Hyde & Hare | Why Meta customers are worth less

We went looking for the reason. It is not the audience and it is not the tracking. It is which products we put in front of people, which is the one thing here we fully control.

Three things are the same. One is not.

MetaGoogle

Entry price paidtypical smaller first order

£75

£78

Orders in 12 monthsper customer

1.37

1.28

Cost to acquiread spend per customer

£54.47

£49.90

First basketaverage first order

£132

£194

Meta customers arrive at the same price, buy more often across the year, and cost about the same to win. The whole difference is the size of their first order.

And the difference sits entirely at the top end

MetaGoogle

Under £100entry pieces

58%

43%

£100 to £300

36%

42%

£300 and overrugs, hides, beanbags

14%

Share of each channel's first orders. Both sell the entry pieces at a similar rate. Google sells far more of the expensive ones: 14% of its first orders are £300 or more, against 5% on Meta.

Here is why

Where the £35770 of Meta budget actually went, August 2024 to July 2025.

Sheepskin

80.0%

Whole catalogue (dynamic)

9.0%

Beanbags

5.1%

Cowhide

2.9%

Fashion and accessories

2.1%

Multi-product collection

1.0%

BerberGoogle runs four berber campaigns

nothing

Four fifths of Meta spend advertised sheepskin, our entry product. Beanbags and cowhide together took 8%, and berber never appeared on Meta at all. The Google catalogue, by contrast, shows every rug, hide and beanbag to every shopper automatically. Google has been doing the expensive selling for us. Meta was never asked to.

The small test we already ran worked

Recruited by sheepskin ads578 customers

£182

Recruited by beanbag ads26 customers

£245

Google averageall Google customers, for comparison

£244

The customers our handful of beanbag ads brought in are worth £245 over twelve months, level with Google. They did not all rush out and buy a beanbag either; their first orders were small too. Simply showing the expensive pieces attracted a better customer. This is 26 customers, so we treat it as a strong hint rather than proof, which is precisely why it deserves a proper test.

What we would do next

Build a high value lane on Meta. Beanbag, cowhide and berber creative, lifted from roughly 8% of spend to somewhere near a fifth.
Judge it on the customer, not the first sale. These ads will look expensive on day one and pay back over the year, so we hold them to twelve month value.
Leave the sheepskin engine alone. It recruits cheaply and those customers come back more often than Google's. It is not the problem.

Two explanations we tested and ruled out

Meta warms people up and Google closes them

If that were happening, Google's biggest orders would show earlier Meta clicks. They do not. A paid Meta click appears in 3.1% of Google's small orders and 3.5% of its largest, which is flat. Ads that are seen but never clicked stay invisible to this test, so we cannot dismiss the effect entirely, but there is no sign of it.

Google simply harvests people who already know us

Brand searches account for only 10% of Google's new customers, and they are worth £239, no more than the £248 from shoppers who had never heard of us. Google's high value customers are genuinely new.