Reproduced from Jack Kinsella’s Guide to Conversion Rate Optimization
A High-Level Catalogue of The Main Areas to Fiddle With When Optimizing Your Conversion Rate
This is the thinking person’s guide to conversion optimisation. Instead of listing 998 disconnected, free-floating tips, this essay packages up the field into 10, high-level strategic themes. The idea is that you’ll leave this article with a solid mental framework for thinking about, analysing, and coming up with your own conversion optimisations.
This is written for e-commerce companies, SAAS businesses, and even the neighbourhood dentist.
Without further ado, here are the ten main areas to fiddle with:
A. Conversion funnel length
Every stage of a conversion funnel has some leakage (40% of traffic seep away here; another 22% dips out there…). Website owners invest untold energies into patching up these leaks so that even more leads reach the end of the funnel and convert.
But no stage of a conversion funnel will ever be completely sealed against leakage, and, for this reason, the most efficient possible modification to a funnel is usually to rip out entire stages.
Do you really need a separate page to collect billing addresses when you deliver your service digitally and can attain the necessary taxation information about that customer from Paypal’s API? Is that extra “confirm your order” page really necessary, given that (1) the customer has already paused to enter in their credit card details on a prior page and (2) you offer 100% refunds, no questions asked? Is it really necessary for the customer to register before purchasing? Sure, this may be standard practice, but perhaps the crowds are mistaken? Jared M. Spool writes in The $300 Million Button that a mandatory registration step intended to ease repeat purchases only served to drag down their conversion rates. His usability tests found that customers were “not here to be in a relationship,” and that the registration requirement deterred them, causing them to abandon their purchases. The fallout was phenomenal: By removing this registration step, sales increased by 45%, and the retailer increased revenue by $300,000,000 that year.1
Amazon’s one-click order feature seems to be motivated by similar considerations; it must surely count as the ultimate exercise in minimising funnel length.
Trimming funnel length isn’t just about cutting the number of pages in the flow on your website; more generally, it’s about reducing the amount of tedium and distraction that a customer must trudge through before completing their order. From this point of view, you should consider the following factors when optimising your funnel:
1. Steps behind the scenes
There exist conversion steps that customers pass through even though these steps don’t correspond to individual pages on your website.
Perhaps the most commonplace example of this is the confirmation email click required by some websites to finalise new user account registrations. This step is a 10-ton conversion dampener.
Another (more subtle) dampener would be the very presence of a coupon box on a checkout page. This box alerts the potential customer to the possible existence of discount codes and invites them to leave your website and google for coupon codes, or subscribe to your mailing list and wait until a discount becomes available. In both cases, this distraction might cause you to bleed leads and lose sales.
Yet another conversion dampener in this category is asking a lead for information not normally held in memory—e.g. passport numbers, VAT numbers, scans of documents, or credit card limits. This information demand compels anyone without a reference at hand to drop out of your funnel. If at all possible, ask for such information after the sale has been completed.
2. Number of form fields
Funnel length is not just measured by the number of distinct pages; after all, even a one-page funnel can be exhaustingly long. Each additional field that needs to be filled in counts as an extra step in your funnel, regardless of whether this additional field shows up on the same or a different page. Do you really need to ask the customer how they heard about your website? Do you really need to ask them to upload a document that they probably won’t have at hand right now?
The more form fields you add within a funnel, the more you risk losing your potential customer. This suggests that, when optimising your funnel, you ought to remove any fields that aren’t totally necessary, or least make as many fields as humanly possible into optional ones. You can always ask for this information after the lead has converted into a paying customer.
3. Painfulness of form fields
There are all sorts of things you can do to make your form fields less onerous.
Then there are autofill fields: If someone enters their street name, you should populate their ZIP code without the user having to type anything. Indeed, maybe you can even prevent the need to type in ZIP codes at all, at least when you’re able to detect it from their browser’s geolocation. As well as all this, you should always autofill any form fields with any existing data you have about that user, saving them the need to type again.
The third measure in reducing painfulness of form fields is making your input validations more forgiving. You really shouldn’t be rejecting people’s credit card numbers or phone numbers just because there were hyphens or spaces in places you didn’t expect. Take a little time to write validation code that’s a little more accepting of your customers’ input. There’s no point in having a bouncer so strict that no one gets into to your club anymore.
