How referral schemes work: a study of 27 live UK schemes

Referral Plug publishes a referral link for each of the 27 schemes in this study, and earns from them when readers join. No brand paid to be included, and this study ranks none of them.

Written for businesses deciding whether to run a referral scheme, or how to fix one, this reads 27 live UK schemes against their own terms, then sets them against the published research on referral rewards and the UK rules that apply. If you want to use a referral code rather than run a scheme, start with what a referral code is.

How we built this study

This study reads the 27 refer-a-friend schemes Referral Plug had live on 29 September 2026, across ten categories from current accounts and energy to broadband, cashback sites and car hire. Every week, each scheme's advertised reward is checked against the brand's own terms page, and I click every link to confirm it lands in the right sign-up flow with the referral attached before signing that brand off. The figures here are the ones signed off on 29 September.

• In scope: the UK consumer and small-business schemes live on the site that day. The two paused schemes, Starling and Affordable Mobiles, are left out.
• Source: each brand's own terms or scheme page, recorded row by row in the CSV. Quidco and THE OUT publish no standing page that states the reward, so their rows rest on the referral landing page, and Quidco's referrer side on what its account showed and actually paid me.
• What we recorded: how the friend is identified, what each side gets and in what form, what has to happen before either is paid, any minimum time as a customer before payout, whether the referrer must be a customer, caps, and whether the amount is random. The minimum-time count excludes restrictions on keeping or withdrawing an already credited reward. Money paid into a bank account counts as cash; credit restricted to a brand's bills or services does not.
• Judgment calls: "the same reward both sides" compares standard terms. EDF's temporary £75 and £100 (for switches started by 6 October 2026) counts as its standard £50 each, giffgaff (£5 of credit to the friend, £5 of cashable Payback to the referrer) and THE OUT (£50 off for the friend, £50 of credit for the referrer) count as the same amount in different forms, and so do Surfshark, whose referrer may take cash instead of free months, Zzoomm, whose referrer's £75 rises to £100 after five referrals, and Tide, whose referrer's £100 rises to between £125 and £200 on a paid Tide plan. A cap is a ceiling, so Monzo, which pays £5 each side after a referrer's first 100 referrals, is not counted as one.

Limitations, stated plainly: 27 is a small sample, and not a random one. Referral Plug lists only schemes that pay the person joining, so this describes how dual-sided UK schemes are built, not how common any feature is across the market. Amounts move, too: EDF, VOXI and TopCashback were all paying time-limited figures on the day we read them.

What is a refer-a-friend scheme?

A refer-a-friend scheme pays an existing customer (the referrer) for bringing a new customer to a brand, and usually pays the new customer (the friend, or in some terms the referee) as well. Every scheme in this study has the same three working parts: an identifier unique to each referrer, a condition the friend must meet, and a condition the referrer must meet.

A unique link or code for each referrer

The identifier is how the brand tells which customer sent which sign-up, a step called attribution. If attribution fails, nobody is paid, however genuine the recommendation. 24 of the 27 schemes carry the identifier in a link the friend clicks; 3 make the friend type a code, which the code-or-link section (opens in new tab) comes back to.

What the friend has to do

The friend's condition proves a real customer before money moves. In 25 of the 27 schemes it costs the friend something or needs a service to go live: a first direct debit at Octopus Energy, £500 of card spend within 90 days at Tide, a car actually sold through Motorway, broadband connected for 21 days at Quickline. Only Zilch (identity checked and a card issued) and Incard (a business account approved) reward the friend for signing up alone.

9 of the 27 also hold the reward until the friend has stayed a set time: toob requires 14 days connected, and Surfshark 31 days subscribed.

What the referrer has to do

23 of the 27 pay only a referrer who is a customer. All four exceptions are broadband networks: toob, Quickline and Airband let anyone refer, and Hyperoptic accepts former customers and people who have registered interest at their address. EDF and Rakuten also ask something of the referrer: EDF pays once both people have made a payment, and Rakuten once both have linked a bank account.

The terms do not always say what happens when a referrer leaves. E.ON Next paid me two referrer rewards in July 2026, after my own E.ON Next account had closed.

