Most small business owners I meet have the same two fears about upselling. The first is that they'll come across as the shop assistant who won't let you leave without a warranty. The second is that if they don't do it, they're leaving money on the counter every single day. Both fears are reasonable, and the good news is that automation, done properly, solves the first one rather than making it worse.

I've set up automated upsells and cross-sells for online shops, trades businesses, a couple of clinics and more than one accountancy practice. The ones that worked all had something in common. The offer arrived at a moment when the customer would plausibly have asked for it anyway, it was clearly optional, and it stopped when the customer said no. The ones that failed were the ones where somebody copied a tactic from an American growth blog without asking whether a customer in Stockport would find it helpful or irritating.

The Short Answer

Automating upsells and cross-sells without being pushy comes down to three things. Trigger offers from what the customer actually did, not from a calendar. Make every extra a clear, unticked choice at a fair price. And cap how often anyone hears from you, so a customer who ignores two suggestions isn't sent a third.

If you sell products online, the tools to do this cost somewhere between nothing and about £30 a month at a small scale. If you sell services, the same logic works through your CRM or booking system, and in some cases a well written follow up email template does the job better than any software. And there's one more piece that most guides written outside the UK skip entirely: since April 2025 the Competition and Markets Authority can fine you directly for the pushy version, and it has already started doing so.

What Pushy Actually Looks Like To A Customer

"Pushy" is a feeling rather than a rule, so it helps to be specific. From the customer's side, the offers that grate share a few features. They interrupt something the customer was trying to finish. They assume a yes, either through a pre-ticked box or a button that's much bigger than the "no thanks" link. They keep coming after the customer has declined. And they use pressure, whether that's a countdown clock or a "three people are looking at this" banner.

Automation multiplies whichever behaviour you build into it, so a slightly annoying pop up becomes a business wide habit within a week. That cuts both ways. Build a considerate offer once and it's considerate ten thousand times over.

An upsell is offering a better version of the thing they're buying, such as the larger size, the longer service plan, or the premium tier. A cross-sell is offering something that goes with it, such as the filter for the machine or the annual review alongside the tax return. Both are fine. Both become pushy the moment they stop being about what the customer needs and start being about your average order value.

The Law You Need To Know Before You Automate Anything

Two sets of UK rules govern this, and I'd rather you heard about them from me than from a regulator's letter.

The first covers how you contact people. The Privacy and Electronic Communications Regulations, usually shortened to PECR, say you can't send marketing emails or texts to individuals without consent. There's one exception that matters enormously for upselling, and the Information Commissioner's Office calls it the soft opt-in. As the ICO's guide to electronic mail marketing explains, it lets you email your own previous customers about similar goods or services, provided you collected their details during a sale or negotiation for a sale, gave them a chance to opt out at the time, and give them a chance to opt out in every message.

Read those conditions carefully, because the soft opt-in is the legal foundation of nearly every automated cross-sell email. It does not cover bought lists, it does not cover people who only signed up for a newsletter, and it does not cover offers that are unrelated to what they bought. A customer who bought a coffee machine can be emailed about beans and descaler. Emailing them about your new range of garden furniture is a different conversation, and one that needs proper consent. The ICO also notes in its guidance on complying with the PECR email rules that "refer a friend" schemes can't rely on the soft opt-in at all.

The second set of rules covers what the offer itself looks like. The Digital Markets, Competition and Consumers Act 2024 replaced the old Consumer Protection from Unfair Trading Regulations from 6 April 2025, and the CMA's published guidance on unfair commercial practices sets out what's now banned outright. The CMA can now investigate and fine businesses directly without going to court, with penalties of up to £300,000 or 10 percent of global turnover, whichever is greater.

Here's why that matters to a small business rather than just to airlines and ticket sites. In June 2026 the CMA fined Marks Electrical, an appliance retailer in Leicester, £720,000 and ordered refunds of around £600,000 to nearly 40,000 customers. According to the CMA's own announcement, the problem was that customers were automatically opted into paid extras, specifically an old appliance recycling service and a packaging removal service, through pre-ticked boxes at checkout. The refund per customer was about £15. That is exactly the kind of cross-sell a small online shop might add without a second thought, and it is now the subject of the CMA's second ever direct enforcement decision.

