Product descriptions from manufacturer text — without duplicate content.

The problem

A distributor onboards 400 products from a manufacturer catalog into their shop. The manufacturer ships the descriptions. The distributor copies them — because it's fast and because the text is technically correct. Two months later, the shop doesn't rank in the top 20 on Google for a single product.

Why: twenty other distributors onboarded the same catalog and copied the same text. Google notices, flags it as duplicate content, and picks a canonical version. Usually that's the manufacturer itself or the oldest reseller. Rarely is it a new shop.

Misconceptions about duplicate content

Misconception 1 — "Swapping a few words is enough"

No. Since at least 2018, Google has been able to detect semantically equivalent text even when individual words have been swapped. "The stove weighs 120 kg" and "This stove has a weight of 120 kilograms" are recognised as the same statement. If the whole paragraph mirrors the source in sentence structure and order of claims, it counts as duplicate.

Misconception 2 — "SEO doesn't matter as much in B2B"

Partly true, partly dangerous. Yes, B2B revenue often comes through existing customer relationships. But new-customer acquisition in B2B is increasingly driven by Google search ("wood stove 8 kW dealer Austria", "promotional items with logo express"). A shop that doesn't show up for those queries loses the whole top of the funnel.

Misconception 3 — "Uniqueness comes from length"

No. A 2,000-word text that's just a long paraphrase of the manufacturer description is still duplicate. Uniqueness comes from additional facts (use cases, typical deployment scenarios in the specific market), from structural reorganisation (different order, different weighting), and from your own tone.

What "structurally new" actually means

A manufacturer description is typically organised as: intro → product features → technical data → accessories note. The reseller's text should carry the same facts but organise them on a different principle.

Example rearrangement

• Lead with the use scenario ("This stove fits living rooms between 40 and 80 square metres") instead of the product line
• Technical data as a list, not as prose
• Add reseller-market-specific notes — delivery windows, installation notes, compatibility with other products in the assortment
• Cut manufacturer marketing claims ("revolutionary", "world-class") entirely; replace with verifiable attributes

How Otto handles it

The SEO generator doesn't receive "the manufacturer text" as input. It receives the extracted facts in structured form: SKU, dimensions, weight, power rating, materials, mechanism, scope of delivery. That fact list is the basis — the source text is never copied, never paraphrased, just ignored.

From that fact list, the model generates shop-ready copy on a predefined structure: a one-sentence lead on use case, a paragraph on product features in proportional weighting, a tabular spec block, a one-sentence note on compatibility or delivery.

The output has minimal semantic overlap with the manufacturer source — different order, different weighting, different tone, sometimes additional facts.

The limits

Otto doesn't replace specialist editors. In niches with high subject-matter demands (medical devices, industrial safety gear), the generated text should get a pass from a domain editor. For standard B2B categories (promotional goods, household appliances, wood stoves, hand tools), the output is ready for productive SEO without rework.

How to measure the effect

Three indicators, 90 days after publish:

• Indexed pages in Google Search Console — should match >95% of the published products
• Keyword positions for SKU-based long-tail queries — typically top 10 within 3 weeks
• Organic impressions — measurable lift versus the prior duplicate-content state, usually a factor of 3 to 8 in the first 90 days

The first indicator is the most reliable. Duplicate content doesn't get explicitly flagged in Search Console, but the indexation rate collapses — from a normal 95-100% down to under 30% on aggressively copied shops.

In short

Building a structurally new text from manufacturer facts takes exactly twice the effort of copy-paste — and a factor of ten less than fixing SEO after the fact. It pays to put that step in the first pass, not in cleanup.