Increase average order value

Cross-selling in a phone call depends on whether the person on your side remembers to mention the accessory. Some do, on a good day, when the queue is short. A shop mentions it every time, in the same place, for every customer. Once disciplined pricing sits underneath it, the same number of orders produces more margin.

Why this is harder than it sounds

B2B cross-selling is not "customers who bought this also bought". A buyer ordering a pump needs the seal kit for that pump, not the seal kit that is statistically popular. Suggest the wrong one and you have insulted a professional. Technical buyers extend very little patience to a supplier who appears not to know their own products, and one bad suggestion is enough for them to stop reading the block entirely.

That means relevance has to come from product relationships you maintain on purpose: this article is an accessory for that one, this consumable fits that machine series, this assembly contains four of these. Which is the real obstacle. Nobody has maintained those relationships, because on a printed page the accessory was the next item down and no data was needed. Building them is catalogue work, and it is the whole cost of this goal.

Pricing discipline has a different kind of difficulty. It is political. Every mid-sized B2B company carries discounts that a sales representative granted years ago to save a specific deal, that were never reviewed, and that now apply to everything that customer buys. Nobody has added them up. Adding them up and then deciding which ones stay is a management decision that a pricing feature cannot make for you. The software makes the leakage visible and applies the policy once it is decided.

Two smaller traps. Volume breaks only work where the customer can absorb the volume. A buyer with no storage, a shelf-life limit, or a budget that resets in December will not take 500 because 500 is cheaper per piece, and pushing them there produces returns. And larger baskets often mean fewer orders. A customer who used to place four small orders a month now placing two larger ones is a success that will look like decline on an order-count chart. Fix the reporting before you start, or the project will be judged on the wrong number.

The levers

Put the whole assortment online, not the top 100

The cheapest incremental revenue in B2B is the article the customer did not know you carried. Offline, a customer typically knows the part of your range their sales contact talks about. Online they can see all of it.

What good looks like: every article you are willing to sell is findable, priced and orderable, including the slow movers, the alternatives and the higher-grade version of what they usually buy.

What it depends on: completeness. A product that is technically in the catalogue but has no attributes, no image and no unit of measure is not online in any useful sense.

1. Completeness and readiness
2. Categories
3. Variants and family variants
4. Import products in bulk

Suggest what belongs with it

What good looks like: on the page for a DN 25 valve, the buyer sees the seal kit, the mounting set and the actuator that fit it, and where the relationship is quantified, the basket already proposes the right number of them.

What it depends on: modelling relationships at the level of the family rather than the individual product. Maintaining accessory links one product at a time across 40,000 articles is a project that never finishes; deriving them from families, series and reference entities is one that does.

1. The catalog data model
2. Families
3. Reference entities
4. Assets and asset families
5. Edit pages and blocks

Make volume breaks visible at the point of decision

A buyer who can see that 500 pieces cost less per piece than 400 will often order 500. The arithmetic explains itself and nobody has to negotiate.

What good looks like: the scale is shown on the product page and in the basket, in the buyer's own currency and their own condition, with the next break and what it saves.

What it depends on: scale conditions being modelled rather than approximated, and being combined correctly with customer-specific prices. Getting the interaction wrong is a fast way to sell below your floor.

1. Price models in B2B
2. Scale and volume prices
3. Check which price a buyer sees

Stop margin leaking through manual overrides

What good looks like: discounting is a rule with an owner, a reason and an end date. When somebody wants to deviate, that is a decision someone makes, and it is visible afterwards.

What it depends on: first being able to see realised price against list price, per line, per customer. You cannot govern what nobody has measured.

1. How pricing works
2. Price lists
3. Contract prices for a customer
4. Common pricing problems
5. Reports and exports

A sensible order

Phase 1: make everything visible and correctly priced. Get the long tail of the assortment online and make sure the price shown is the buyer's price. This phase produces order-value growth on its own, with no recommendation logic at all, because buyers discover articles they did not know you carried. It depends only on catalogue completeness, so it can start immediately and does not need any pricing rework.

Phase 2: relationships and scales for the families that matter. Do not attempt the whole catalogue. Take the ten to twenty families that carry most of your revenue, model the accessory, consumable and spare-part relationships there, and switch on the scale display. This phase depends on Phase 1: suggesting an accessory that turns out to be incomplete or unpriced is worse than suggesting nothing.

Phase 3: price governance. Measure realised margin against list across a full quarter, find where it is leaking, and convert the defensible discounts into rules while retiring the rest. This is last for a reason. It needs the data that Phases 1 and 2 generate, and it is the phase that requires management agreement rather than configuration.