4. Steps added by third-party providers
Many websites collaborate with third-party web services which require the customer be sent away to the third party’s website at some point during the sales funnel. A common example would be Paypal briefly sending customers off to Paypal.com for credit card payments. These off-site service providers add a number of steps to the sales funnels, so it’s important that these extra steps be reckoned when calculating total funnel lengths. Thus, when choosing third-party software service providers, carefully consider the number of steps they insert into your conversion funnels. It might well be a better idea to choose a service provider that charges double or even triple as much, so long as they remove a step of your sales funnel and thereby improve overall conversion rates.
5. Number of choices a customer must make
Professors Iyengar and Lepper found through their experiments that people are more likely to buy products (jams or chocolates, in their case) when presented with a small number of choices (6 items) compared to when presented with more options (24 or 30 items).3 By extension, stocking excessive numbers of similar products on your website can also hamper conversion rates. I hypothesise that this is because having additional products introduces an implicit extra step in your conversion funnel: the comparison stage. Potential customers who are straining to tell the difference between many almost indistinguishable products may suffer a bout of choice overload that could poison an otherwise sure-fire sale. Remove this comparison stage by stocking less items.
B. Funnel order
Once you’ve trimmed down your funnel length, it’s time to start thinking about the order in which you display the various retained stages.
Much like difficult family relations, certain stages of your conversion funnel are here to stay—regardless of your feelings towards them. Whether it’s the need to ask for credit card details before payment, scans of passports for government-mandated identity verification, usernames and passwords for account creation, or log-in details for a connected Twitter account, these annoying stages often serve to scare potential leads off the path to conversion.
As a website owner, you don’t realistically have the option to banish these steps outright, but you can shuffle them about so that they rear their annoying heads only when their corrosive effect on conversion rates is minimised. Just as the waiter presents the restaurant bill only after giving you a free shot of mango schnapps, it may be more prudent to ask your customer for payment details after they’ve scheduled a hairdresser, masseur, or cleaner to arrive this afternoon. Relieved that their desired service is already on its way and afraid of having to go through the whole process again on some other website, the customer will be motivated to endure the pain of entering in credit card details.
How do you decide how to reorder conversion funnel steps? This choice involves two classes of considerations. The first is the effect on conversion rates (to be determined by experimentation). The second class of considerations are those pertaining to technical/programming issues. Allow me to make two general observations as to the flavour of these technical issues:
Observation 1: The further you stray away from that which is well-worn in the world of web architecture, the more you should expect trouble and expense with implementation.
Here is one of many examples: Most websites use “batteries-included” libraries for managing their login and logout functionality instead of building their own solutions from scratch. This saves days or even weeks of development time. But these libraries, being standardised, assume typical, everyday website flows (e.g., the user creates a user account before having access to special features or being allowed to pay). Generally speaking, these libraries are unsuitable for alternative website flows (e.g., ones which delay user account creation until the last possible moment). If you wish to keep using the standard account management libraries, you’ll need to resort to custom hacks like the creation of zombie user accounts that must later be awoken. But all this customisation causes development time and bug incidences to balloon.
Observation 2: The more your architecture allows a user to postpone inputting information, the more you should expect problems attributable to this information being missing.
This is because some proportion of your users will never return to provide this info, causing postponement to drift into chronic absence. This causes a schism to form between the users with and without this data, which in turn requires that your website’s algorithms and administrators be trained to also handle the cases when data X or data Y is missing. What’s more, you’ll also have to decide how to respond when data which has long been missing finally appears. For example, the authors in my notes publishing platform are paid monthly royalties to their Paypal accounts. But because I allow authors to set up shop without inputting their Paypal email straight away, it sometimes takes authors up to a year to send us their Paypal account details. Once these details are received, our software needs to calculate the royalties accrued up to that date and pay this amount. The software needed to do this requires a fairly substantial amount of code, none of which would be necessary if we had everyone’s Paypal account from the get-go. (But, on the flip side, requiring a Paypal account up front would drive away too many potential authors, so I believe that this trade-off is justified. This shows that the bump in conversion rates came at a cost in technical complexity…)
Some software companies have adopted what at first seem like outlandishly long conversion funnels. Central to their strategy is a free demo approach, which involves an inversion of the normal business sequence of paying today and receiving tomorrow. Their funnel sequence goes like this: “Create your account today, enjoy our service for the next 30 days free of charge, then begin paying monthly, but only if you want to continue using us.” The power of this “free demo” approach is that the user becomes gradually dependent on the service, having grown accustomed to the interface, trained their employees, integrated the software with the rest of their toolkit, and generated a history of data sitting within the platform. Assuming that the software works reasonably well, inertia all but guarantees that the user won’t swap the supplier for another when the bills eventually start arriving. The effort involved in switching becomes too high to justify.
