Luxury and high eCommerce purchases rarely occur in one sitting. Customers can spend weeks (sometimes months) researching, comparing, reading reviews, visiting stores and speaking to sales associates before they commit.
By the time a transaction completes, a lot of the touchpoints that influenced that decision have dropped out of the marketing attribution model being used to measure them.
If budget decisions are being made from GA4, eCommerce platforms, or ad platform data, they’re being made on an incomplete picture, and that’s the backdrop against which we’d encourage you to read the numbers below.
We discuss:
- Traffic vs purchase share by channel
- Forms versus calls
- Purchase rate by channel
- Ad spend benchmarks
💡TL;DR
• Direct and paid search take the highest share of purchase in luxury retail (30.1% and 27.8% respectively), but that’s often the last touch taking value for work done by channels higher up the funnel.
• Calls make up nearly a quarter of all purchases (25.1%). Missing that data out of reporting doesn’t just create a blind spot, it feeds bidding and optimisation with an incomplete view of value, and the same risk applies to any offline purchase, not just calls.
• Google Paid delivers purchases at £24.2 each, while Instagram Paid and Facebook Paid sit at £312.2 and £329.4. Social paid often plays a different role to direct response though, so it’s worth measuring against that, rather than search, as the benchmark.
Traffic share and purchase share
| Channel | Share of purchases | Share of sessions |
| AI Referral | 0.3% | 0.1% |
| Direct | 30.1% | 28.0% |
| 6.2% | 6.0% | |
| Organic Search | 22.4% | 24.1% |
| Paid Search | 27.8% | 28.6% |
| Referral | 1.6% | 1.7% |
| Social Organic | 0.9% | 1.2% |
| Social Paid | 10.6% | 10.2% |
Direct traffic is the standout here, taking 30.1% of purchases from 28.0% of sessions, which on the surface looks like the best performing channel in the account. In our experience, that’s rarely the full picture.
Direct traffic in a luxury context is often the final step of a much longer journey that started somewhere else entirely, a customer who saw a paid social ad three weeks ago, browsed a couple of times from organic search, then typed the brand name straight into their browser once they were ready to buy.
Attribution models that only look at the last touch will hand all the value to direct, when really it’s absorbing the work done by channels further up the funnel.
Paid search tells a similar story in reverse. It converts almost in line with its session share (27.8% vs 28.6%), which suggests it’s doing a genuinely proportionate job of capturing existing demand, though it’s worth remembering that some of that demand was created by other activity the customer saw first.
Social paid is worth a closer look too. It brings in 10.2% of sessions and converts 10.6% of the time, a small but real over performance, and one that often gets undervalued in luxury retail and eCommerce because social is assumed to be a brand awareness channel rather than a purchase driver. The numbers suggest it’s doing more of the latter than it gets value for.
💡 Pro tip
Channels like this are exactly where a narrow, click based view of performance falls short. We’ve put together a guide on unifying measurement for ecommerce, covering how to bring online and offline channels, long consideration journeys and touchpoints that influence a sale without ever earning a click into one picture, so budget decisions are based on what’s actually working rather than what’s easiest to see.
Download the eCommerce unified measurement guide
Share of forms versus calls
| Industry | Share of forms | Share of calls |
| eCommerce & Retail | 74.9% | 25.1% |
A quarter of all purchases in luxury retail and ecommerce still come in as phone calls rather than forms. For a sector that’s often assumed to be purely transactional and self service, that’s a meaningful chunk of demand happening in a channel that’s much easier to under report than an online form fill.
If call tracking isn’t wired into the attribution model, a full quarter of purchases risk being lost, or worse, misattributed entirely.
The risk here goes further than a reporting gap though. It changes what the ad platforms themselves are optimising towards. Google and Meta’s bidding algorithms are only as good as the purchase data they’re fed, and if that data only includes what happens on the website, form fills, checkout completions, add to basket events, then that’s the only version of value the algorithm can ever learn from. A quarter of real, high value demand is happening in a channel the algorithm simply can’t see.
Budget drifts towards audiences and keywords that generate form fills, even if those same audiences rarely go on to actually buy, while the traffic that’s genuinely driving revenue through the phone gets no acknowledgement and no additional investment. Over time, that’s not a small inefficiency, it’s the algorithm working exactly as instructed, just against an incomplete definition of what a good outcome looks like.
The same logic applies to any offline purchase, not just calls. In store purchases, appointment bookings, private client enquiries, anything that closes outside a browser session sits in the same blind spot.
Feeding those purchases and their actual value back into the ad platforms, rather than just the online actions that led up to them, is what lets the bidding algorithms optimise towards real commercial outcomes instead of a partial proxy for them. Without that, even a well built campaign is only ever operating at a fraction of its real capacity, because the system making thousands of micro decisions a day about where to spend the next pound simply doesn’t know what actually matters.
💡 Pro Tip
We saw this play out with a health consumer brand running paid media across Google, Meta, connected TV, Amazon, Walmart and QVC at the same time. Last-click attribution was attributing Google for purchases that TV had influenced weeks earlier, while offline retail sales sat in a completely separate silo. Once online and offline spend were brought into a single model, they discovered that Amazon alone contributed 11% of revenue and were able to attribute 53% of revenue back to paid media channels.
