planning a shopping feeds budget without ignoring regulatory shifts

What to take away

Decide now whether your 2027 feed budget funds compliance work, platform change, or both. The three cost lines that move most are personalisation consent, product data quality and crawl budget for large catalogues.
Typical monthly planning ranges for those three lines are £300 to £1,200 for consent documentation, £700 to £2,800 for product data quality, and £400 to £1,500 for crawl and feed engineering. These are indicative ranges for a 40,000 SKU catalogue, not quotes.
Google Ads shopping campaign changes arrive through the Google Ads help announcements page, so treat platform release notes as a recurring line item rather than a one-off project.
The Data (Use and Access) Act 2025 will reshape how you document lawful basis for feed-driven personalisation. Build the paperwork before the rules bite.
Product safety duties sit with the seller, not the feed tool. The GOV.UK A to Z guidance is the reference your catalogue team should be checking against.
Large catalogues face a hard technical ceilingcrawl rate. Plan feed file structure and sitemap discipline as a budgeted engineering task.

Why 2027 feed costs are changing

Cost lines that change fastest

Most feed budgets are built from platform fees, agency time and internal hours.

Where the 2027 increase lands

Data quality and consenthalf
Platform subscriptionrest

Take a mid-sized retailer with 40,000 SKUs. The table below sets out typical monthly planning ranges for each line, with the factors that drive the figure. These are planning assumptions, not quotes.

Cost line	Typical monthly range	What drives the figure
Consent and personalisation documentation	£300 to £1,200	Number of personalisation touchpoints, whether a DPIA is written in-house or bought in, and how often the logic changes
Product data quality	£700 to £2,800	SKU count, channel count, attribute rules per channel, and whether fixes are automated or manual
Crawl management and feed engineering	£400 to £1,500	Catalogue size, server response times, thin or duplicate product URLs, and seasonal peaks

Platform subscription sits outside these lines and tends to stay flat or fall as tools compete on price.

The reason is straightforward. Feed tools have become commoditised. The work that now differentiates performance is the quality of the data going in and the legality of how it is used afterwards. Both cost internal time.

Channel count multiplies the effect. Every extra marketplace, affiliate network or comparison site adds another set of validation rules to satisfy. A single attribute fix can take an afternoon per channel to verify, which is why data quality quietly becomes the largest line in the plan.

Where you buy agency time, ask for a breakdown between platform subscription, data work and reporting. Without that split you cannot tell whether a fee rise reflects extra channels or extra compliance work. Those two need different decisions.

If you want the wider market context before you set numbers, the shopping feeds market outlook in England explains how demand and channel mix are shifting.

Fundamentals hold, tolerance narrows

Feed fundamentals do not change. Accurate titles, correct GTINs, sensible categories and clean availability data still drive performance. No platform change removes that requirement.

The teams that absorb 2027 changes best are the ones with a named owner for feed accuracy. Ownership matters more than tooling here. The common failure modes are human: a missed supplier email, a stale category map, an approval that never happened.

The regulatory shifts to plan around

AI personalisation and the ICO

The Information Commissioner's Office has published guidance on artificial intelligence that covers how personal data may be used in automated decision-making and profiling. Feed-driven personalisation sits squarely inside that scope when it targets individuals.

ICO personalisation documentation

What data feeds personalisation logic
Which lawful basis applies
How a customer can object
Write a data protection impact assessment
Update it when logic or sources change

The practical effect for feed teams is documentation. You need to be able to explain what data feeds the personalisation logic, what lawful basis applies, and how a customer can object. That is not a feed tool setting. It is a governance task.

A data protection impact assessment is the usual starting point where profiling is involved. Write it once for the personalisation programme, then update it whenever the logic or the underlying data sources change. Treat it as a living document, not a project artefact.

Within that guidance, the sections on lawfulness, fairness and transparency and the data protection impact assessment section are the ones to read first.