The same enforcement wave covers urgency. The CMA's guidance says that if you use a countdown timer you must be able to back up the urgency, and that a promotion which quietly restarts after the timer hits zero is likely to be misleading. As law firm Ashurst Perkins Coie summarises in its note on the DMCC Act's treatment of manipulative online design, pre-ticked boxes, default selections and complicated opt-outs are all firmly in the regulator's sights. Emma Sleep, the mattress firm, gave binding undertakings in May 2026 to stop using misleading countdown timers and false high demand messages after a High Court endorsed settlement with the CMA.

So the rule of thumb is short. Every extra must be an active choice, never a default. Every urgency claim must be true. And every automated email must go to someone you're allowed to email about that kind of product. Get those three right and the rest of this article is about being useful rather than about staying out of trouble.

Four Moments Where An Automated Offer Feels Like Help

Timing does more work than copy. The same offer that feels grasping on a landing page feels like good service when it arrives at the right moment. These are the four moments I build around, in the order I'd add them.

The first is the thank you page, immediately after payment. The customer has finished the hard part, their card details are already entered, and they're in a relaxed state. A single, relevant, one-click offer here is the least intrusive upsell that exists, because saying no costs nothing and the customer was leaving anyway. On Shopify this runs on the order status page after checkout, so it never touches the checkout flow itself. As the team behind the Oxify app points out in their comparison of post-purchase upsell apps, you don't need Shopify Plus for this; only in-checkout offers need the more expensive plan.

The second is a few days after delivery or after the service is done. This is when a cross-sell email lands well, because the customer has the product in their hands and knows what they're missing. A cycle shop I worked with sends one email five days after a bike is delivered suggesting a pump, lights and a lock, with a plain sentence explaining why each matters. It converts well because it reads like advice from a mechanic, which is essentially what it is.

The third is replenishment. If you sell anything consumable, from dog food to printer toner to contact lenses, the reorder reminder is the most welcome automated message you'll ever send. The trick is timing it to when the product actually runs out, not to when your sales target needs a nudge. Klaviyo's predictive tools estimate this from purchase history; a simpler tool can use a fixed number of days per product that you set yourself.

The fourth is the service milestone. For trades and professional services, the equivalent of replenishment is the annual boiler service, the twelve month review, or the renewal. An automated reminder that says "it's been eleven months since we last serviced your boiler, here's a link to book" isn't an upsell in the traditional sense, but it's the one that brings in repeat revenue most reliably for the trades businesses I've worked with.

Three Rules I Set Before Building Any Offer

Relevance comes first, and I mean this strictly. The offer must relate to what they just bought, both because the soft opt-in requires it and because unrelated suggestions are what customers remember as spam. If your system can't tell what someone bought, you're not ready to automate cross-sells yet. Fix the data first.

Price ceiling is second. Keep the add on well below the value of the original purchase. The Appbrew team's round up of Shopify upsell apps cites Scandiweb data suggesting that upsells priced above roughly 30 percent of the original item see abandonment rise sharply. I've seen the same pattern in my own accounts. A £12 case on a £90 pair of headphones gets taken. A £60 extended warranty on the same headphones gets resented.

Frequency cap is third, and it's the one most tools make you configure yourself. Decide in advance how many suggestions a customer can see or receive before you stop, and what "stop" means. My default is one offer on the thank you page, at most two cross-sell emails in the thirty days after purchase, and then nothing promotional until they buy again or a replenishment date arrives. If someone clicks through an offer and doesn't buy, that counts as a no. Don't retarget it.

Tools For Online Shops And What They Cost

I'm going to name prices, because most articles on this subject don't, and because the cost is the thing small business owners ask me about first. Several of these platforms bill in US dollars, so the pound figures are approximate and will move with the exchange rate.

For email flows, Klaviyo is the tool most e-commerce agencies push, and it's good at this. As the SME Compare pricing breakdown notes, there's a free plan covering 250 profiles and 500 emails a month, and the paid Email plan starts at around £15 a month for up to 500 active profiles. Omnisend's guide to Klaviyo's pricing confirms the entry tier at 20 dollars a month and points out that since February 2025 Klaviyo bills on every profile that could be emailed, not just the ones you do email. That makes list cleaning a cost control exercise rather than an optional chore.

