How to Increase Ecommerce Sales: 20 Techniques [2026]

The visitors who click your offers most often are the ones who spend the least. Across our customer stores, traffic from paid ads engages with onsite offers more than twice as often as direct traffic, and places orders about half the size.

Facts like that decide which work is worth doing, which is why the tactics here are grouped rather than listed. There are only five places an ecommerce store can find more revenue:

• The offers shoppers see while they browse

• The carts they fill and abandon

• The size of the orders they do place

• How often they come back and buy again

• The store itself, meaning the pages all of the above sits on

Each of the 20 tactics below gives the reasoning, the steps to apply it, and what we measure across our customer stores between August 2025 and July 2026. Campaigns from Nutrimuscle, Pierre Hardy, Maison Lejaby and others show what each looks like in place.

The tactics assume a store with traffic and orders already coming in, where the constraint is conversion, order value and repeat purchase rather than awareness.

Find the area costing you most, then work down from there:

Jump to a section:

• Where to start

• Improve personalization across the shopping experience

• Recover abandoned carts

• Raise average order value

• Increase customer lifetime value

• Improve product pages and store design

Where to start

Working through all five areas at once spreads effort thin and makes it hard to tell what worked. The five solve different problems, so the useful starting point depends on what your own numbers already show.

Find the row that matches your situation and start with the tactics listed:

Improve personalization across the shopping experience

One offer shown to every visitor averages very different shoppers into a single message. The gaps between groups are wide enough to act on, and none of these tactics need extra traffic.

1. Match the offer to where the shopper came from

Someone who clicked a paid ad, someone who typed your domain, and someone who opened your newsletter are three different shoppers with three different budgets.

Across the stores we work with, visitors from paid ads and social engage with onsite offers most often and place the smallest orders. Visitors arriving directly engage least and order around twice as much.

Order value by traffic source, with direct visitors set at 100 so the rest can be read against them:

How to split the audiences:

• Build three audiences in the platform that runs your onsite campaigns: paid and social, direct and organic, and email. The referrer and UTM parameters are read automatically, so no tagging work is needed

• Show paid and social visitors the product or category they clicked through for, not a sitewide discount, since a percentage off a small order is mostly lost margin

• Show direct and organic visitors bundles, higher tiers, or a free shipping threshold, because they already spend more

• Exclude email traffic from signup offers entirely, and give them new arrivals or loyalty rewards instead

2. Put offers where shoppers are closest to buying

Many stores put their offers on the homepage and blog, because that is where the traffic is. Order value points somewhere else.

Order value by the page an offer appears on, with the cart page set at 100 because that is the page every store can freely message:

Where to move each offer:

• List every offer you run and note which pages it appears on. Sitewide offers count as appearing everywhere, including the blog

• Move revenue offers onto product, cart and checkout pages by adding a page rule, which takes minutes and needs no new creative

• Keep content-page offers, but judge them on email addresses collected rather than on revenue

• Change the message per page: category pages suit bestsellers and buying guides, product pages suit stock and delivery detail, cart pages suit anything about cost and finishing the order

Page targeting is one of the simpler rules to apply, and most campaign platforms handle it with a URL match rather than any code:

3. Make product recommendations available instead of interrupting with them

Recommendations work when a shopper wants to see more products, and fail when they arrive in the middle of a decision about the item already on screen. Timing decides which of those happens.

Several placements keep them available rather than intrusive. A block further down the product page, reached by scrolling. A row on the cart page itself. Or a website feed, meaning a small icon in your header that opens a list of suggestions and current offers, much like a notifications feed on social media.

In our campaign data across customer stores, where the recommendations sit matters more than which products they show:

Which products appear is worth automating rather than curating. AI recommendations read what each shopper has browsed and rank the catalog per visitor, so every customer sees a different list without anyone building segments by hand.

Pierre Hardy does exactly this, generating each list from browsed items and showing it through the website feed:

How to place the recommendations:

• Add a recommendation block below the fold on product pages and on the cart page, before you consider anything that interrupts

• For a first visit with no history, show bestsellers from the last 90 days, which is long enough to reflect real demand rather than a short spike

• On sale and new-arrival pages, show products trending over the last 7 days. 4murs runs exactly this split across different page types

• Once a shopper has browsing history, switch to viewed items and cart contents, which outperform both of the above

• Test related items against frequently bought together rather than assuming which wins

4. Help shoppers choose when the catalog is large

Some stores lose sales to indecision rather than to price. A shopper facing 40 similar products often leaves rather than picking one, and a discount does nothing about that.