Read the full case study here
Forms vs calls by channel
| Channel | Forms | Calls |
| AI Referral | 89.2% | 10.8% |
| Direct | 89.8% | 10.2% |
| 92.0% | 8.0% | |
| Organic Search | 77.9% | 22.1% |
| Paid Search | 74.9% | 25.1% |
| Referral | 83.6% | 16.4% |
| Social Organic | 95.7% | 4.3% |
| Social Paid | 94.2% | 5.8% |
The split by channel is where this gets genuinely useful for planning.
Paid search and organic search customers are far more likely to pick up the phone (25.1% and 22.1% of their purchases respectively) than customers arriving through email or social, who overwhelmingly convert through forms.
Our read on this is that search traffic tends to carry higher intent and more specific questions, the kind of thing a customer wants answered by a real person before they spend a significant amount of money, whereas social and email traffic is often earlier in the journey and happier to fill in a form and wait.
Social organic and social paid sit at the other end, converting through forms 95.7% and 94.2% of the time. That’s consistent with what we generally hear from clients, that social driven customers are still browsing and comparing rather than ready for a detailed conversation.
Purchase rates by channel
| Channel | Purchase rate |
| AI Referral | 4.5% |
| Direct | 2.4% |
| 2.3% | |
| Organic Search | 2.1% |
| Paid Search | 2.2% |
| Referral | 2.0% |
| Social Organic | 1.7% |
| Social Paid | 2.3% |
| Total | 2.5% |
AI referral’s 4.5% purchase rate is nearly double the account average of 2.5%, and it’s the standout figure in this entire dataset.
Someone arriving from an AI assistant or AI powered search tool has usually already done a fair amount of the comparison and research work upstream, inside the conversation itself, before they ever land on the website.
They’re arriving closer to a decision than someone starting a search from scratch, which would explain why they convert at such a higher rate. Worth keeping an eye on as this channel grows into 2027, both in terms of volume and how it’s tracked.
Direct sits second highest at 2.4%, again likely benefiting from being the last step in a longer journey rather than being inherently more persuasive as a channel. Social organic is the lowest converting channel at 1.7%, which fits its role earlier in the funnel, building awareness and consideration rather than closing sales.
Ad spend benchmarks
| Platform | Average spend | Total CPC | Total cost per purchase |
| Bing Paid | £1,407.47 | £0.61 | £62.42 |
| Facebook Paid | £6,672.77 | £3.04 | £329.44 |
| Google Paid | £22,869.02 | £4.17 | £24.17 |
| Instagram Paid | £1,730.16 | £7.80 | £312.17 |
| Pinterest Paid | £446.46 | £1.27 | £48.03 |
| YouTube Paid | £401.36 | £17.94 | £208.11 |
This is probably the table that will get the most attention from anyone holding a budget. Google Paid takes by far the largest share of spend, at an average of £22,869.02, and it earns that position, returning purchases at a genuinely efficient £24.17 each, comfortably the lowest cost per purchase of any platform in the table.
Bing Paid is the quiet performer here. It takes a fraction of the budget of Google, around £1,407.47 on average, but delivers purchases at £62.42, and with a total CPC of just £0.6, it’s clearly an underused channel in a lot of luxury accounts.
Pinterest sits in a similar bracket, £48.03 per purchase on relatively modest spend, which tracks with what we tend to see, that visually driven platforms perform well for products that benefit from being seen rather than searched for.
Facebook and Instagram are the two channels where cost per purchase runs highest, at £329.44 and £312.72 respectively, despite Instagram commanding a notably higher CPC of £7.8. YouTube’s CPC of £17.94 is the highest in the table by some distance, though its cost per purchase of £208.11 puts it ahead of both Meta platforms.
None of this means these platforms aren’t worth the investment, luxury brands often use paid social and video for brand building rather than direct response, but it does mean the return needs to be judged against the right objective rather than compared directly to search.
Final thoughts on luxury retail and eCommerce measurement
Taken together, these numbers point to the same underlying issue. Direct and search look like the strongest channels, but that doesn’t necessarily mean they’re responsible for the full weight of the purchases they receive. Other channels are likely influencing customers earlier in the decision-making process, without that influence being reflected in the attribution.
A quarter of purchases are happening over the phone, and offline purchases add another layer that most attribution set ups simply can’t see, which means both reporting and the bidding algorithms behind it are working from a partial picture.
AI referral’s early over performance and the wide spread in ad spend efficiency only reinforce the point, the channels that look best or worst on paper aren’t always the ones actually driving the value.
The fix isn’t complicated in principle, even if it takes some work in practice.
You need to capture every purchase type, calls and offline sales included, match that back to real customer and order data, and feed it back into the platforms making bidding decisions.
But that still only covers the interactions you can directly observe. Statistical modelling can help account for the influence of impressions, upper-funnel activity and offline interactions that never produce a trackable click, giving you a more complete picture of what’s actually driving demand.
Once that’s in place, budget can start moving towards what’s genuinely working, including the activity that influences customers without being the easiest thing to measure.
If you’d like to see how this kind of measurement looks for your own luxury or high value ecommerce brand, we’d be happy to walk you through it. Book a demo with Ruler and we can show you what your own numbers are telling you.
