I also removed the coupon code box completely from checkout. I used to get bombarded with requests for discount codes and people not getting their codes to work. In reality the way I used codes were in email promotions. I switched to instant redeeming clickable codes in emails and removed it all the boxes on the site. I also know it was a significant factor in checkout flow because I could google "mybrand.com " into google and google instantly starting adding in "mybrand.com coupon codes" "mybrand.com promo codes". People see the coupon code box and immediately head to Google to search for codes.

PayPal Express checkout is still a huge choice for customers. They can choose express checkout or standard checkout flow. Many people use it (30-40%). It removes the need to enter their address as well.

If your system would still benefit from a discount code and you're not selling low margin products, it's a good idea to "seed" those discount sites with permanent codes for very minor discounts (even just 4%, say). People feel like they've had a win even when the discount is minor.

When I was working with one that was primarily female and cost-conscious, that trick worked great. The fact that a code existed helped drive organic referrals (the code alone was enough to route traffic to me instead of a competitor).

Don't make it permanent or you definitely are giving money away to Googling cheapskates. Change it often-- people are excited by novelty anyway.

There's a number of benefits to supporting both IMO:

If someone is talking about your product on a podcast or audio form, being able to say "enter in promo code BENISCOOL at checkout to save 10%" is useful. But then in the show notes, they can link directly to the URL based coupon link. That is much cleaner than having to explain how the show notes link already has the coupon, and you must click that link specifically to get the applied discount. It also works in cases where show notes aren't available (ie. radio -- although that use case is very rare for a digital store).

Another benefit of having it as a text input is during email promotion launches. For example, you might decide to put up your product for sale as part of an initial promo, but then in your email copy say something like "By the way, the next 15 people who use coupon code ABC123 will save an extra $10". You can't really do that with URL-only codes, unless you happened to include a second link in the email (technically could work I suppose, but it loses its appeal and becomes less clear on how it works).

Lastly, another potential use case for input based coupon codes is doing some form of price parity based on geo location. You may want to show people from XYZ country a way to get a discount by entering in a code at checkout, but having it in non-URL form allows you to word it in such a way that it's optional. So if the person happens to live in XYZ country but they are well off enough not to need the discount, they might pay the normal price to show support.

But like you said, the down side is people might hesitate to find a coupon. That's a very hard thing to measure.

Checkout is super interesting to me because I'm currently developing a platform to sell digital items and the check out flow is what I'm currently developing.

Do they need to have the URL to do so?

Hmm, I suppose one could have a feature, without accounts, where you enter your email address and get emailed to you (at that address), your order history with links to individual orders.

I think you are right that at most ecommerce sites, registration does not provide too much value for the user (and is a barrier to sales). I know I am always annoyed when an ecommerce site asks me to or makes me register, and if given the option almost always choose not to register. (Exceptions for sites I use very regularly; maybe everyone is hoping their site will be one of these. :) )

It really is a big advantage, for whatever reason customers choose it.

Relatedly, just yesterday I had my first web-based Apple Pay experience and man, was it ever smooth and fast. I can see it becoming an equally important 'checkout brand' experience for people in the Apple ecosystem.

Ie does the flow take them straight to checkout, or does it apply the code to any relevant product they add to their cart?

Or you mean I could link to a product page, and when the product is added to the cart the discount is active?

The frustrating thing is this is such an anti-pattern that we _still_ get emails from potential customers requesting the cost of shipping to their country.

If it’s not present (or too well hidden) I usually try another website.

I too hate the unpleasant surprise of high shipping costs

The next question I always get is about how annoying it is to go make an email address in the middle of checking out is. I personally always have a shell open - I'm a command line person. So I just run a script, `nem spammystore-com190621`, done. (And then `dem spammystorewhatever` when they start annoying me.)

I've fantasized about writing a browser plugin that scans the page for entered email addresses and shells out to create them on submit, but (a) I'm not even sure that's entirely possible anymore, and (b) wiring my browser up to something that runs sudo on remote machines seems like maybe not the best idea I've ever had.

But I guess the easiest way would be to throw all the email into a catch-all inbox and filter it by mailbox.

Anyway, I've stopped doing that after I got locked out of a store because I couldn't remember the address I registered, and it would neither let me reset my password without I entering my email or create a new account with the same data of one that already existed. Luckily this one a shitty store so getting locked out was a gain, but the odds of this happening to something important are not low.

These days I tend to find that Google tabs work well enough that I don't really get overwhelmed by all the email going to the Promotions tab.

I think if you’re on OS X that it would be easier to write a small automater action that you register as a service that after you select some text and right click it it parses the text and calls the shell command.