The honest caveat about Klaviyo is that it's built for shops with a decent order volume. If you're doing thirty orders a month, its predictive features have nothing to predict from, and you're paying for machinery you won't use. At that scale I'd look at Mailchimp, which starts at just under £10 a month on Essentials but bills on total contacts including unsubscribes, or Brevo, which starts at around £6 a month and bills on emails sent rather than list size. Omnisend, from around £12 a month, is the closest like for like alternative to Klaviyo for a Shopify or WooCommerce shop.

For the thank you page offer on Shopify, the app market is crowded and the prices are low. Selleasy, made by Logbase, has a free plan and a flat paid tier of 29 dollars a month, roughly £22, covering product page, cart, checkout and post-purchase placements in one app. ReConvert starts at 4.99 dollars a month, about £4, and is the strongest option if you want deep control over how the thank you page looks. AfterSell is the premium choice at 34.99 dollars a month, about £26, for up to 500 orders, rising to 99.99 dollars at 2,000 orders. Logbase's own comparison of Selleasy and AfterSell is written by one side, but its mid 2026 pricing table is accurate.

One thing I'd warn you off. Several of the AfterSell tiers and some competitors offer to show third party adverts on your thank you page in exchange for a small payment per order. In the UK this arrives via the Rokt Thanks programme. It's revenue, but it's revenue you earn by letting someone else upsell your customer with an unrelated product the second they've paid you. That is the definition of pushy, and it isn't yours to control. Turn it off.

Tools If You Sell A Service Rather Than A Product

Product shops get the glamorous apps. Service businesses get better margins from a much simpler set up, and I'd argue they have the easier job.

The core tool is whatever holds your customer records. For a trades business that's often a job management system such as Tradify or Jobber, both of which can send automated follow ups after a job is marked complete. For a clinic, a salon or a personal trainer, the booking system usually has the same function. For a professional practice, it's the CRM, and for very small firms it's sometimes just a spreadsheet and a calendar reminder to a human. All of these work if the message is right.

The pattern is the same as for products. When the job closes, a message goes out after a sensible delay. It thanks the customer, checks everything is as it should be, and mentions one related thing you can do. A plumber's version might be "we noticed your stopcock is stiff, we can replace it for £85 next time we're passing". An accountant's version might be "now the year end is filed, most clients at your size find a quarterly check in saves them the January panic; here's what that costs". Neither is a hard sell. Both are things the customer would probably have asked about if they'd known to.

Deployteq's guide to structuring cross-sell and upsell emails makes the point that the tone has to be consultative rather than promotional, and I'd go further. For a service business the offer email should sound like the person who did the work wrote it, because ideally they contributed the line about the stopcock. Automation handles the sending, the timing and the record keeping. The human supplies the observation that makes it feel personal.

A Workflow I'd Actually Build For A Shop With Under Ten Staff

Here's the sequence I set up most often, with the settings, so you can copy it.

Day zero, at checkout. No pre-ticked extras of any kind. If you offer gift wrap, delivery protection or a warranty, each one is an unticked box with the price shown next to it. After payment, the thank you page shows one offer that's clearly related to the main item, priced under 30 percent of the order, with a "no thanks, continue" link that's as easy to find as the accept button. One offer. If they decline, they see their order confirmation and nothing else.

Day one. The order confirmation and shipping emails carry no offers at all. These are transactional emails, and stuffing them with promotions both annoys people and muddies the line between service messages and marketing under PECR.

Day five to seven, depending on your delivery times. One cross-sell email, triggered by the specific product category they bought, sent only to customers whose details you collected at checkout and who didn't opt out. Two or three related products, one sentence each on why they're useful, no discount unless you'd offer it anyway. Include the unsubscribe link, which your platform adds automatically.

Day fourteen. A review request. This isn't an upsell, but it matters here because the DMCC Act now bans fake reviews and requires you to have a proper process for genuine ones. Don't incentivise the review and don't filter which customers get asked.