The ICO artificial intelligence guidance sets out the expectations your personalisation programme will be measured against, so read it before you brief a vendor. If your roadmap includes automated product selection, the shopping feeds AI applications in England piece covers where those applications are already in use.

The Data (Use and Access) Act 2025

The Data (Use and Access) Act 2025 changes parts of the UK data protection framework. The ICO maintains a summary of the Data (Use and Access) Act 2025 and what it means for organisations.

For feed teams the relevant provisions are the new Articles 22A to 22D on automated decision-making, the recognised legitimate interests route in new Article 6(1)(ea), and the Article 30 records of processing requirements as amended. If your personalisation relies on inferred preferences, the documentation burden rises.

Nothing in the Act changes how a product feed is structured or transmitted. What it changes is what you must be able to show about the personal data behind your targeting decisions, and how quickly you can produce that evidence if asked.

Scotland, Wales and Northern Ireland operate under the same UK data protection regime, so this is a UK-wide planning point rather than an England-only one.

Product safety duties

Product safety obligations rest on the business placing goods on the market, not on the feed platform. The GOV.UK product safety guidance for businesses is organised as an A to Z by product type, which makes it usable as a catalogue QA checklist.

Map your top categories to the relevant guidance and record the check. When a marketplace asks for evidence of compliance, a dated internal review is more useful than a supplier assurance email.

Budget 16 to 24 hours a year for this work, split across two half-day reviews. The task is not heavy, but it needs a named owner and a place to store the evidence. Feed attributes such as age suitability, warnings and manufacturer details are where the duty surfaces in catalogue work.

Platform and technical pressures

Google Ads shopping campaign changes

Google ships changes to shopping campaigns continuously. The Google Ads new features announcements page is the primary record of what has changed and when.

Treat that page as a standing agenda item in your monthly feed review. Changes to campaign structure, asset requirements or feed attribute support can invalidate a working setup without warning.

For 2027 planning, allocate 8 to 12 engineering hours a month rather than a large annual project. Platform change is now continuous, and annual replatforming cycles leave you exposed between releases.

Note the direction of travel. Shopping surfaces have moved towards more automation in bidding and asset selection. That shifts effort from manual campaign build to attribute completeness, because an automated system can only work with the data you supply.

If you are building the business case, the shopping feeds England market guide for 2027 sets out the channel and cost assumptions worth using.

Crawl rate and large catalogues

Large catalogues hit a technical limit that no feed tool removes: how often search engines will fetch your pages. Google publishes guidance on how to reduce Google crawl rate for sites that need to manage server load.

That guidance matters in two directions. If your feed points to thousands of product URLS, crawl demand is high. If your server is slow, crawl frequency drops and product pages fall out of date in search.

The fix is not a bigger server by default. It is cleaner URL structure, accurate lastmod signals and removing thin or duplicate product pages from the crawlable set. That is engineering time, and it belongs in the 2027 budget.

Measure before you change anything. Log file analysis shows which product URLS are crawled often and which are ignored. That evidence lets you argue for engineering time with numbers rather than instinct, and it gives you a baseline for next year.

Seasonal peaks complicate this further. A catalogue that triples in size for a promotion can overwhelm crawl capacity exactly when visibility matters most. Plan crawl headroom before the season starts, not during it.

How to build the plan

The plan below turns those pressures into a working calendar. It assumes a feed tool is already in place and that the question is what to change around it.

Quarterly review scoring

How does this change rate?

Yes

No action -> skip

Monitor -> watch only

A quarterly review cycle

Set four review points across the year. At each one, check three things: platform release notes, regulatory updates and your own feed error rates.

Use a simple scoring sheet. Rate each change as no action, monitor, or budget. Only the third category reaches the finance conversation.

This keeps the plan honest. Most platform changes need monitoring, not money. The ones that need money tend to be regulatory or structural, and those are visible months ahead.