I'd be surprised if there isn't a service for that.

All they had to do was not be greedy-- I probably would have paid up to $10 for shipping even knowing it was inflated, even knowing I could probably find the item cheaper, just to get it over with. If you can't land that kind of sale you're doing something wrong. Of course, maybe I'm the idiot and they're making bank by ripping people off, but I wouldn't want to have anything to do with a place like that.

I think you meant "all they had to do was use enormous volumes to benefit from being a logistical priority and huge economies of scale in their shipping process."

Without further context (or an exact weight, or whether they were gonna use an expedited or overnight service) it's hard to say how cheap they could go, but $25 seems too high. Either the seller is really small scale or they need to renegotiate their rates.

----

\* We're big enough to have a bit of negotiating power, especially with smaller/regional carriers, but we're nowhere near Amazon-scale.

In my view, there are 3 bits of effort for shipping:. Pickup, transit/routing, and dropoff.

Pickup and dropoff are the parts which could benefit from volume, but dropoff can't benefit from volume in the e-commerce case. Pickup and dropoff should cost the same.

That means, in the worst case, a small e-commerce seller sending only 1 item per day could expect to pay double, since they pay the pickup and dropoff price with no volume discount.

Prices for delivery services don't reflect that though, suggesting market inefficiency.

It can come close. Most US carriers (or at least most of the bigger ones) offer a USPS hybrid service (e.g. UPS SurePost, FedEx SmartPost, etc.) that delegates to USPS for the last-mile delivery/dropoff. Unsurprisingly, these services tend to be cheaper than standard "ground" services, since USPS is already driving mail trucks to every mailbox (or at the very least every neighborhood) so their costs are more constant (not entirely constant, since a package takes up more space on the truck than the usual junk mail, but more constant than UPS or FedEx having to drive a truck for a one-off delivery).

But yeah, absolutely brand-new startups will have higher costs. Those go down pretty quick, though, and it might even be okay to take the loss on those initial customers (and assign it to "marketing" or "customer acquisition" or however else you wanna record that in the books).

People absolutely expect to pay for fast shipping; people just expect baseline slow shipping to be free.

> What if they're in another country?

Depends on the industry you're in, but it seems a lot less reasonable to expect free international shipping, unless you're buying something that costs high-three-figures or four-figures.

> If they only order one small item, the shipping cost is a very large percentage of the total cost to the seller.

USPS and hybrid shipping options can be relatively cheap, for domestic shipping.

> you've now priced all your items with the assumption that they're only buying that one item

You could always offer multiple-item discounts at that point.

Actually, they really aren't, not for private shippers. If your item is tiny, you can ship it for around $3 by 1st class; if it's over 1 pound, it's not going to be cheap. For something that only costs $5-10, that's very significant.

>You could always offer multiple-item discounts at that point.

People look at initial price first, and won't see your discount until later. The competitor with a cheaper per-item price will get the sale.

If there are no shipping options, I shop around for the same product elsewhere, express post is usually only $10AUD more and will sometimes arrive 1-2 days later, usually never more than 4 days. Couriers for $30AUD will usually definitely arrive 1-2 days later. Even ordering internationally can be faster than standard post to my address these days.

So I usually bail out of a transaction if there's no shipping options, its easier to just find another source than it is to ask and arrange for shipping personally.

But a similar sized package from an Australian seller will often take three days for the vendor to send then five days to get to Launceston and cost $15 for shipping.

I understand all the factors at play, but still...

I've had parts come from Germany to Perth in 3 days. On the mid-cost freight option. Via Frankfurt and then Singapore.

Australia Post is an essential institution and somehow they manage to be entirely useless.

A shipping estimator should be available on the product page. That's where I need it.

For Austria they use the German shop, but Germany has different VAT and a lot of sellers don't ship to Austria.

When ordering it didn't show how much it cost and if shipping is possible at all.

I don't want to see a Google shopping result saying "Get these shoes for only $10", only to find shipping is $35.

It should be illegal to advertise a product online for anything less than the cost to have it delivered.

Eventually, you get trained to just go to Amazon.

I would suggest that the idea is that losing X% of first-time customers isn't such a big deal if you can increase LTV by Y% through marketing efforts otherwise not available to you.

That being said I don't really think the average marketing team has thought it through to this point (or measures the outcomes). But nevertheless it's perhaps why they see positive results.

It's a good way to battle people otherwise preferring to use a platform they already have an account on (e.g. eBay; Amazon) where you lose more on commission.