The cheapest fix is to put bestsellers in front of new visitors at the moment they are already paying attention, which is when the first-order discount lands. The code and a short list of proven products in the same message removes the choosing problem and the price objection together.

Nutrimuscle uses that final step to deliver the code alongside recommended products:

For a catalog where the right product depends on the person, a quiz does more. Two or three questions narrow the range, and the answers tell you how to talk to that shopper later.

Maison Lejaby runs one as a two-step campaign, asking the shopper to choose first, then offering to send the matching selection:

Comparison tools do the same job for technical catalogs, putting the differences between similar products side by side. Nutrimuscle built one for proteins and one for collagens rather than adding another discount:

How to build the quiz:

• Pick the two or three questions your customer service team answers most often. Those are the decisions shoppers cannot make alone

• Make the first question a choice between buttons rather than a text field, so answering costs nothing

• Route each combination of answers to a set of three to five products, not to a single one, since a single result reads as a sales pitch

• Store the answers against the contact, so you can segment later campaigns on them

5. Judge every offer on revenue, not on clicks

Click-through rate is the default report in most tools, and it points the wrong way. The visitors who click most often are the ones who spend least, so a campaign tuned for clicks drifts toward your cheapest traffic.

Sticky bars are the clearest case in our data: almost nobody clicks them, and the people who do place above-average orders. Judged on clicks alone you would switch them off.

What to report instead:

• Report revenue per thousand times an offer was shown, which lets a low-volume, high-value offer show its worth

• Hold back a share of eligible shoppers who see nothing, so the figure reflects what the offer added rather than what happened around it

• Compare order value alongside conversion rate. An offer that raises orders and lowers their size can leave you flat

• Give any test at least two weeks, or one full purchase cycle if yours is longer. At mid-market volumes most offer tests resolve inside that window, so there is little reason to call one early

Set the success metric to revenue before the test starts, not after the results arrive:

Recover abandoned carts

A shopper with something in their cart has already done the hard part. They picked the product, accepted the price, and stopped somewhere between there and paying, which makes this the cheapest revenue in the guide.

6. Win back shoppers who abandon a cart before checkout

Separate your two groups before writing anything. An abandoned cart means items were added and checkout was never started. An abandoned checkout means the shopper entered the flow, usually gave an address, then left before paying.

This tactic covers the first group. You have no contact details for most of them, so an onsite message is the only channel that reaches them, and it is the highest-earning kind of onsite offer we measure, at roughly twelve times the revenue per showing of a general-purpose popup.

The reason is selection. The offer only appears for people who chose something, so it never spends attention on visitors who were never going to buy.

What goes in the message is the part worth automating. Rather than fixing one offer for everyone, an AI campaign picks per shopper from the options you allow, weighing cart contents and what has worked on similar sessions, which converts more than three times better than a single fixed exit offer.

How to configure the campaign:

• Trigger on cart value above zero combined with an exit or idle signal, rather than on the page alone

• Show the items in the cart inside the message. A reminder that names the product reads as helpful, while a bare discount reads as an offer looking for a taker

• Lead with the objection rather than the price: delivery date, return window, payment options, stock

• Hold the discount back, since leading with money rewards the behavior you are trying to remove

• Cap frequency at once per session so a returning shopper is not chased

• Give the AI a set of offers to choose between rather than one, since the choice is where the gain comes from

The message names the cart and carries the offer chosen for that shopper:

7. Treat abandoned checkouts as a separate, higher-priority group

A shopper who started checkout was seconds from paying and, in most cases, has already handed over an email address. Both facts make this group worth a different message from the one you send cart abandoners.

Reporting the two together hides the difference. Split them and the checkout group almost always shows the higher recovery rate, which is where the follow-up effort belongs.

