Ask a B2B leadership team why their digital commerce investment stalled and you will usually get a version of the same answer. It was hard to prove the return.
That is rarely because the return is not there. It is because the return was measured in the wrong place. Digital commerce projects get evaluated on incremental online revenue, which frames the website as a new sales channel competing with the existing one. Under that framing, every dollar the site earns looks like a dollar it took from a rep, and the project becomes a threat to the people whose cooperation it requires.
There is a more accurate framing available, and it changes both the math and the politics.
The number nobody calculates
Most manufacturers and distributors do not know what percentage of their orders are entered manually by an employee. Not orders influenced by a rep, which is a different and legitimate thing. Orders where a customer communicated what they wanted through a phone call, an email, or a spreadsheet, and an employee typed it into a system.
In a lot of organizations that number is well over half. In some it is nearly all of it.
Now add the second number: the loaded hourly cost of the people doing that typing. In technical B2B, order entry and information retrieval are frequently handled by sales engineers, application specialists, and inside sales staff whose expertise is expensive and genuinely differentiating.
The product of those two numbers is your manual order-processing cost. It is almost always larger than the digital commerce investment being debated, and it is currently sitting in a cost line nobody has ever attributed to the website's absence.
Why this framing changes the politics
The incremental-revenue framing makes self-service a competitive threat to sales. Reps hear "the website will take your orders" and they are not wrong to resist, because in that framing they are being asked to fund their own displacement.
The cost-to-serve framing makes it a capacity argument, and the capacity argument is one sales leadership tends to support.
Nobody's best sales engineer wants to spend Thursday afternoon telling a customer the lead time on a part. They want to be solving an application problem for an account that is about to specify a competitor's component, which is work that requires their expertise and produces margin. The routine work is not protecting their job. It is consuming the hours they would rather spend on the work that actually protects it.
The manufacturers who get this right do not position self-service as a replacement for the rep relationship. They position it as the thing that finally lets reps do the job they were hired for. That reframing is not a communications trick. It is a more accurate description of what happens.
What actually has to be true
Self-service does not reduce cost to serve by existing. It reduces cost to serve when it is genuinely faster than calling. That is a high bar, because calling a rep who knows you and answers in ninety seconds is a good experience.
Four things have to be true before buyers switch.
Prices are visible after authentication. Contract pricing, resolved for that specific buyer, on the page. If a buyer has to request a price, the phone is faster and always will be.
Inventory and lead time are real. Not a generic in-stock badge. Position by location and a date they can put in a bid. Unreliable availability data trains buyers to verify by phone, which means you now have both a website and the call. This is where the integration and data foundation underneath the experience stops being a technical concern and becomes a commercial one.
Reorder is trivial. Order history, saved lists, one-click repeat. In high-repeat businesses this single capability moves more volume to self-service than everything else combined. It is a large part of what made Pierce Manufacturing's aftermarket parts experience work for buyers who order the same components repeatedly.
Specifications are on the page. Not attached to it. If confirming a part fits requires downloading a PDF, opening it, and searching for a value, the buyer will call instead. One mission-critical manufacturer we worked with moved every specification out of PDFs and onto structured pages, and picked up AI search visibility as a second-order benefit. Specifications trapped in attachments are invisible to buyers in a hurry and to search engines at the same time.
Miss any one of these and buyers correctly conclude the phone is faster, because it is. Partial self-service does not deliver partial savings. It delivers a website that nobody uses and a support line that still rings.
Building the case
Four steps, none of which require a consultant to start.
Categorize a week of inbound. Split every call, email, and chat into relationship work and information retrieval. The second pile is your addressable volume. We have written a fuller walkthrough of how to read support volume as usability data, including what the composition of that pile tells you about where the friction actually lives.
Cost it honestly. Loaded hourly rate of the people handling it, multiplied by realistic time per contact, multiplied by annual volume. Resist the urge to inflate this. A defensible number that survives scrutiny from finance is worth more than an impressive one that does not.
Establish the manual order percentage. What share of order lines were keyed in by an employee from a customer communication. This is usually the number that gets attention in the room.
Name the capacity, not just the savings. The strongest version of this case is not "we will spend less." It is "we will free roughly this many sales-engineering hours per year, and here is the account work we will point them at." Cost savings arguments compete with every other cost savings argument. Capacity arguments have a growth story attached.
The part that takes longer than the build
Buyers who have been trained for a decade that calling is faster do not change behavior because a portal launched. They change when calling stops being faster, and then only after they have tried the site once and it worked.
That means the rollout matters as much as the build. Reps introducing the portal to their own accounts. Order history migrated so a buyer's first visit already contains their last two years of purchases. Something concrete the buyer gets on the site that they cannot get on the phone, such as real-time availability across every warehouse at eleven at night while they are assembling a bid.
The organizations that get self-service adoption are not the ones with the best portal. They are the ones who made the first visit succeed.
If you already have a portal and the phone still rings, the fastest way to find out why is to look at where buyers actually stall. Our UX site audit combines heuristic review with your own behavioral data and returns a ranked list of what to fix, which is a more useful starting point than another round of internal debate about whether the site is good.
FAQs
Q: How do you calculate cost to serve in B2B eCommerce?
A: Start by categorizing a representative week of inbound contact into relationship work and information retrieval. Multiply the information-retrieval volume by realistic handling time and by the loaded hourly cost of the staff involved, then annualize. Separately, establish what percentage of order lines are manually keyed by employees from customer communications. Together those figures give a defensible baseline. Keep the assumptions conservative, since the case only helps if it survives finance review.
Q: Will self-service cannibalize our sales relationships?
A: The evidence in technical B2B generally points the other way, provided self-service is positioned as absorbing routine work rather than replacing the relationship. Buyers still want expert help for application questions, problem-solving, and negotiation. What they do not want is to phone someone to learn a lead time. Organizations that shift routine transactions to self-service typically redeploy sales capacity into account development, and reps who initially resisted tend to become advocates once the routine volume drops.
Q: What percentage of B2B orders should be self-service?
A: There is no universal benchmark, and chasing one is a distraction. The meaningful measure is movement against your own baseline, particularly the ratio of information-retrieval contacts to relationship contacts and the share of order lines keyed manually. Measure both before the work and again ninety days after launch. Businesses with high repeat-order volume can shift a large majority of transactions to self-service. Businesses dominated by new configurations with quoting attached will shift less, and should target the reorder and status-checking volume specifically.
Q: We already have a portal and nobody uses it. What now?
A: Low adoption is usually a signal that calling is still genuinely faster, not that buyers are resistant to change. Audit the four conditions: visible contract pricing after login, reliable inventory and lead time, frictionless reorder, and specifications published on the page rather than attached to it. A gap in any one of them is generally enough to keep buyers on the phone. It is also worth checking whether existing customers had their order history migrated, since a portal that opens empty gives a returning buyer no reason to come back. A structured UX site audit will identify which of those gaps is costing you the most volume rather than which one is most visible.