If my SO or I are talking about something there, it’s an almost certainty that one of us will press thumb to purchase. It’s too easy.

What proved conclusively positive was adding a discount to the checkout process (a time-limited one did not prove successful). Just to make the customer feel that they're getting a good deal.

Once the customer was on the checkout form we'd display: "This product is almost out of stock," and when they for some reason come back to the product page or listing, we'd show: "43 people are looking at this product."

Which created a sense of urgency. This strategy wasn't legal in all countries, because it was 100% fake, but it worked wonders where it could be displayed.

Customer satisfaction skyrocketed, so... evil as it might be, it only created happy people.

Hotels.com and a couple other websites were doing that to me the other day. I had a feeling it was bogus.

You did have me at discount. I shop the same brands that I like and know. If I’m offered a discount I’ll buy 90% of the time. If it’s a brand I haven’t heard of or bought previously it’s much lower.

I've always ignored those sorts of things entirely, on the theory that I don't believe the entities in question are capable or willing to put the engineering effort to provide a number accurate to two significant digits at scale.

Also the numbers are often implausibly large. If there's "one" of an item left and "43 people" are "currently looking" at it, then the problem isn't that I need to hurry up and buy it, the problem is that you've utterly failed at one of your basic tasks, stocking correctly for demand. In general, businesses are actually really good at that; it's literally life and death for them, so the idea that every single thing on their site is just on the verge of selling out unless you get it now is just absurd.

I've been trying to train myself to ignore "discounts" too, unless I can prove to myself that there really is a difference between the "real" price and the "discounted" price and there is some specific reason that I'm getting the latter. Or to put it another way, "customers in general" don't get discounts. You can get a discount if you are "a price discriminating customer willing to hunt for coupon codes", or "a friend of the proprietor", or something specific, but you are never just handed a discount. Those discounts are discounts... they're just the price.

Everything you do from start to finish in the sales process needs to be believable. Like in a sales page, if you admit that you screwed up and ordered 100,000 sunglasses instead of 10,000 and now you want to get rid of them at whole sale prices, suddenly if you sell $39 glasses for $14.99 it's believable that this really happened and it's a good deal, so the discount makes sense.

Now of course that could all be a lie and they cost $5 and they really wanted to sell them for $15 so they cooked up a total bullshit story but something like that is a lot more believable than trying to throw up a bunch of high pressure sales tactics based on scarcity. Picking out those lies really comes down to your reputation too, which is why sites that offer 24/7 discounts have no credibility.

Basically, just be honest and it's hard to go wrong.

Depends on your country, but in the US you can't claim something has a regular price if the regular price is never charged for it.

But if you get a discount a non-negligible amount of the time, you may stop buying unless you're getting the discount. The discounted price will become the 'true' price.

Definitely guilty of this, if I'm on the fence about a purchase, finding a discount code (Honey extension is great) can often make the difference. Even knowing that the discount is probably baked into the price

Only articles about the finer points of segment tuning and tweaking x86 assembly registers! Or all the reasons why classes can never really approach the elegance and terseness of classic K&R C structs. Or how Common Lisp and Scheme have this and that and the other but only that and no more :)

The former are really life skills that make even the most hardcore techies more effective human beings!

Or ... perhaps many HN folks just hate the concept of hacking the human brain to generate more sales.

Just 2 days ago I purchased very expensive laptop via multi page checkout process which I really appreciated.

The key was to remind customer what he is buying, all options, warranty, shipping etc - make sure the complicated configuration is delivered in clear way.

Gigantic one page thing would be less welcome

A big one page checkout could work on desktop machines since they have the real estate to work with. This can be a poor experience on mobile as you need lots of scrolling to access all the fields and options.

Whereas if you're just buying some simple items that have no options at all, you don't need a complicated checkout, just a list of the items and the taxes and shipping cost.

Multistep forms trick the user whether it's intended or not: you're either collecting a little bit of information knowing the user might not complete the process, or you're pressuring the user into completing a process they wouldn't otherwise complete because they've already sunk so much time into it, when they wouldn't have ever embarked if they had known from the beginning all you were going to ask for.

There's a segment of HN that argues that market incentives will always lead to good results for consumers, but here's a clear example where people clearly only care about maximizing conversion and collecting the most data, with no concern for the consumer.

I wonder how the 'create an account' problem is best solved. Some sites (like Bandcamp) explicitly don't require accounts, and just associate purchases with e-mail addresses until you create an account. Other sites sometimes will give you an account with just an e-mail address, then allow either resetting password or 'magic links' (a la Slack) that log you in from an e-mail. I think for simplicity, it's probably best to abstract e-mails from accounts, so from a data model PoV, which thing do you ideally associate a purchase with, a phantom account or email addresses directly?

