{"id":8580,"date":"2026-08-04T22:15:10","date_gmt":"2026-08-04T22:15:10","guid":{"rendered":"https:\/\/dextora.agency\/?post_type=insight&p=8580"},"modified":"2026-08-05T11:26:48","modified_gmt":"2026-08-05T11:26:48","slug":"less-traffic-more-revenue-ecommerce-in-2026","status":"publish","type":"insight","link":"https:\/\/dextora.agency\/en\/insights\/less-traffic-more-revenue-ecommerce-in-2026\/","title":{"rendered":"Less Traffic, More Revenue: What Actually Happened to Ecommerce in 2026"},"content":{"rendered":"

A store owner opened the year-on-year comparison in June and found something that did not fit the usual script. Visitors were down by double digits. Revenue was up. Nobody had changed the product range and nobody had run a heroic campaign. The obvious question, asked in a slightly suspicious tone, was whether the analytics had broken.<\/p>\n

They had not. This is the shape of trading in 2026, and it is visible in more than one dataset. Fewer sessions, better sessions, more money per session. What follows is what the numbers actually say, where they come from, and the one place where a genuinely spectacular statistic is being used to justify decisions it cannot support.<\/p>\n

The June 2026 picture<\/h2>\n

IRP Commerce publishes a free monthly market data centre, and the June 2026 table is the clearest single snapshot of the pattern. The figures below are taken from that public market data page<\/a>.<\/p>\n\n\n\n\n\n\n\n\n\n\n\n
Metric<\/th>\nJune 2026<\/th>\nJune 2025<\/th>\nYear on year<\/th>\n<\/tr>\n<\/thead>\n
Visitors<\/td>\nWithheld<\/td>\nWithheld<\/td>\nDown 12.23%<\/td>\n<\/tr>\n
Session conversion rate<\/td>\n2.03%<\/td>\n1.85%<\/td>\nUp 9.74%<\/td>\n<\/tr>\n
Revenue per session<\/td>\nGBP 1.89<\/td>\nGBP 1.47<\/td>\nUp 28.48%<\/td>\n<\/tr>\n
Average order value<\/td>\nGBP 127.06<\/td>\nGBP 124.19<\/td>\nUp 2.31%<\/td>\n<\/tr>\n
Cost per acquisition<\/td>\n9.34%<\/td>\n7.93%<\/td>\nUp 17.78%<\/td>\n<\/tr>\n
Cost per session<\/td>\nGBP 0.16<\/td>\nGBP 0.11<\/td>\nUp 45.82%<\/td>\n<\/tr>\n
Bounce rate<\/td>\n41.83%<\/td>\n38.87%<\/td>\nUp 7.59%<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Three caveats before anyone puts this in a board pack. First, IRP’s market data centre covers B2C ecommerce in Great Britain, Northern Ireland and Ireland, calculated from operational trading data recorded inside the IRP platform. It is first-party trading data, which is a strength, and it is a specific set of independent and mid-market merchants, which is a limit. It is not UK ecommerce as a whole, and IRP does not claim it is. Second, the participation percentages published alongside the June table fall on exact fractions of a small denominator, which suggests the benchmark set is a modest panel rather than a market census. Read the direction, not the decimal. Third, it is a live dashboard rather than a dated report, so the same link will show a different month by the time you follow it. The figures above are June 2026.<\/p>\n

One more thing not to do: do not multiply the conversion rate by average order value and expect to land on revenue per session. The metrics use different denominators and the arithmetic will not reconcile. This is normal in published benchmark sets and it is a good reason to treat each line as a trend indicator rather than a component of a model.<\/p>\n

Why revenue per session outran conversion<\/h2>\n

Conversion rose about ten percent. Order value rose about two. Revenue per session rose nearly thirty. Those do not add up unless something else changed, and what changed is the composition of the traffic.<\/p>\n

When total visits fall by twelve percent while the number of orders holds or grows, the visits that disappeared were disproportionately the ones that were never going to buy. That is exactly what the zero-click shift in search produces, and the scale of it is not small: SparkToro’s analysis of Similarweb clickstream data<\/a> put 68.01% of US Google searches in early 2026 as ending without a click to any site. The informational query that used to send a visitor now gets answered on the results page, and the visitor who still clicks through is further down the decision. We went through the mechanics of that shift in our piece on what zero-click search means for a site that lives on organic<\/a>, and the ecommerce consequence is the one in this table. Traffic became a worse proxy for demand, and revenue per session became a better one.<\/p>\n