How to work abandoned checkouts:

• Separate abandoned carts from abandoned checkouts in your reporting before changing any campaign, since the two recovery rates are not comparable

• First email inside the hour while intent is live. Cart contents, product image, a direct link back, and no discount

• Second the next day, carrying the reassurance that stops checkouts in your category: returns, sizing, delivery dates, or duties on cross-border orders

• Third after two more days, with an incentive only if the first two produced nothing

• Stop the sequence on purchase, and exclude anyone who bought a different item in the meantime

Leading every sequence with a discount builds the opposite of a loyalty program. Shoppers who learn that abandoning produces a code will abandon deliberately, and your recovery rate will look strong while your margin falls.

8. Stop offering a discount to everyone who tries to leave

An offer that appears when the cursor heads for the browser bar, usually called exit intent, is widely used. In our data it is the weakest of the common approaches on both engagement and order size.

The mechanism explains it. A discount shown to anyone leaving selects for the most price-sensitive shoppers on the site, so the orders it saves are the smallest ones you have.

Narrowing the audience is also what makes the offer tolerable. Fewer shoppers see it, the ones who do are holding a full cart, and nobody reading a blog post gets interrupted with a discount.

What to change in the campaign:

• Keep the exit trigger, and add a condition: cart is not empty, or cart value is above a threshold you set

• Split the message by cart value. Above roughly the value of two average items, free shipping tends to work better than a percentage off

• For visitors with an empty cart, swap the discount for something that costs nothing: a size guide, a buying guide, or a restock alert signup

• Measure the recovered orders separately, so a fall in average order value shows up rather than hiding inside a conversion gain

Scoped this way, the offer reads as a reminder about a specific cart rather than a blanket discount:

9. Answer last-minute doubts on the cart page

Cart pages carry the largest orders of any page you can freely message. The shopper is closest to paying and most likely to be weighing a single final concern, which makes the message worth getting right.

What belongs on the cart page:

• Delivery dates and cutoff times, phrased as a date rather than a number of days

• Return window and who pays for return shipping

• Accepted payment methods, including instalment options if you offer them

• The remaining spend to free shipping, updated as the cart changes

• Where the discount code field is, if shoppers regularly ask

Nothing here should send a shopper back to browsing, so keep product recommendations off the cart. The one exception is the post-purchase page, where an add-on cannot cost you the order.

Raise average order value

Average order value, or AOV, is the quickest number to move, because every tactic here works on someone who has already decided to buy. You are not persuading them, only giving them a reason to add one more thing.

The four tactics below are the ones with the clearest mechanism. For the wider set, including pricing and merchandising angles, see our guide to increasing average order value.

10. Run a free shipping progress bar, and set the threshold on margin

A threshold a shopper has to work out for themselves is a threshold most shoppers ignore. A progress bar naming the exact amount still to spend turns it into a decision they can act on now.

Where to put the threshold is where most published advice goes vague. Recommendations range from 10% above your average order value to 40% above it, which is a wide enough spread to be useless as a rule. Work it out from your own margin instead.

Nutrimuscle updates the message as the cart changes:

How to calculate the threshold:

• Take three numbers: your average order value excluding shipping, your average shipping cost per order, and your gross margin

• Pick a candidate threshold and work out the extra margin the larger basket earns you, which is the gap above your AOV multiplied by your gross margin

• Subtract that from your average shipping cost. What remains is what the free shipping offer actually costs you per order at that threshold

• Raise the threshold until that cost is one you would happily pay for the extra volume, then check it against your order distribution so the target sits inside reach for a normal basket

Worked through with an AOV of $40, average shipping of $8 and a 40% gross margin: a $45 threshold earns $2 of extra margin and leaves $6 of shipping cost per order, while a $60 threshold earns $8 and covers the shipping entirely.

How to run the progress bar:

• Show the gap as a cash amount whenever an item is added, not as a percentage of the way there

• Fire again when the cart crosses the line, confirming free shipping and linking to checkout, since a shopper who qualified without noticing gets no benefit

• Recalculate each quarter, because the threshold should move as your AOV and carrier rates do

• Set thresholds per market rather than converting one number, since shipping cost and basket size rarely move together across borders

11. Offer the add-on at the two moments that work

Two points in the journey take an add-on offer without resistance. The seconds after an add to cart, when the decision has just been made, and the seconds after payment, when the shopper has nothing left to lose by saying yes.

