Buying Media When Every Impression Costs More
Media-buying-rising-costs-first-party-data
Paid social costs are rising and AI search is squeezing clicks.
How home shopping retailers can use first-party data to buy media more efficiently and prove what actually worked
Most retailers have felt it without needing a report to tell them. There's no doubt, the same budget buys less than it did. The cost per new customer creeps up. And the platforms' own dashboards insist everything is going brilliantly.
This post looks at why media is getting more expensive, why the usual response makes it worse, and how first-party data changes the maths.
Why it costs more
Paid social is more crowded. Industry benchmark studies for 2026 put the median Meta CPM up by around a fifth on last year, as more advertisers compete for the same feeds. Those are third-party benchmarks rather than Meta's own figures, but almost every dataset points the same way.
There is a new UK surcharge. Since March 2026, Meta has passed on a 2% location fee for ads delivered to UK audiences, to cover the Digital Services Tax. It sits outside your campaign budget, so it is easy to miss.
Q4 is the most expensive time to buy. UK benchmark data typically shows Meta CPMs in Q4 running 20 to 40% above Q1. Every retailer is fighting for attention in the same eight weeks.
Search is giving away fewer clicks. SparkToro's June 2026 study found 68% of US Google searches now end without a click to any website, up from 60% in 2024. When an AI Overview appears, studies from Ahrefs and Seer Interactive have found organic click-through rates falling by around 60%. Transactional retail searches are less affected than research queries, but the buying guides and "best of" content that used to feed the top of the funnel are being hit hard. Fewer free clicks pushes more advertisers into paid search, and that pushes up costs there too.
Why the usual response makes it worse
When costs rise, the instinct is to chase volume. Broaden the targeting. Let the algorithm find more people. Retarget harder.
The problem is that the platforms optimise for the conversions they can see and claim.
That often means:
• Spending money on people who were going to buy anyway.
• Claiming credit for orders your catalogue or email actually drove.
• Prospecting to your own existing customers and calling them new.
The platform report looks healthy. The P&L does not.
Where first-party data changes the maths
Your customer file is the one advantage nobody else in the auction has. Used properly, it lets you buy less and buy better.
1. Suppress before you spend.
Upload your active customers and exclude them from prospecting. Exclude recent buyers from retargeting for a sensible window. It is the simplest efficiency gain there is, and a surprising number of retailers still do not do it consistently.
2. Seed audiences with value, not volume.
A lookalike built from everyone who ever bought from you finds more one-time bargain hunters. A lookalike built from your top-value, repeat, full-price customers finds people who behave like your best customers. Feed the platforms value data where you can, so they optimise for the customers you actually want.
3. Enrich what you know.
Layering lifestyle, life-stage and demographic data onto your own file shows you who your best customers really are: where they live, how they live, what else they buy. That sharpens creative, audience selection and channel choice at the same time.
4. Treat offline as part of the media plan.
Paid social is not the only way to acquire customers. JICMAIL's 2025 tracker showed cold retail direct mail response rates up 21% year on year, with cost per acquisition falling. Inserts in partner parcels and publications, partner mailings and well-selected cold mail all reach the audiences home shopping retailers want, often at a cost per new customer that compares well with social once you measure it honestly.
5. Judge channels on the customers they bring, not the first order.
A channel that brings in customers at £25 who never buy again is worse than one that costs £40 for customers who buy four times a year. The DMA's 2026 Customer Engagement Report looks specifically at how customer value varies by acquisition channel, and the answer is rarely the same as the cost-per-order ranking.
Proving what actually worked
None of this matters if you cannot tell what is working. Platform attribution will always flatter the platform. Last-click attribution will always flatter search. You need a view that sits above all of them.
Holdouts. For CRM, email and mail, leave a random group uncontacted and compare. The gap is the true effect.
Geo tests. Switch a channel on in some regions and off in matched regions, then compare total sales, not platform-reported sales. Postcode-level planning data now makes this practical for mail and door drops as well as digital.
Matchback. Match every order back to every customer contact in the window. With 56% of mail-driven purchases now completed online, according to JICMAIL, this is the only way to stop digital channels taking credit for print.
New customer economics. Report cost per genuinely new customer (checked against your own database, not the platform's) and the 12-month value of each cohort by channel.
You do not need a full econometric model to start. A few well-designed tests a quarter will tell you more than a year of platform dashboards.
The bottom line
Media is getting more expensive, and nothing suggests that will reverse.
The answer is not to spend more to keep up. It is to spend where your data tells you the right customers are, stop paying for customers you already have, and measure the result against the only number that counts: profitable growth in your customer base.
[GB example: a client where suppression and value-based audiences cut cost per new customer.]
Ginger Black helps retailers enrich their customer data to power more precise social, search and offline campaigns, and to measure what actually drove the sale. Speak to the team at info@gingerblack.com.
Sources
Comments