The reward

20 of the 27 pay the friend in something other than cash: bill or account credit (5), vouchers or gift cards (6), points (2), investments (2), Bitcoin (2), free subscription months (Surfshark), a booking discount (THE OUT) and a perk (Incard's doubled cashback). The seven that pay the friend cash are Monzo, Tide, Airband, Zzoomm and the three cashback sites, TopCashback, Quidco and Rakuten.

Payout runs from instant to months. THE OUT takes £50 off the friend's first booking at checkout; Zzoomm confirms its reward from 60 working days after the first direct debit; and the Tide reward I joined for in 2022 arrived about four months after I opened the account.

What is a dual-sided referral scheme?

A dual-sided referral scheme, also called double-sided or two-sided, rewards both people: the referrer and the friend who joins. A one-sided scheme rewards only one of them, and in practice that is the referrer.

All 27 schemes in this study are dual-sided, but that is our selection rule rather than a market share: Referral Plug lists only schemes that pay the reader, who is the friend. In a wider sample, dual-sided is the most common design but not the only one: of 351 live referral programmes scanned for a 2020 study, 55% rewarded both sides, 40.5% only the referrer and 2.6% only the friend (opens in new tab). Well-known UK brands run one-sided schemes. Revolut's UK terms (opens in new tab) call its promotion "a one-sided campaign, where only the Revolut customer who was invited to refer a friend (not the invitee) can be awarded the Reward". Admiral's page (opens in new tab) offers the referrer "a reward of up to £75" and names nothing for the friend, and PensionBee's (opens in new tab) offers "£100 for each friend that opens an account and adds £100 or more", again with nothing stated for the friend. All three read as quoted on 1 October 2026.

Dual-sided does not mean equal. 22 of the 27 pay both sides the same on their standard terms. Of the five that do not, three tilt towards the referrer: TopCashback (£10 to the friend and £25 to a Plus member who referred them, under the campaign running to 31 October 2026), Quidco (£10 and £25) and Incard (a cashback boost to the friend, £250 to the referrer). Yonder tilts towards the friend, and the BA Amex varies the referrer's Avios by card. Trading 212 shows a third shape: once a referrer has been paid five times in a campaign, the friend is still paid, so the scheme carries on as one-sided in the friend's favour.

Do dual-sided referral schemes work better?

Four field experiments, reported in two studies, found that giving the friend part or all of a referral reward recruits more new customers than giving all of it to the referrer. The gain is on the friend's side: referrers invite about as readily either way, and more of their friends sign up when some of the reward is theirs. A fifth, at a US cashback site, found that tilting a two-sided reward towards the referrer drew more invitations. Two caveats cover all of them: no experiment we found tests a full reward to each side against the same reward to one side, and none is British or tests a bank or an energy supplier.

Three practical points follow. The friend's reward protects the referrer too: in Verlegh and colleagues' studies, a friend who knows the recommendation was paid for suspects it, and rewarding both sides removes that. The cashback exception matches UK practice: two of the three UK cashback sites in our study, TopCashback and Quidco, also tilt their rewards towards the referrer. And keep the friend's reward certain, since in Li, Belo and Li's experiment the lottery cost referrals only on the friend's side; Monzo, InvestEngine and Trading 212 randomise the friend's reward above a guaranteed £10, £20 and £8, a design none of these studies tested.

Two figures often quoted as proof answer a different question. Dropbox's 2010 slide (opens in new tab) saying its "2-sided incentive permanently increased signups by 60%" compares the programme with having none, and the German bank behind the much-cited "16% more valuable" referred customers paid only the referrer (opens in new tab). Three statistics that circulate on referral-marketing sites, referred customers' "37% higher retention", their being "4x more likely" to buy, and dual-sided schemes' "53% more referrals", trace back to a 2009 conference slide, to nothing we could find, and to a vendor survey with no published method.

What conditions and limits do UK referral schemes set?