A post-purchase upsell is the safer of the two, since it cannot cost you the original order. Anything offered before payment carries a risk of reopening a decision that was already closed.

How to pick the add-on:

• Suggest accessories, refills and consumables, never an alternative to what was just chosen, because a competing product reopens a closed decision

• Show one item rather than a grid. Several options at that moment become a new browsing task, which is the opposite of the intent

• Price the add-on well below the main item. An add-on at similar cost reads as a second purchase and gets declined

• Pull the pairing from your own order data rather than from a category rule, which is where a large catalog beats a small one

• For the post-purchase slot, offer one click to add to the existing order rather than a new checkout, since a second payment step removes most of the gain

One add-on, priced below the item already in the cart:

12. Bundle products that already sell together

A bundle raises order value by removing a decision rather than by cutting a price, which is why it protects margin better than a discount does.

How to build the bundle:

• Export your order history and find the pairs and triples that already appear in the same basket. Build from those rather than from stock you want to move

• Price the bundle below the sum of its parts by a visible amount, and show the saving as a cash figure, since a percentage of a multi-item total is hard to judge

• Give the bundle its own product page so it can rank and be linked to, rather than existing only as a cart offer

• For catalogs that need explaining, build a starter kit, which doubles as a recommendation for someone who does not know where to begin

Reading the pairs by hand stops scaling past a few hundred products. AI bundling apps generate frequently-bought-together sets from your own order history and keep them current as buying patterns move, which is the part manual merchandising cannot keep up with.

A generated bundle still needs promoting, since a set nobody sees changes nothing:

13. Reward bigger baskets with tiers, not a flat discount

A flat 10% off applies equally to a small order and a large one, so it costs margin without changing anyone's behavior. A tier gives the shopper a reason to reach the next level, and shows them how close they are.

L'Atelier d'Amaya ran tiered pricing during Black Friday, shown both as a popup and as a message in the feed:

How to structure the tiers:

• Set the first tier just above your average order value and the second at roughly 1.5 times it, then check the tiers against your actual order distribution

• Vary the reward rather than only the percentage: free delivery, then a gift, then the larger discount

• Show the current tier and the next one together, with the amount between them

• Give every code an expiry, because a code without one gets saved and used weeks later on a purchase that was coming anyway

Increase customer lifetime value

A second order costs far less to win than a first one, since you are not paying to be found again. That margin difference is what makes this the most profitable of the five areas, and every tactic here exists to give a shopper a way back to you.

14. Treat a second visit as worth more than another ad click

An audience you can reach directly keeps working after the spend stops. Paid traffic has to be bought again every month.

Visitors who arrive directly, meaning they typed your domain or used a bookmark, place the largest orders of any group in our data, around twice the size of orders from paid ads.

Those are mostly people who already know you, which is the argument for spending on the return visit rather than on another first one.

Where to start measuring:

• Calculate the share of customers who have placed two or more orders, then track it monthly rather than as a single figure

• Split your revenue between first orders and repeat orders, and watch the ratio rather than the total

• Compare the order value of direct and returning visitors against paid visitors, which usually makes the case for the work below on its own

Then give returning shoppers a reason to finish rather than browse again. The offer lands best on a product or category page, once a returning visitor has looked at two or three items, because by then you know what they came back for.

L'Atelier d'Amaya shows this while returning visitors are browsing products, not on arrival:

Three conditions separate this from a generic welcome offer: the visitor is returning rather than new, they are on a product or category page rather than the homepage, and the offer differs from the one they already declined on their first visit.

15. Collect a contact detail where shoppers will actually give one

Where a signup form sits changes how many people finish it. In our data, forms inside a website feed are completed around eight times per hundred shown, popup forms around five, and forms embedded quietly in a page fewer than one.

Embedded forms are not failing, they are passive: a form sitting in a page collects from people who went looking for it. Use them for steady background capture, and a popup or feed when you need volume.

How to raise the completion rate:

• Ask for the phone number in the same flow as the email. Across popup campaigns we see, phone numbers come in at roughly a quarter of email volume, which is a second channel for almost no extra work

• Put an embedded form in your footer and on your about page as a permanent baseline, then run a popup or feed offer on top for volume

• State what the subscriber gets and how often, since vague signup copy collects addresses that never open anything

• Suppress the form for anyone already subscribed, which is usually a single rule

Maison Lejaby asks for the phone number as its own step, after the email address is already in:

16. Offer restock and price-drop alerts

Both of these are requests the shopper makes rather than messages you decide to send, which is why they get accepted and why they convert well.

