Skip to main content

Search Visibility and SEO

Understand how OKNowShop's typed product search ranks results and which inputs you can improve. This is the operational detail behind the Fair Go ranking principles.

Search is personalised: two shoppers can receive a different order for the same query. A subscription plan never changes organic search rank; plans may unlock tools, not placement.

  1. Text matching — Meilisearch finds products matching the query, with typo tolerance and filters.
  2. Personalised reranking — a boost-only multiplicative pipeline reorders the matched hits using merchant quality, shopper context, and locality.

A product that does not match stage one cannot be rescued by a stage-two boost. Accurate title, description, category, attributes, and stock are therefore the foundation.

Published reranking factors

Each matched product starts at 1.0. Applicable factors multiply that score; no factor in this pipeline is a penalty, and ties preserve the original match order.

#FactorMaximum / thresholdControlled byGroup
1Stock availability1.2xSellerMerchant quality
2Seller trust score1.1x above 0.8Seller + marketplace outcomesMerchant quality
3Qualifying customer-rating count1.05x above 10CustomersMerchant quality
4Brand preference matchup to 1.2xShopper behaviourPersonal fit
5Category preference matchup to 1.15xShopper behaviourPersonal fit
6Price-range affinity1.05xShopper behaviourPersonal fit
7Session brand signalsup to 1.2xShopper behaviourPersonal fit
8Session category signalsup to 1.15xShopper behaviourPersonal fit
9Geo proximityup to 2.5x at 0 km; decays across 50 kmSeller + shopper locationLocality
10Recent interaction recencyvariable, 30-day decayShopper behaviourPersonal fit

Deliberately absent: subscription tier, advertising spend, sponsored products, seller GMV tier, and “lowest marketplace price” as a typed-search multiplier. Price-range affinity means fit with that shopper's usual spend band.

Merchant inputs you can improve

Keep sellable stock accurate

An in-stock matched product receives the 1.2x availability boost. Keep stock feeds current and stop selling a listing you cannot fulfil.

Protect seller trust

The trust projection combines source-backed risk and operating signals such as verification history, rejected lines, returns/fulfilment issues, disputes/chargebacks, identity collisions, and security events. Some customer experience inputs are coverage-dependent, so an unavailable score is not zero.

Accept only fulfilable lines, dispatch within the commitment, and resolve returns/disputes through recorded workflows.

Earn qualifying customer ratings

More than 10 qualifying customer ratings activates a small 1.05x boost. The numeric count is a ranking input and is also shown when a customer-only rating is published; review text remains private. Self-service review-request campaigns are not currently available.

Deliver a reliable experience. Do not incentivise ratings or selectively ask only customers likely to rate positively.

Improve stage-one matching

  • Title: use the real brand, product type, variant, and distinguishing attribute in readable language.
  • Description: explain what the product is, what is included, important specifications, compatibility, and exclusions.
  • Category: choose the most specific accurate category.
  • Attributes: complete brand, size, colour, material, dimensions, and other filterable data that applies.
  • Images: use accurate primary and supporting views. Images support listing readiness and conversion; image count is not a separate reranker factor.

The listing-readiness view helps locate missing basics. Its score is not an independent search multiplier.

Seller dispatch and the buy box

Typed-search ranking and buy-box resolution are different contracts. When multiple sellers offer the same product, current fallback buy-box eligibility uses a 30-day seller dispatch window:

MetricGate after at least 5 recent orders
Order acceptance rateIneligible below 70%
Dispatch SLA rateIneligible below 50% when a rate can be computed

New or low-history sellers receive a grace period unless their recent activity is only rejected/cancelled risk outcomes. There is no separate “two consecutive SLA failures” gate.

For eligible sellers, the resolver establishes a competitive window within 10% of the lowest known landed price (or item price when shipping is unknown), then orders qualifying offers by price, estimated delivery, dispatch SLA, and remaining tie-break context. It does not invent a price × SLA discount.

Dispatch tiers

TierHandling commitment
Standard3 business days
Express1 business day
Next DaySame business day

Current measurement runs from order acceptance (or creation where acceptance is absent) to the first purchased shipping label. Carrier transit and delivered-on- time performance are separate. Choose a tier you can consistently meet.

The current portal has no dedicated Fulfillment Health screen. Use Orders and Shipping label state; see Shipping Setup.

Monitoring current visibility

Analytics → Search Insights is not currently available. For now:

  1. keep the listing approved, complete, and in stock in Listings;
  2. test representative shopper queries and filters against the exact metadata;
  3. check seller dispatch and order outcomes in their source workbenches;
  4. treat future search analytics as unavailable until Partner Portal identifies the measured activity, timestamp, and sufficient data.