Beyond the three working parts, schemes add limits, and three of them can be counted across all 27:

The rest are not stated consistently enough to count, but they recur:

• Excluded sign-up channels. Hyperoptic pays only on orders placed directly with it, not through a price comparison site. Surfshark excludes subscriptions billed by the App Store or Google Play; ExpressVPN, which does the same, says that is because "in-app payments are processed by Apple and Google and outside our control". Quickline's link does not cover an order placed by phone, where Airband's does. Ledger pays only on a device bought at full price, not in a sale.
• Spend that does not count. Amazon orders and gift cards do not count towards Rakuten's £50 threshold, and gift cards and no-spend offers such as free SIMs do not count towards Quidco's £5.
• Rules against referring yourself. Quickline and toob require the friend to live at a different address from the referrer, and the BA Amex excludes anyone who has held a personal Amex card in the previous 24 months.
• Rewards that lapse unclaimed. Quickline's voucher is forfeited if it sits unclaimed for 60 days, and toob's gift card has to be chosen within 180 days.
• Clawbacks. InvestEngine (opens in new tab) normally credits the bonus within two business days of the qualifying £100 investment. Both people must then keep £100 plus the bonus invested for 12 months from receipt or InvestEngine reclaims the reward. Outside the 27, Good Energy takes the credit back from both accounts if the friend leaves within six months, one of several energy clauses in our energy refer-a-friend league table.

The limit that most decides how far a scheme can spread is who a link may be shared with. Barclays' terms for its June to October 2026 offer define the friend as someone the referrer "knows personally and trusts"; So Energy's allow no "messages to strangers"; and Capital on Tap's rule out "creating websites … for the purpose of promoting the Refer a Friend Programme". Lloyds sets no such rule, but its in-app link expires 14 days after it is generated. All four were read in August and September 2026. Barclays and Lloyds are covered in our bank referral bonus round-up.

Which UK rules apply to a referral scheme?

Five UK rulebooks reach a refer-a-friend scheme. None stops a mainstream business running one, but each rules out a design that is easy to reach for:

Gambling operators have one more: since 19 January 2026 the Gambling Commission (opens in new tab) caps bonus wagering at 10 times and allows one product type per incentive, a rule written for any incentive and so for a referral bonus. Two questions remain open. The FCA's guidance on financial promotions on social media (opens in new tab) says someone paid when a consumer clicks or buys through a firm's link is likely acting "in the course of business", which is what makes a post a financial promotion, but it does not mention refer-a-friend links; and the ASA has not ruled on whether a customer who posts a code for a reward must label it as an ad.

Should a referral scheme use a code or a link?

Use a link by default, and add a code wherever the link cannot reach the moment of sign-up: an iPhone app install, a sale made by phone or in a shop, a switch of device, or a recommendation made out loud.

We tested a narrower assumption first: that a typed code only makes sense for a product that lives in a mobile app. It describes our data but fails as a rule. Three of the 27 live schemes need a typed code, and so did Starling's, paused when we read it. Three of those four are app products: Tide, where the code goes in during sign-up or within 7 days; THE OUT, whose public referral page carries no personal identifier, so the code typed at in-app checkout is the only thing tying a booking to the referrer; and Starling. But Monzo, Yonder and Zilch are app products as well, and attribute by link. The fourth code scheme, Hyperoptic, counts a referral only on an order placed on its website, and still asks the friend to tick "I was referred" and type the code the link already carries.

The app pattern has a documented cause. On Android, Google Play passes the link's referral details (opens in new tab) to the newly installed app and keeps them readable for 90 days. Apple has no equivalent: if the app is not installed (opens in new tab), a link opens in the browser, and Apple documents no way to hand it to the app after a normal App Store install.

Vendors can bridge the gap, with timing that depends on the integration. AppsFlyer's unified deep linking (opens in new tab) has a 15-minute window, but its extended deferred deep linking handles installs beyond that window using conversion data. Branch's default is 2 hours (opens in new tab), configurable in account settings or overridden for a link. Apple's rules still constrain the methods: it bans identifying a device (opens in new tab) from its signals, and since iOS 16 the phone asks the user before an app can read the clipboard. Google's free route, Firebase Dynamic Links, shut down on 25 August 2025 (opens in new tab). A code typed into the app after installing depends on none of this, which is why Chase UK, an app-only bank, has the friend submit the code in the app (opens in new tab) after joining.