Email works for this, and browser notifications, sometimes called web push, reach shoppers who will not join a list at all.

How to run the alerts:

• Put a notify-me option on every out-of-stock product page, above the fold rather than in the description

• Ask for permission after the shopper does something that shows interest, never on the first page they land on, since a request on arrival is easy to dismiss and hard to repeat

• Send the restock alert immediately and only once, with a direct link to the product rather than to the category

• Keep price-drop alerts to items the shopper actually viewed, since a general sale announcement is a different message

émoi émoi uses browser notifications for restock and launch news, alongside email rather than instead of it. The channel earns its place by reaching people who never subscribed, not by carrying the bulk of revenue.

A restock notification from that store, sent to shoppers who asked for one:

17. Ask for one detail beyond the email address

An email address supports one kind of campaign. A birthday, a size, or a favorite category supports several, and each can be collected with a single question.

L'Atelier d'Amaya invited registered customers to add their birthday through a short campaign:

How to collect the extra field:

• Take the email on the first visit and the extra detail on a later one, using a visit-count rule rather than one long form

• Ask logged-in customers directly, since they have already committed and the question reads as account maintenance

• Pick one field and use it within a month, since a field with no campaign attached is data collection rather than marketing

• Give returning shoppers a different offer from the welcome discount they already declined: early access, loyalty rewards, or restock news

Improve product pages and store design

Everything above sits on top of your store. If a product page leaves an obvious question unanswered, no offer will fix it, and the shopper leaves before a cart exists.

These three are craft judgments rather than measured findings, which is worth stating plainly.

18. Answer the questions your category actually raises

Shoppers usually leave a product page because something they needed to know was missing, not because the page was not persuasive enough. What is missing differs by category.

Questions worth answering, by category:

• Clothing: size and fit guidance, the model's height and the size they wear, how the fabric behaves, and the return terms

• Beauty and supplements: full ingredients, how to use it, when to expect results, and how it differs from your similar products

• Furniture and hardware: dimensions, materials, assembly, delivery window, and whether it fits through a doorway

• Any category: read the last 50 customer service messages and count how many were answerable from the product page

Most of these answers are visual, so several angles and one short video usually do more than longer copy. A small block inside the page can carry the answer at the moment of hesitation. L'Atelier d'Amaya put one directly beneath the Add to Cart button:

19. Put reviews and customer photos next to the price

Reviews answer the one question your description cannot, which is whether the product matches what you said about it. Position matters as much as volume, so star ratings belong beside the title and price rather than far below the fold.

Collecting them is the harder half of the job.

How to build up review volume:

• Send the review request after delivery rather than after the order, timed to the carrier's confirmation

• Ask for a photo explicitly, since a general review request almost never produces one

• Offer loyalty points or a small credit rather than a discount code, which keeps the incentive off your margin

• Show the review count next to the rating. A 4.8 from six reviews reads as weaker than a 4.4 from four hundred

• Put the reviews you have in front of more shoppers rather than leaving them on the product page. A rating and one short quote inside a leaving-visitor campaign reaches people at the moment they are deciding against you

Social proof carried by an exit campaign, rather than a discount:

20. Keep navigation and speed out of the way

Store design fails quietly. A shopper who cannot find the category they are looking for does not report the problem, they leave and buy elsewhere.

Four checks worth running:

• Name categories the way shoppers describe them rather than the way your merchandising team files them, and keep any product within three clicks of the homepage

• Test the whole purchase on a mid-range phone on mobile data, not on a desktop browser window made narrow

• Audit your tag manager twice a year. On a store of any size the container accumulates scripts nobody owns any more, and each one costs speed

• Check every market, not just your largest. Localized sites drift, and a broken size guide or a missing payment method in a secondary market is easy to miss for a year

Check for collisions while you are testing. An offer overlapping a chat widget or a cookie banner on mobile is a common reason a campaign shows plenty of impressions and almost no clicks.

Frequently asked questions