Side note: I worry that skipping the cart review mostly loses customers because they begin having second thoughts. While adding this step may reduce conversion, it may be more ethical to include it as it may improve user's decision making.

I’ve personally abandoned plans to purchase because it wasn’t clear if I would get a chance to review at the end. Its especially important if its unclear what the final price will be, after applying shipping and such, since I live outside the US. Anecdotally, I’m much more likely to buy if its very clearly stated that there is a review step.

I know that more impatient people might be the opposite way around, though.

Most checkout pages I've seen have that information, so even if the confirm and pay steps are on the same page, I don't mind since it's still the same information anyway.

And yeah, using a mail address for login is certainly better than coming up with a user name, which is free and which you remember in a year's time.

If low friction truly is the goal, why even ask before you clear the payment? Make it an after-checkout option.

Honestly, this sounds like a feature of GDPR, not a bug. It makes pervasive data collection come at a cost (here in terms of friction).

A _lot_ of data is necessary to call a result if the effect is relatively small. If not enough samples are taken, there will still be a "formal" result, but the level of certainty of it will be low. Of course, depending on how accessible that page is, it might take a very long time to achieve significance.

How much certainty do we need in the result? P-values are generally considered not a great measure, but they are standard, and they are what they are using (if I understand correctly). This can give you an idea of the numbers we'd like to see for that value: https://xkcd.com/1478/

In other words, only the first A/B test has conclusive results: accordion is worse than wizard by 30% with 0.01 p-value. Of course, you can see that there's a significant result _because the effect is large_. Note that this gives us an idea of the _resolution_ of their test system (i.e., how many As and Bs they have and how much effect they can see with their test). So we know they see an effect of about 30% change.

Now: all the other results are wrongly reported. The next four are mistakenly reported as "no statistically significant difference in performance." This is a false statement. Having high p-values means that any effect is below the resolution achievable with those numbers. In other words, the correct statement would be either "the experiment was inconclusive" or "the effect, if any, is less than 30% change". This is quite different from claiming "no change."

The last result has 28%, and it has a p-value of 0.05. I would _not_ accept it at face value. Such a p-value means the result has 1-in-20 chances of being due to a random fluctuation of data. There are six such experiments just in the post, so it's not unthinkable that the effect could be a fluke. More info on this here: https://www.xkcd.com/882/

All in all, the only result found is that _in their case_ the wizard was much more effective than the accordion. The rest is not supported by the facts presented and furthermore any generalization of the result outside their particular case is unwarranted.

No...that's what "no statistically significant difference" means.

In these kinds of statistical tests, literally all you can look for is a statistically significant difference. Either you find one, or you don't. Once you know that you've found one, you can say what size it is—a difference of 0.1% is still plenty statistically significant if your p-value is <0.001. It's just not very practically significant.

There are a lot of different reasons you might not find a statistically significant difference, and simply knowing the p-values cannot tell you which ones are true here. That's why a lot of journals (at least in Psychology, the field I have some knowledge about) have been pushing for reporting additional tests and measures.

They mean a lot less than people assume, and in some cases, as you suggest, something different than they assume. (Ie, "statistical significance" relates to confidence that the results showed a reproducible correlation (and are never 100% 'proof' of this, just a level of confidence; and are totally ruined if you've been doing data dreding/p-hacking), rather than any kind of magnitude of effect).

Nothing can be said of changes of less than 30%, and that's a hell of a lot of uncertainty. One of the variants could give you 25% more results and yet, those experiments would not see such a result.

As such, it is clear that the experiment is simply not thorough enough to imply the two alternatives are equivalent. Not in the least.

Simply saying that the statistically significant effect they found was 30% has absolutely no bearing on how large an effect they can detect...

If group B had a result of 10% with the same confidence, that would also not indicate that their measures have a resolution of 10%—those things literally have nothing to do with each other. You are reading far too much into the very limited data presented—not merely more than you can prove, more than there can be any possible logical support for.

I'm designing a checkout page right now where I planned to do a small 1 pager with an email address + minimal billing info required to make a transaction while minimizing CC fraud. Knowing the email (aka. account) before the checkout actually happens is essential because it would allow you to prevent someone from buying the same item twice. If you don't collect an email address pre-transaction, how would you do that for a digital item?

---

[0]: https://web.archive.org/web/20190621031723/https://capitalan...

EDIT: this is a clear GDPR violation - you can't store and use user data without a lawful basis. Marketing requires consent. Consent has to be given freely.