The cost lines say the same thing from the other direction. Cost per session up nearly forty-six percent and cost per acquisition up almost eighteen percent mean that buying the missing volume back through paid channels is a materially worse trade than it was a year ago. The rational response to more expensive traffic is to convert more of what arrives, which is where the checkout work sits. Our breakdown of the Baymard checkout data<\/a> covers where those percentage points are actually available.<\/p>\n

The AI traffic number everyone is quoting<\/h2>\n

Here is where the year gets misread. In July 2026 Orbit Media published an analysis of GA4 data from 97 sites covering 28.9 million sessions<\/a> between July 2025 and June 2026. The finding that travelled is that AI-referred visitors convert far better than search visitors. That part is real.<\/p>\n

The part that did not travel is the volume. Across 29 million visits, roughly 140,000 came from AI sources. That is 0.5% of all traffic.<\/strong> Orbit’s own framing is that visitors from AI are around three times more likely to convert into leads than other organic traffic, that the median site sees a much wider gap still, and that ChatGPT converts around 2.1% of its visitors against about 0.5% for Google Search, while Google sends roughly a hundred times more of them.<\/p>\n

Two qualifications the study itself supplies and most summaries drop. The sample is business-to-business and lead generation, not ecommerce, so the conversions are enquiries rather than orders. And the 0.5% is almost certainly an undercount, because Google AI Mode and AI Overviews report as organic search and traffic from AI apps often lands in direct.<\/p>\n

You will also see a comparison of 1.91% against 0.71% attributed to this study. Those two values come from two different charts inside it, one an aggregate across the 97 sites and one a screenshot of a single account used as an opening anecdote. They are not a matched pair and we are not going to repeat them as one.<\/p>\n

The arithmetic that ends the excitement<\/h2>\n

Take the friendliest reading of those numbers and do the multiplication.<\/p>\n\n\n\n\n\n\n\n\n
Scenario<\/th>\nAI share of visits<\/th>\nRelative conversion<\/th>\nShare of total conversions<\/th>\n<\/tr>\n<\/thead>\n
Today, as measured<\/td>\n0.5%<\/td>\n3x<\/td>\nRoughly 1.5%<\/td>\n<\/tr>\n
AI traffic doubles<\/td>\n1%<\/td>\n3x<\/td>\nRoughly 3%<\/td>\n<\/tr>\n
AI traffic grows tenfold<\/td>\n5%<\/td>\n3x<\/td>\nRoughly 13%<\/td>\n<\/tr>\n
Conversion improves 10% on everything else<\/td>\nn\/a<\/td>\nn\/a<\/td>\nWorth more than the first two rows combined<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

A tripled conversion rate on half a percent of visits is a rounding error with excellent public relations. It is worth preparing for, because the growth rate is real and the preparation is cheap. It is not worth reallocating a quarter’s budget to, and it is certainly not a reason to under-invest in the ninety-nine and a half percent of traffic that is paying the bills.<\/p>\n

Adobe’s own Digital Insights team reports the same directional story from a much larger panel. In its April 2026 analysis, traffic from AI sources to US retail sites was up 393% year on year across the first quarter, visitors from those sources spent 48% longer on site, and in March 2026 they converted 42% better than non-AI channels such as paid search and email. We have deliberately left out the widely quoted 37% revenue-per-visit figure, because we could not find it in Adobe’s own write-up. Those are vendor figures drawn from Adobe Analytics customers and should be read as such. Adobe’s own summary of the analysis<\/a> describes AI-referred traffic as growing at triple-digit rates and as consistently more engaged, converting better and bouncing less, across more than a trillion visits. Note what it does not say: that the volume is large.<\/p>\n

What this means for the reporting line<\/h2>\n

A dashboard built on sessions will describe 2026 as a bad year for a store that made more money than it did in 2025. That is not a small problem, because budgets get cut on the strength of it.<\/p>\n