Day thirty. If they haven't bought again and the product is consumable, a replenishment reminder timed to the product's actual life. If it isn't consumable, nothing. The flow ends. They go back into your normal newsletter list, if they consented to it, and that's all.

In Klaviyo, all of this is one flow with a conditional split at the top on number of orders. First time buyers get the full sequence. Repeat buyers skip the introductory content and get a shorter version, because as one agency put it in their post-purchase playbook, treating a loyal customer like a stranger trains them to ignore you. In Mailchimp or Brevo it's two or three separate automations, which is fine.

What The Numbers Look Like, Honestly

I'll give you benchmarks, but with a warning first. Nearly all the widely quoted figures come from Klaviyo's own customer base, and Klaviyo has an interest in automated flows looking impressive. Digital Applied's playbook on lifecycle flows puts this well: read every number as "Klaviyo reports", not as independently audited fact. There is no UK specific benchmark I'd trust for small shops, and I'd be suspicious of anyone who claims one.

With that said, the shape of the data is consistent. Klaviyo's 2026 benchmark report, summarised by the Ecom Heads newsletter, found that across more than 180,000 brands, automated flows made up around 5 percent of sends but roughly 41 percent of email revenue, with revenue per recipient of about 1.94 dollars on flows against 0.11 dollars on one off campaigns. Agency portfolios I've read put post-purchase flow revenue at somewhere between 2 and 5 dollars per recipient when the cross-sell is well targeted. Thank you page apps commonly claim average order value lifts in the 10 to 30 percent range, though the Oxify team more cautiously say single digit to low double digit percentages is typical for a well targeted one-click offer, and that's closer to what I see.

Translate that to a small UK shop and it looks like this. A shop doing 200 orders a month at a £45 average, with a well set up thank you page offer and one cross-sell email, might reasonably add £300 to £800 a month. That's real money for a business that size. It is not the doubling of revenue the app listings imply, and if your first month shows a 40 percent lift, check you haven't counted the original orders in the upsell figure.

For service businesses the numbers are lumpier and there's no benchmark worth quoting. What I can say is that the businesses that add a single follow up with one related offer usually see it pay for the software within the first quarter, because the offers are higher value and the customer relationship already exists.

When Not To Automate This At All

Some of the most useful advice I give is to leave this alone, at least for now.

If you have fewer than about fifty customers a month, the maths rarely works for a dedicated upsell app. A conversion rate of a few percent on fifty offers is one or two extra sales, and you'd get more from writing a personal email to each customer. The free tiers cost nothing to try, but your setup time isn't free.

If your products don't naturally go together, don't force it. A shop selling one hero product with no accessories has nothing to cross-sell, and inventing a "complete the look" bundle from unrelated stock is precisely the kind of offer customers remember as pushy. Sell the replenishment or the upgrade instead, or accept that this lever isn't yours.

If your customer data is a mess, fix that first. Automated cross-sells depend on knowing what someone bought, when, and whether you're allowed to email them. If your order records and your email list live in different places and don't talk to each other, the automation will send the wrong offer to the wrong person, and under PECR "we didn't realise they'd opted out" is not a defence.

And if you're tempted by any tactic that involves pressure, take the DMCC Act as the deciding vote. The CMA opened its first eight investigations in November 2025 specifically into drip pricing, urgency messaging and automatic opt-ins, and sent advisory letters to a hundred more businesses. Two of the first eight have already settled with substantial fines. The regulator has said it will focus early on the most egregious cases, which means a small business isn't first in the queue, but it also means the tactics themselves are now unambiguously unlawful. The gentle version isn't just nicer. It's the only version that's legal.

Something To Do This Week

Open your own checkout and buy something from yourself. Look for any box that's ticked by default, any charge that appears after the headline price, and any timer or stock warning you couldn't prove if asked. Untick, remove, and delete accordingly. That's the compliance half done in an afternoon.

Then pick the one moment from the four above that matches your business best, and build a single offer for it. One product, one message, one trigger. Run it for a month, count the extra revenue honestly, and only then decide whether it's worth a second. Automating upsells and cross-sells without being pushy isn't a system you buy. It's a habit of asking, before every offer, whether the customer would have thanked you for it. The software just makes sure you ask every time.