EDIT: even if you worm out of the fine people hate getting unsolicited email like this so some non zero percentage of potential customers will avoid your business in the future.

That would exclude Remarketing services from GDPR-compliance, wouldn't it?

> even if you worm out of the fine people hate getting unsolicited email like this so some non zero percentage of potential customers will avoid your business in the future.

I'm not so sure about this. It may well be true for you and me and our peers, but I don't believe that companies send out annoying "newsletters" and marketing emails just to drive away business, annoy consumers and wake up regulators. They do work for a certain audience.

As an old school mail order pro once said "there is no such thing as junk mail, just a poorly targeted list". If McLaren (to name one of my favorite companies and website) started spamming me with the latest low-down on their hypercars I'd be in heaven... Paul Graham Essays, Jay Abraham ebooks; same deal--pure bliss.

> It may well be true for you and me and our peers

You and me and our peers are a non-zero percentage.

As for creepy: I wouldn't consider these creepy (i.e. where I've entered my email), but maybe annoying. I had the creepy experience with Pavlok when I first looked into getting one. Browsed around on their site but didn't put one into the cart. Hours later I get an email from them. Turns out they were using some service that loads a beacon and tries to connect the browser via various third party services to an email address. That was creepy. I still ended up buying a Pavlok though, but at a later time.

If I abandon a basket in a retail store, I don't expect a phone call about it hours later after I've left the store and gone home. The first time an online-store followed up with an email about my abandoned basket, I stopped using that store.

Not naive enough to think other stores aren't tracking these metrics, but it's still creepy unless you've been subjected the obviousness of it a lot.

Very curious about your experience! Thanks.

Good to see you're still around, I've enjoyed using my Pavlok quite a bit.

Can't even tell you how many times I dealt with complaints about shipping costs, shipping times, customs costs, returns, etc. Whenever we'd get a complaint we'd just cancel the order and send them a link to a competitor.

Good riddance.

I could invisage a SaaS without ANY personal information, but if that even legal to do if you're taking credit card payments?

In e-commerce cases, you're usually just fine with an address (since you need to ship), an email or phone or some kind of order tracking ID.

In most SaaS cases you can ask for nothing - literally - as visitors try out your product. Allow them right in, and sprinkle a few "sign up" links or forms where appropriate. Build trust, basically.

More often than not, you can delay creating an account proper until you're beginning the formal trial period or processing the credit card.

/s...?

They got the legal compliance as a checkbox during the process. The whole process had 15+ steps and it was when I learned that we were all wrong about how many clicks and actions a user is willing to go through to get a result.

Feels like filling out a paper order form to buy something but not mailing it.

I think the whole "if something can happen, it will happen" rule works nicely here. You can write on paper all you want, but the recipient will never be able to know what you wrote unless you explicitly mail them or send them a copy of the paper. On the other hand, for the web, it is very possible for the client to send the server information at any time.

Say you phoned a company to enquire about some new windows and receive an estimate, if they call you back a few days later you wouldn't be that surprised. However you wouldn't want them to phone you every month until you buy from them, or to give/sell your phone number to another company without consent (maybe you need roofing too?).

This is based of an assumption that cart abandonment is some sort of mistake. What I would guess usually happens (based on my experience with shopping online) is that stopping just one step shy of actual payment is the only way for the customer to get the total cost of orders, that includes discounts and shipping costs. When I close the page at this point, it doesn't mean I got distracted - it means that once I finally got to see the actual costs, I decided not to proceed.

> Say you phoned a company to enquire about some new windows and receive an estimate, if they call you back a few days later you wouldn't be that surprised.

I wouldn't be surprised, but I wouldn't be happy about it either. If I asked for an estimate, I asked because you didn't publish the costs up-front, and I need them to make my decision (whether by comparing with my budget, or the price offered by competition). If I wanted to buy from you, I'd call again.

You would need to show that a person entering their email into an input expected that email to be stored by the other party without submitting the form.

Unless you have a commercial relationship with a customer, don't assume you can email them anything or phone them up.

If the company calling you back isn’t unsolicited (ie I asked them to or they told me they would and I said OK), then its an expected part of the interaction. Cart abandonment emails are not expected or even necessary (from the customers point of view). If I called a company and they called me back days later, without telling me they would or asking if I mind, I would absolutely report them for that too. (At least in my country, data protection says they can’t do this)

Cart abandonment is just like any other marketing and honestly, I report any cart abandonment emails I receive as GDPR violations to my local data protection commission.

[1] https://europa.eu/youreurope/business/dealing-with-customers...