DEMO – Illustrative analysis
My Shopify sales have dropped 20%. What should I do?
All figures, findings, arguments and conclusions in this analysis are fictional and for demonstration only. No live research was performed.
Research summary
Sales are down about 20% on the previous 30 days while sessions are almost flat. The fall comes mainly from lower conversion, concentrated on mobile and at checkout. Traffic mix has also shifted towards lower-converting paid social, which explains part of the drop. Several possible causes cannot be tested with the data supplied.
Revenue fell about 20% over the last 30 days versus the previous 30 days (£41,200 to £32,960).
User-supplied sales data (demo)
Sessions were roughly stable (32,500 to 31,800, about -2%). Orders fell from 780 to 636 (about -18.5%). Conversion fell from 2.40% to 2.00%. Average order value moved from £52.80 to £51.80.
User-supplied traffic and conversion data (demo)
Mobile conversion fell from 1.9% to 1.5%; desktop from 3.4% to 3.2%. Mobile is about 70% of sessions.
User-supplied conversion data by device (demo)
Paid social rose from about 21% to 30% of sessions and converts at roughly 0.9% to 1.0%. Conversion from all other sources fell from about 2.77% to 2.47%. A rough estimate is that the traffic-mix shift explains about 0.16 of the 0.40-point conversion fall (around 40%).
Demo calculation from supplied traffic-source data (illustrative estimate)
Checkout completion fell from 61% to 54%. The free-shipping threshold was raised from £40 to £60 about four weeks ago.
User-supplied funnel data and store settings (demo)
Mobile product-page load time (largest contentful paint) rose from about 2.1s to 3.4s after a product-review widget was installed about six weeks ago.
User-supplied product-page performance data (demo)
Other channels remained comparatively healthy: email about 3.6%, direct about 2.6%, organic search about 2.3%.
User-supplied conversion data by source (demo)
Independent perspectives
Strategist
Conversion and offer friction is the primary problem. Fix the funnel before spending more to bring people into it.
Initial recommendation
Hold advertising spend. Prioritise mobile page speed and the shipping threshold.
Analyst
The decline has two components: a traffic-mix shift and a genuine on-site conversion fall. Neither should be assumed without segmenting the data.
Initial recommendation
Segment conversion by source and device before committing budget to any fix or campaign.
Sceptic
Traffic quality is the main issue. Stable session volume hides a change in who is arriving.
Initial recommendation
Review paid social targeting and quality before touching the site.
Risk Analyst
The biggest risk is spending the £5,000 on advertising into an unresolved conversion problem. A second risk is changing several things at once and losing the ability to see what worked.
Initial recommendation
Do not scale ad spend. Make one change at a time and measure each.
Alternative Perspective
The cause may not be in the funnel at all. Seasonality, competitor promotions or stock availability could explain some of the fall, and the £5,000 might do more for retention than acquisition.
Initial recommendation
Run quick external checks, and consider directing part of the budget towards email and retention.
Challenge + debate
Round 1 · Sceptic challenges Strategist
Conversion is the primary problem and traffic quality is unchanged.
Challenge
You concluded conversion is the primary issue. However, you have treated traffic quality as unchanged. What evidence supports that assumption, given paid social's share of sessions rose from about 21% to 30%?
Response
The traffic-mix shift is real and I did not weight it enough. However, conversion also fell from about 2.77% to 2.47% among non-social sources, which the mix shift cannot explain. The drop is also concentrated on mobile and at checkout, which lines up with the slower product page and the higher shipping threshold. I accept that roughly 40% of the fall may come from mix, but the larger share appears to be on-site.
Position refinedRound 2 · Analyst challenges Sceptic
Traffic quality is the main issue.
Challenge
You say traffic quality is the main issue. But conversion fell among sources that were not affected by the paid social increase, and the mobile and checkout drops are specific to the site. How does audience quality explain those?
Response
It does not explain them. On the evidence, mix accounts for a meaningful part, but not the majority. I am changing my position: conversion friction appears to be the immediate issue, while traffic quality remains a real secondary factor that should still be tested.
Position changedRound 3 · Risk Analyst challenges Alternative Perspective
Seasonality, competitors or stock may explain part of the drop, and retention may deserve budget.
Challenge
Your alternative explanations are not supported by anything in the evidence supplied. What would distinguish them from a funnel problem, and why should any budget move to retention before the funnel is understood?
Response
Fair point: none of these can be confirmed from the current data. They can be checked cheaply: last year's same period, branded search trends, competitor prices and a stock-out log. I would keep the position but narrow it: do the checks first, and treat any retention spend as a small, capped test rather than a reallocation of the main budget.
Position refinedReassessment
Strategist
Changed: partlyConversion friction is still the main issue, but around 40% of the drop may come from a traffic-mix shift.
The Sceptic's challenge exposed an under-weighted factor; the source-level and mobile data kept conversion as the larger cause.
Analyst
Changed: noSame view, with a sharper priority on source-by-device segmentation as the first step.
The debate supported the original decomposition.
Sceptic
Changed: yesConversion appears to be the immediate issue, but traffic quality should still be tested.
Conversion fell even among sources not affected by the paid social shift, and the mobile and checkout drops are site-specific.
Risk Analyst
Changed: partlyDo not scale ad spend; a small capped test of retention or paid activity is acceptable once fixes are in.
The Alternative Perspective's retention point showed a cheap test carries little risk.
Alternative Perspective
Changed: noRun cheap external checks first; retention is a small capped test, not a reallocation.
The position was narrowed but not reversed; the alternative explanations are unproven rather than disproven.
MultiDI conclusion
Don't increase ad spend yet.
The strongest available evidence points to a conversion and offer problem, concentrated on mobile and at checkout, alongside recent changes to page speed and the shipping threshold. More traffic would flow into the same friction. Traffic quality remains a real but smaller and still-unresolved factor.
Arguments against
- About 40% of the fall may come from a traffic-mix shift, which site fixes will not address.
- A small ad test is cheap and could produce useful data.
- Email converts well, so a modest retention test might return value sooner than site fixes.
- Waiting has a cost while sales are down.
Remaining uncertainty
- How much of the on-site fall is caused by the shipping threshold versus page speed.
- Whether seasonality, competitors or stock issues contribute.
- Whether the paid social audience or creative has changed.
Way forward
Analyse conversion by traffic source and device
This separates the traffic-mix effect from the on-site effect and settles the main disagreement.
Speed up the mobile product page
Mobile is about 70% of sessions and load time rose from about 2.1s to 3.4s after the review widget was added.
Test the shipping offer
Checkout completion fell from 61% to 54% after the free-shipping threshold rose from £40 to £60.
Run cheap external checks
Seasonality, competitor promotions and stock-outs cannot be excluded from the data supplied.
Reassess advertising spend
Once the funnel fixes are measured, a capped test can show whether extra traffic now pays back.
MultiDI provides structured decision support, not professional advice. You remain responsible for the final decision.
Try MultiDI free