The same gap opens wherever the sign-up leaves the browser the link was clicked in. Virgin Media (opens in new tab) runs both carriers: a link for online orders and a code its phone and shop staff take "at the point of sale". Link-only schemes exclude those channels instead: E.ON Next's terms (opens in new tab) do not count a switch made over the phone, and nor does Quickline. A web journey that changes device, or waits a week, can lose the link too, because Safari caps cookies set by a page's scripts at 7 days (opens in new tab).

Codes cost something a link does not. The friend has to type one, and Baymard Institute's checkout testing (opens in new tab) finds that a visible promo-code box sends shoppers off to hunt for codes elsewhere; it recommends applying discounts automatically, which is what a link does. A typed code is invisible until it is applied: THE OUT has paid me for referrals my analytics never recorded, because no click ever reached a counter. And a reusable code works for whoever finds it, which is why Chase issues single-use codes, up to five per customer.

Where the identifier sits inside a link matters less than it might seem. Firefox's list (opens in new tab) of the 22 tracking parameters it strips contains no referral parameter, and an identifier in the URL path, as at share.octopus.energy or join.monzo.com, is untouched by every stripping mechanism we found.

Cite this study

This study is original research, read from each scheme's own terms in Referral Plug's weekly verification pass of 29 September 2026, and you are welcome to cite or link to it with attribution. The underlying data is a CSV, one row per scheme, with the source we read each one from. A ready-to-use line:

A September 2026 Referral Plug study of 27 live UK refer-a-friend schemes found that 22 paid the referrer and the new customer the same reward, 25 withheld it until the new customer had paid for something or had a service connected, and 24 identified the referral by link rather than by a typed code.

Full citation: Seb Place, "How referral schemes work: a study of 27 live UK schemes", Referral Plug, October 2026. https://www.referralplug.co.uk/blog/how-referral-schemes-work (opens in new tab)

The weekly pass is described in our testing methodology. Two further Referral Plug datasets sit on the best UK referral codes page: an index of every refer-a-friend scheme we track, one row per brand and generated from the brand files themselves, and a dated log of every change the weekly re-read has found in one. Both download as CSV, under the same CC BY 4.0 terms as this study.

Frequently asked questions

How much do UK referral schemes pay the new customer?

Among the 27 live UK schemes we read in September 2026, fixed rewards to the new customer ran from £5 (giffgaff and Zilch) to £100 (Tide and Quickline), and the most common amount was £50: Octopus Energy, E.ON Next, Motorway, Airband and THE OUT pay it, and so does EDF outside its temporary boost. Three randomise the amount (Monzo £10, £20 or £50; InvestEngine £20 to £200; Trading 212 a share worth £8 to £100), and seven pay in something with no fixed sterling value: points at Yonder and the BA Amex, Bitcoin at Ledger and Bitrefill, dollar credit at Proton, free months at Surfshark and a cashback boost at Incard.

Do referred customers stay longer?

The best-known evidence says yes, with limits. Schmitt, Skiera and Van den Bulte (2011) (opens in new tab) followed about 10,000 customers of a German bank and found referred customers roughly 18% less likely to leave, and at least 16% more valuable over six years, than comparable customers who joined another way; the bank's scheme paid only the referrer, a €25 voucher. A 2015 replication (opens in new tab) confirmed the lower churn but not the higher value, and a 2018 follow-up (opens in new tab) at the same bank found referred customers churned less only while their referrer stayed.

What is the difference between a referral scheme and an affiliate programme?

A referral scheme rewards a customer for introducing someone they know, and usually rewards the new customer too. An affiliate programme pays a publisher, such as a comparison or cashback site, commission for the customers it sends, under a business agreement. The FCA draws the same line: its 2022 ban on referral bonuses (opens in new tab) for high-risk investments was "not intended to impact business to business relationships such as those operated by affiliates or comparison websites".