Strategy
Part of How to plan a shopping feeds strategy that puts profit before clicks
Why a shopping feeds ninety day plan beats ad hoc tweaks
A practical ninety day plan for shopping feeds, with a before and after table, UK compliance links and a clear sequence for testing, measuring and scaling.
What to take away
- The ICO's guidance on storage and access technologies, updated 2024, means cookie consent choices shape what feed and site data you can rely on, so plan around consent from day one. The law behind it is the Privacy and Electronic Communications Regulations 2003 (PECR) and the UK GDPR.
- A ninety day shopping feeds plan works because it forces a sequence: audit, fix, test, then scale. Ad hoc tweaks move several things at once, so you can never tell which change did the work, and you cannot repeat a win you cannot explain.
- Days 1 to 30 are for the feed itself, days 31 to 60 for channel fit, days 61 to 90 for measured scale.
- Every claim you make in a feed or ad is covered by the CAP Code, written by the Committee of Advertising Practice and enforced by the ASA, and by the Consumer Protection from Unfair Trading Regulations 2008, so keep evidence for prices, stock and delivery times.
- Write down the objective before you touch product titles, because the plan is only as good as the number it is chasing.
Days 1 to 30: the feed audit
The first month is unglamorous. You are not optimising yet. You are finding out what your feed actually contains and what your site is allowed to collect. The alternative, ad hoc tweaks made whenever someone notices a dip in sales, feels faster, but it leaves no baseline, no control and no way to repeat whatever seemed to work.
Start with a full export of the feed and a list of every channel it reaches: Google Merchant Center for Google Shopping, Microsoft Shopping, Amazon, eBay, Meta catalogue ads, TikTok Shop and the main comparison sites. Check titles, GTINs, availability, price and shipping fields against each channel's own specification; in Google Merchant Center that is the product data specification, and every other channel publishes an equivalent. Fix the errors that block products entirely before you touch anything else. A feed with 12% disapproved items will distort every test you run afterwards, so clear those first.
Consent sits underneath all of this. If your site relies on cookies and similar storage for analytics or remarketing, PECR 2003 and the UK GDPR set the standard, and the ICO guidance on storage and access technologies explains what counts as consent and how it must be obtained. Feed work that depends on tracking will fail if that layer is wrong.
Days 31 to 60: channel fit
Once the feed is clean, decide where it should go. Not every channel suits every catalogue. A furniture retailer with long delivery windows will struggle where shoppers expect next-day dispatch, such as Amazon or TikTok Shop, while a small accessories brand may do well on marketplaces with low listing fees. Marketplaces, comparison sites, social catalogues and your own store each set their own rules on titles, images, delivery promises and returns.
Match catalogue to channel
Does the channel suit your delivery promise?
run one controlled test
choose a different channel
This is where the sequence matters more than the tool. Feed management platforms such as Channable, DataFeedWatch, Feedonomics, GoDataFeed and Shoptimised can push one catalogue into several channels and hold the structure steady, but they only amplify decisions you have already made. Use the middle month to match categories to channels, then run one controlled test per channel. Keep the feed structure identical across channels so you can compare results honestly. If you change titles and channel at the same time, you learn nothing.
Comparison sites deserve their own line: they often convert well on price-led ranges but take your brand out of the decision, so test them first on a category where you already compete on price. Your own store is the only channel where you keep the customer data, which matters when you plan the next ninety days.
Compliance belongs here too. Any claim about a price, a discount or a delivery promise must be accurate and evidenced. The CAP Code, written by the Committee of Advertising Practice and enforced by the ASA, governs the ads themselves, while the Consumer Protection from Unfair Trading Regulations 2008 govern the commercial practices behind them. Both apply to feed-based ads as much as to any other format.
Days 61 to 90: measured scale
The last month is about evidence. Pick two or three metrics that match your objective, such as revenue per click, return on ad spend, or new customer rate. Then scale only what has already worked in the test window.
Scale only proven channels
- Pick two or three objective metrics
- Scale only tested channels
- Increase budget in steps
- Hold other channels steady
- Pause failures and note why
- Record every change and date
Resist the urge to scale everything at once. Increase budget on the best-performing channel in steps, and hold the rest steady so you can see the effect. If a channel is not working after a fair test, pause it and note why.
Keep a written record of what changed and when. Dates, channels and outcomes turn the ninety days into evidence you can argue from, and stop the same arguments returning each quarter. That record is also the input for the next ninety days.
One more check before you scale: make sure your ad copy and feed claims still match the rules the ASA enforces. The AdviceOnline library for marketers is a searchable reference for common questions about pricing, availability and comparative claims.
Before and after the ninety days
Before
- Feed errors
- Unchecked, blocking some items
- Consent
- Assumed
- Channels
- All at once
- Reporting
- Clicks only
- Scaling
- Reactive
After
- Feed errors
- Cleared, with a weekly check
- Consent
- Documented and tested
- Channels
- Tested one by one
- Reporting
- Objective-led metrics
- Scaling
- Evidence-based steps
Common questions
The questions below come up in most feed projects, usually somewhere in the second week.
How long should a feed test run?
Long enough to cover a full buying cycle for your category. For fast-moving goods that may be two weeks; for considered purchases, four to six weeks is more realistic. Stopping early usually produces a false answer.
Do I need product feed software to run this plan?
No. A spreadsheet and a clear process will get you through the first ninety days. Software such as Channable, DataFeedWatch, Feedonomics, GoDataFeed or Shoptimised earns its place once you are managing several channels and need scheduled updates, but it is not the starting point.
What if my products keep getting disapproved?
Check the channel's specification against your feed fields one by one, starting with identifiers, availability and price; in Google Merchant Center, that means the product data specification. Most disapprovals come from missing or mismatched data rather than policy problems.
Can I run this plan without a dedicated feed manager?
Yes, if someone owns the task each week. A named owner matters more than a job title. Without one, the ninety days drift and the feed reverts to its previous state.