Common mistakes

• Recommending on behaviour instead of fit. For technical products, "also bought" produces suggestions that are visibly wrong to the buyer and damage credibility without anyone reporting it.
• Maintaining accessory links product by product. It is achievable for 500 articles and impossible for 50,000. Model it at the family or series level or do not start.
• Discount rules without an owner. A policy nobody is accountable for reverts to ad-hoc pricing within two quarters.
• Pushing volume the customer cannot store. Scale breaks raise value where consumption is real. Elsewhere they produce returns and credit notes.
• Watching average order value alone. The average moves for reasons that have nothing to do with this goal. One large project order distorts a whole month.

How you know it is working

• Lines per order, per customer segment. This moves before revenue does and it is the most direct evidence that cross-selling is working.
• Attach rate for each accessory relationship you built. Of the orders containing the base article, what share contained the accessory. Measure per family, because one bad family will hide inside a global figure.
• Share of order lines placed at or above the first scale break, before and after you made the scales visible.
• Realised price against list price, per line. Look at the distribution rather than the average. The customers sitting far below the rest are the leakage, and you want their names, not a mean.
• Share of the assortment that has been ordered at least once online in the last twelve months. If it stays close to your traditional top sellers, the long tail is technically online but not findable. Go back to Reach more customers.
• Order count alongside order value. Falling order count with rising value per order is consolidation, and it is a good outcome. Label it as such in the dashboard so nobody has to argue about it later.

Related goals

• Move existing customers online: do this first; order value is a second-order lever.
• Decide on real numbers: you cannot govern pricing you cannot see.
• Publish a digital product catalog: the catalogue work that Phase 2 depends on.

What you're building

A catalogue where every sellable article is complete enough to be found and priced, where the relationships between articles (accessory, consumable, spare part, contains four of) are modelled once per family and rendered on every product page, and where scale prices and contract conditions combine correctly in the basket. Underneath it, a reporting view that shows realised price against list per line, so discounting becomes a policy with owners rather than a habit.

flowchart LR
  ERP["Your ERP"]
  subgraph IS["Integration Studio"]
    WF["Workflows"]
  end
  subgraph CS["Commerce Studio"]
    P["Products"]
    AT["Association types"]
    PR["Prices"]
    C["Customers"]
    O["Orders"]
  end
  subgraph XS["Experience Studio"]
    SF["Storefront"]
  end
  ERP -- "articles, conditions" --> WF
  WF --> P
  WF --> PR
  P -- "accessories, fits-with" --> AT
  AT -- "related articles" --> SF
  PR -- "scales, contract price" --> SF
  C -- "segment" --> PR
  SF -- "order" --> O
  O -- "lines, realised price" --> WF
  WF -- "order" --> ERP
  AN["Analytics Studio (coming soon)"]:::planned
  O -.-> AN
  PM["Promotions (planned)"]:::planned
  PM -.-> SF
  classDef planned stroke-dasharray: 5 5

What you need

• The assortment complete for the web channel, including slow movers — Completeness and readiness
• Families that group what behaves alike, so relationships can be modelled once — Families
• Association types for accessory, consumable and spare-part relationships — Association types
• Scale prices modelled as scales, not as separate articles — Scale and volume prices
• Contract prices per customer that combine correctly with scales — Contract prices for a customer
• A report of realised price against list, per line — Reports and exports
• Order-count and lines-per-order reporting in place before the project starts — Metrics that matter

Build it

1. Measure lines per order, order count and realised-versus-list price for one quarter as a baseline — Metrics that matter
2. Publish every article that clears the completeness bar, and chase the ones that do not — Completeness and readiness
3. Check that the price each buyer sees is their contract price, on the long tail too — Check which price a buyer sees
4. Pick the ten to twenty families that carry most of the revenue — Categories
5. Define the association types you will use (accessory, consumable, spare part, contains) — Association types
6. Model the relationships for those families, with quantities where they exist — Reference entities
7. Place the related-articles block on the product page and in the basket — Edit pages and blocks
8. Load scale prices for those families and switch on the scale display — Scale and volume prices
9. Test scale plus contract condition for three real customers, to the cent — Common pricing problems
10. Read the realised-price distribution per customer after one full quarter — Reports and exports
11. Convert the discounts you keep into price-list rules with an owner and an end date — Price lists
12. Retire the rest, customer by customer, with sales in the room — Price list hygiene

Decisions you'll need to make

Related product areas

• Commerce Studio — Products · Families · Association types · Associations · Completeness · Price lists · Price entries · Orders
• Experience Studio — Pages · Sites
• Integration Studio — Import/export · Workflows