Give the cycle a fixed date in the calendar rather than a rolling intention. A review that slips by a quarter turns into a firefighting exercise when a platform change lands mid-season.

Agree the triggers that force an off-cycle review. Examples include a platform deprecating a feed attribute, a regulator publishing new guidance, or your error rate tripling in a week. Write the triggers down so the decision does not depend on someone noticing.

The table below sets the four review months and suggested trigger thresholds.

Review month	Focus	Trigger threshold for an off-cycle review
January	Platform release notes from the previous quarter, annual feed error baseline, commencement dates for new data rules	A platform deprecates a feed attribute your campaigns rely on
April	Consent documentation refresh, DPIA update, product safety category check	The ICO publishes new guidance or a commencement date moves
July	Crawl log analysis before the seasonal peak, URL cleanup, lastmod accuracy	Crawl rate on your top 100 product URLs falls by 20% or more
October	Budget submission for the following year, channel mix review, skills gap check	Feed error rate triples in a week or a marketplace rejects more than 1% of items

Set your own baseline for each threshold, then record it in the same sheet you use to score changes.

Skills and ownership

The work above needs three capabilities: feed engineering, data protection literacy and product data governance. Few teams hold all three internally.

Decide which you buy and which you build. Buying feed engineering is common. Buying data protection literacy is riskier, because accountability stays with you regardless of the contract.

Write the ownership map down. For each capability, name the person accountable and the person doing the work. Where those are the same person, that is a capacity risk worth flagging in the budget.

The shopping feeds skills forecast in England looks at which of these capabilities are hardest to hire for, which should inform whether you train or recruit.

A training budget is often cheaper than a recruitment fee. For an illustrative comparison, a team paying £4,000 to upskill two people may spend less than the agency fees attached to one hire. It also keeps the knowledge in the business, where it can be applied to the next platform change.

Data sources to trust

Build your plan on primary sources. Regulator guidance, platform release notes and official statistics beat vendor blogs every time.

That is not a criticism of commentary. It is a question of what you can defend in a budget meeting. A release note with a date is evidence. A summary of a release note is an opinion about evidence.

The shopping feeds 2027 trends: data and sources article lists the sources worth monitoring and how often to check them. Bookmark those pages and review them on the quarterly cycle rather than reading commentary about them.

When a source and a vendor disagree, the source wins. Vendor summaries date quickly and rarely record what actually changed in a platform release. Keep a short list of the pages you trust, and check the list itself once a year, because guidance pages move.

Glossary

Feed: a structured file of product data sent to a channel such as a search, social or comparison platform.
GTIN: Global Trade Item Number, the barcode identifier used to match a product across channels.
Crawl rate: how frequently a search engine fetches pages from your site.
Lawful basis: the reason under UK data protection law that justifies processing personal data.
Personalisation: tailoring what a shopper sees using data about them or their behaviour.
Lastmod: the date a page was last changed, used by search engines to prioritise crawling.
Data protection impact assessment: a documented review of the privacy risks a project creates, used to show how those risks are managed.

Common questions

Does the Data (Use and Access) Act 2025 change feed compliance?

It changes the wider data protection framework that feed-driven personalisation sits inside, particularly around records and automated decision-making. Feed formats themselves are unaffected. Check the ICO's summary of the Act for the current position.

Do I need to change my feed for Google Ads in 2027?

Possibly, but not on a fixed schedule. Google publishes changes through its announcements page as they ship. Review that page monthly and act on what applies to your campaign types rather than planning a single annual migration.

Who is responsible for product safety data in a feed?

The business placing the product on the market. A feed tool transmits data but does not assume the legal duty. Use the GOV.UK A to Z guidance to check your categories and keep a dated record of the review.

How do I stop crawl problems affecting a large catalogue?

Reduce the number of crawlable URLS before you ask for more crawl capacity. Remove duplicate and thin product pages, keep lastmod accurate, and follow Google's published crawl rate guidance. Then measure whether performance improves before spending on infrastructure.