Particular Audience Particular Audience
Talk to us

AI for search, personalisation and retail media

A different store for every customer.

Bring what each shopper wants together with what your business needs.

Search, recommendations, retail media and promotions across your digital store and AI shopping experiences.

How to Monetise Site Search: Coverage, Revenue and Tests

Plan sponsored search with eligibility, query coverage and charged-event economics. Test shopper outcomes and keep retargeting scope clearly defined.

How to monetise ecommerce site search

Monetise site search by selling useful sponsored exposure within relevant product results. Start with query-level eligibility and actual advertiser demand, then measure delivered opportunities, billable interactions and shopper outcomes. The aim is a sustainable service whose commercial value can be reconciled, not the maximum number of ads.

Map where advertising can help

Review queries by shopping task and traffic, including exact product requests, broad categories and descriptive needs. Identify whether each group has relevant products, available stock and advertisers with a reason to participate.

Do not equate every search with saleable inventory. A query may have relevant organic results but no active sponsored candidate. The appropriate response can be an entirely organic result set.

Measure coverage with a useful denominator

Distinguish unique query coverage from traffic-weighted request coverage. Covering many rare phrases does not necessarily cover much traffic, while focusing only on popular queries can hide gaps in specific shopping tasks.

Use a synthetic sample of 10,000 search requests. Suppose 6,000 pass the chosen relevance, stock and campaign-eligibility checks. If an eligible sponsored placement is delivered on 3,000 requests, fill among eligible requests is 50%, while delivery across all requests is 30%. These are different ratios.

For this simplified example, each served request contains one sponsored impression. Real layouts need a separate impression count when more than one placement can be served.

Keep the economics transparent

At an assumed 2% click-through rate, the example produces 60 clicks. If each is billable at £0.50, gross media charges are £30. All inputs are invented. The result excludes invalid-event adjustments, platform fees, operating costs and product-sales effects.

Use actual delivered and billable records when evaluating a live programme. A larger opportunity estimate is not a realised revenue result.

Set the experience rules

Keep paid-placement disclosure clear. Test exact-product queries, unavailable variants and queries where the highest bidder is unsuitable. Agree the amount and position of sponsored exposure, then inspect search completion, conversion and category outcomes alongside media delivery.

PA's Sponsored Search describes keyword and vector targeting. This can be evaluated against the same query set; the use of a particular retrieval method does not remove the need for relevance and commercial tests.

Treat retargeting as a separate decision

PA's Search Retargeting describes carrying recent search intent across onsite surfaces with consent and frequency controls. Define the supported surfaces, expiry and suppression behaviour in the implementation plan. Do not infer external-platform activation from this onsite product.

Use a simple table to reconcile query requests through charged events, and plan an appropriate comparison before claiming that monetisation improved the total business. Unified Ranking is a relevant architecture to evaluate, not a promise that conversion cannot fall.

Illustrative example

{"label":"Synthetic one-placement model, not a forecast","requests":10000,"eligible_requests":6000,"served_requests":3000,"impressions_per_served_request":1,"eligible_fill":0.5,"all_request_delivery":0.3,"assumed_ctr":0.02,"billable_clicks":60,"charged_cpc_gbp":0.5,"gross_media_charges_gbp":30}

Sources

How Particular Audience does it

Frequently asked questions

Is a higher ad-fill rate always better?

No. Fill should be evaluated after eligibility and alongside shopper and commercial outcomes. Increasing fill by relaxing relevance can change the experience and may fail the retailer’s goals.

What is the difference between request coverage and impression fill?

Request coverage counts requests meeting a condition. Impression fill counts served impressions against available eligible placement opportunities. They are equivalent only in a deliberately simplified one-placement model.

Does onsite search retargeting mean offsite advertising?

No. The product described here carries recent search intent across the retailer’s own surfaces. Any external activation requires separately verified channel, partner, permissions and measurement.

Related guides

  • Retail media platform: a buyer's guide: How to choose a retail media platform or software: the capabilities that matter, an evaluation checklist, pilot design and free RFP and business case tools.
  • How do sponsored products work?: Follow a sponsored product from campaign setup to eligibility, ranking, billing and reporting. See why a bid ceiling, charged click and attributed sale differ.
  • Vector search vs keyword search for ecommerce: Compare vector, keyword and hybrid ecommerce search. Understand exact matches, semantic recall, filters and ranking, then evaluate them on your own catalogue.

How-to guides · All guides