Picture this: revenue misses the plan and the pressure lands on marketing. Someone wants cheaper clicks, someone wants a better conversion rate and someone suggests cutting the budget until performance improves. With everyone looking for an answer, the team starts changing campaigns before it's explained where the revenue went. That's understandable under pressure, although it can leave you spending three weeks fixing a problem you never properly identified. By the next trading meeting, several variables have moved and nobody's quite sure which change was supposed to solve what.
I'd start with a simple question: where, mathematically, did the revenue go? Although the answer won't establish what caused the decline, it gives the investigation some boundaries. If orders are stable while value per order has fallen, for example, a conversation devoted entirely to traffic acquisition is unlikely to get you far. Equally, buying more traffic could make the problem more expensive if fewer visits are converting, while cheaper clicks won't solve a product availability problem. Before recommending a change, we need to explain which part of the revenue tree we expect it to affect and why.
Give the number some structure
GV Marketing's VVR Decisions framework uses Value, Volume and Rate to organise that diagnosis. We define the value we're trying to create, the volume of outcomes behind it and the rates that explain how those outcomes happen. For an ecommerce business, revenue equals orders multiplied by average order value, while orders can be broken down into sessions multiplied by session conversion rate. Provided the definitions and reporting scope are consistent, those relationships give us a usable tree: sessions × conversion rate × average order value = revenue.
The units need to fit the business, so a travel operator might examine bookings, passengers and yield, whereas an app business might follow acquisition cohorts through activation, first transaction and repeat behaviour. There's little value in forcing a website-session equation onto a journey it can't describe. What matters is making the relationships clear enough for the team to know where to look next, with each metric appearing once in the calculation.
Sessions × session conversion rate = orders
Orders × average order value = revenue
Earlier month: 100,000 × 3.0% × £100 = £300,000
Current month: 90,000 × 3.2% × £90 = £259,200
The conversion rate improved. Revenue still fell.
Consider the hypothetical example below, which compares two months using consistent definitions of sessions and orders. Revenue is net of discounts, refunds and VAT. The figures let us account for the revenue movement, although they don't establish the effect of any particular campaign.
| Metric | Earlier month | Current month | Change |
|---|---|---|---|
| Sessions | 100,000 | 90,000 | −10% |
| Session conversion rate | 3.0% | 3.2% | +0.2 percentage points |
| Orders | 3,000 | 2,880 | −4% |
| Average order value | £100 | £90 | −10% |
| Revenue | £300,000 | £259,200 | −13.6% |
The conversion rate looks healthier, yet revenue is down £40,800 because fewer sessions reduced the opportunity to generate orders and lower value per order compounded the loss. Given that picture, what is a general instruction to “fix conversion” supposed to solve? There could still be a conversion problem within a particular segment, but these totals haven't established one. An acquisition-only explanation would also leave part of the decline unexplained because it overlooks the £10 fall in average order value.
To see how those movements combine, we can build an exact bridge by changing one factor at a time. Starting from £300,000, reducing sessions to 90,000 while holding the other factors constant takes revenue to £270,000, a £30,000 reduction. Applying the higher conversion rate then brings it to £288,000, recovering £18,000, before the lower order value takes it to £259,200, a further loss of £28,800. Together, those movements reconcile to the £40,800 decline.
The order of that calculation matters because the factors interact, which means a different sequence would assign different amounts to the individual movements. The bridge therefore explains how the figures reconcile; identifying the causes requires further investigation.
Find out whether the average is hiding the problem
Before treating the higher conversion rate as evidence of a better customer experience, we'd need to understand whether the traffic mix changed. If the business lost a large share of visits from a market that usually converts below the average, the remaining mix could produce a higher overall rate even if performance within each market hadn't changed. Discounts could also increase conversion while reducing order value or contribution, leaving the team celebrating a green number as the business becomes less attractive. When that happens, it's reasonable to challenge what the target is rewarding, because a more defensible dashboard won't pay for the shortfall.
That makes the next step a breakdown of the biggest movements by market, device, product, customer type and acquisition source, wherever the data supports it. Start with the segments responsible for a material share of the change so the investigation stays proportionate: tiny slices can produce dramatic percentages without adding much commercial insight. Before interpreting a movement as customer behaviour, also check for tracking changes, differences between booking and transaction dates, currency treatment and comparable trading periods. Otherwise, a missing event could send the team looking for a broken checkout.
Turn the diagnosis into a choice someone owns
In this example, I'd investigate why sessions and order value fell before recommending a broad budget cut. Visits might have declined because spend fell, demand weakened or inventory became unavailable; alternatively, a reporting change might have altered the recorded total. Lower order value could reflect discounts, product availability, geography or customer mix. Each explanation points towards a different response, so listing the possibilities is only useful if we also establish what evidence would distinguish them.
The eventual decision might be to restore investment in a market with sound customer economics, fix a product or checkout constraint, or stop rewarding a promotion that buys low-value volume. To reach that decision, someone needs to own the investigation and know when a recommendation is due. “We're all aligned” doesn't resolve an explanation that still fails to account for the numbers, and if the evidence can't support the proposed fix, we should say so before the business pays for it. Any recommendation also needs checking against contribution, cash and operational capacity, since recovering revenue at an unacceptable cost simply creates another difficult meeting further down the line.
When a trading update disappoints, the pressure to act is real. A clear revenue tree helps turn that pressure into a focused investigation by showing the team where the loss sits, what remains uncertain and which decision will follow. That gives people a practical way forward without asking them to defend a confident list of changes built on an untested explanation.
If your revenue review keeps producing campaign changes without a clear diagnosis, GV Marketing can help you identify the constraint and decide what to fund, fix or test next.
Read next: work through conflicting budget evidence or check what your ROAS target rewards.