Open Upsell Strategies, select the date range, and compare Upsell Revenue, AOV Lift, Conversion, Impressions, paid upsells, Uptake, and Rev/Visitor.
Optimisation should start with a reliable customer journey. A strategy cannot be improved from analytics alone if its extension is missing, its products are unavailable, or its targeting never matches.
Before changing a strategy:
Choose a date range with enough traffic.
Record the placement, markets, targeting, modules, price, discount, and status.
Note impressions, paid upsells, uptake, revenue, and revenue per visitor.
Confirm the placement has no installation warning.
Check that the products and variants are still active and available.
Compare like with like. A checkout add-on and a One-Click post-purchase offer have different customer intent and traffic.
If impressions are lower than expected, check:
the strategy is Active;
the surface is installed on the published profile;
intended markets are selected;
targeting is not unnecessarily narrow;
locked-strategy settings are intentional;
a more specific strategy is not winning;
the placement is actually reached in the customer journey.
Use a normal All customers strategy when you need broad coverage, then let more specific matching strategies win for their audiences.
Relevance usually matters more than adding a larger discount to an unrelated product.
Use Single Product for a proven, specific add-on.
Use Group of Products for a controlled choice.
Use related, complementary, recently viewed, bestseller, collection, search, SKU, or integration modules when their source data is strong.
Reorder modules so the most useful source is tried first.
Remove repeated modules that return the same products.
Check module-empty warnings and product eligibility. Hidden products, inventory, market availability, restrictions, and purchased-product settings can remove every candidate after a module runs.
When impressions are healthy but Uptake is low, test one of:
clearer product and benefit copy;
a more relevant product or variant;
a smaller, defensible discount;
a simpler quantity choice;
a better image or layout on supported surfaces;
a different placement that better fits the decision.
Confirm the advertised discount agrees with Shopify's calculated price and compare-at data. A larger headline discount is not an improvement if the final saving is unclear or margin is poor.
Use targeting to make the offer more relevant, not simply smaller.
Product and collection filters work well for accessories.
Customer spending, order count, spending tier, and Shopify Segments can support loyalty offers.
B2B, selling-plan, cart-attribute, discount-code, and order-tag filters can separate operational journeys.
Test All filters match and Any filter matches with real customer and cart data. A highly specific strategy can show excellent Uptake while contributing little revenue because it rarely appears.
Add another placement only when it has a distinct role.
Checkout is suited to low-friction pre-payment additions.
One-Click Upsell is suited to one strong immediate offer.
Thank You provides post-payment discovery.
Order Status provides repeat exposure while customers track the order.
Install, configure, and measure each surface separately. Repeating the same message in every block can make the experience feel aggressive.
Write a simple hypothesis, such as “A single matching accessory will improve Uptake.”
Duplicate or edit the strategy without changing unrelated settings.
Use Draft while preparing the variant.
Run a split test when the strategies can share the same placement and audience.
Allow weekday and weekend traffic to contribute.
Compare Uptake and Rev/Visitor as well as revenue.
Apply the winner or restore the original.
Check installation and availability warnings.
Review the largest impression changes.
Investigate strategies with high views and low Uptake.
Protect strong high-Uptake niche strategies from careless broadening.
Review revenue and Rev/Visitor against product margin.
Choose no more than one or two meaningful tests.
Record what changed and when.
Changing product, copy, targeting, and discount at the same time.
Comparing different placements as if they had equal opportunity.
Optimising from a very short or low-traffic period.
Using discounts before checking relevance and availability.
Treating a zero impression count as a copy problem.
Assuming an Active strategy is visible without checking its extension.
Changing several parts of a strategy at once, which makes the result hard to explain.
Use Upsell Strategy Analytics for the metric definitions behind this process.
