B eCommerce, connect the ecosystem behind it, and define the digital strategy that supports it.

Tell us how your customer buys and we'll tell you what you should be measuring.