C. Number of clickables per page
Next time you are shopping on Amazon, pay attention to how their website sheds navigation menus as you tunnel deeper down their conversion funnel. The reasoning behind this practice is that every clickable link, photo, or button is a potential distraction that can potentially divert a lead away from buying something. This distraction happens in the dullest of ways—for example, someone finishes typing their address in the second-to-last stage of the sales funnel and then decides, out of curiosity, to click the link to the “team” page before paying. Once on the team page, this lead reads that the company’s CEO used to play in a punk band, so off the lead goes off to search for the CEO’s old band on Spotify. Swoosh—the lead has left the site and will probably never return to complete their purchase. There’s something almost cartoonlike in these precautions: If we leave a big red button in the middle of this airplane control panel, someone will inevitably succumb to their curiosity and slam it down.
Cautious about avoiding this kind of funnel leakage, conversion optimisation experts advise restricting each page to a single goal—one that is big, totally obvious, and eminently clickable, and that leads inexorably towards conversion. In the same way that a forking river divides its waters, each clickable on a webpage diverts away a stream of potential conversions. This leakage ought to be dammed: Once the customer is heading in the right direction, limit yourself to only including a clickable that brings them closer to the final goal.4
Since the dawn of the web, internet marketers have loved pop-ups and overlays. The reason why couldn’t be simpler: They work.5 I mention them here because I would venture to guess that a significant part of the reason why pop-ups are so effective is that they reduce the number of clickables on screen and focus attention on some singular goal, such as handing over your email address for a mailing list.
On a similar note, you should avoid linking to external websites in your conversion funnel. Yes, the glowing testimonial on your product page may have been written by a celebrity with a large Twitter following. And yes, you’d like to emphasise the credibility of the testimonial by directly linking to the original utterance. But doing so may have the unwanted side effect of delivering your customer over to the finely tuned engine of infinite distraction that is Twitter, swiftly sweeping your sale away. (Those preferring a compromise might choose to continue linking to the Twitter account, but programming these links to fire up new browser tabs or windows, thereby keeping the customer focused on completing the funnel.6)
D. On-page copy
Wording strongly influences conversion rates. Although finding the best variations takes time, the journey can be sped up by focusing on trying out radically different ideas as opposed to slight variations on what currently works best. The retailer selling shoe polish for the business executive shouldn’t limit their search for the best product ad after pitting eight mildly different catch phrases about perfectly polished shoes against one another; instead, this retailer should test marketing messages corresponding to radically different themes: business success, confidence and status boost, style, convenience when travelling on airplanes, ease of application, ecological friendliness, or money saved by prolonging a shoe’s life.
Calls-to-action are your on-page copy superstars. These are the tiny phrases of imperative-mood language contained within or next to your form buttons and funnel continuation links. Despite their potential to elevate conversion rates, calls-to-action are all too often overlooked by internet marketers who instead leave them set to their ugly defaults. But this is unwise. Asking website members to “join now” conjures up images of a religious cult recruitment drive, and unless you’re in the business of sadomasochism, you probably shouldn’t be asking customers to “submit”. Rather than these uncreative and uncompelling defaults, your calls-to-action should remind the lead of the benefit you promise to provide (e.g., “Remember Everything” (Evernote, a note taking and syncing service), “Start Selling Today” (Square, a credit card processing solution) or “Start Meeting People Today” (OKCupid, a dating website).
I’d like to draw a distinction between copy changes that merely play with words and copy changes that have far-reaching consequences for how you do business. For instance, suppose you modify copy so as to promise the customer “refunds for life”. To honour this promise, you’d better have the technical capacity to refund money months or years later, the administrative capacity to update your accounts (otherwise you may be taxed on money you didn’t make), and the website features to prevent further delivery of your service to a refunded customer. Even though your copywriter only needed to add the three words “refunds for life” to the webpage, the assumptions behind these words could require months of changes tallied in terms of new features, modified workflows, and retrained staff. Not all words are cheap.
It won’t exactly be ground breaking news to you that images are one of the most important elements to experiment with when optimising for conversions.
As a general rule, the more immersive the image, the better. Images with a 360-degree spin feature were found by DueMaternity.com to give a conversion lift of 27%.7 Larger product images were found by car manufacturer Hyundai to improve conversions by 62% compared to thumbnails.8 On a similar basis, having an image zoom feature is likely to improve conversion rates, as does having additional images depicting your products from various angles or used in differing contexts.
Product images should not be reserved for the product sales page alone. When online retailer Brickhouse Security added mini product images to their drop-down on-site search, they boosted their conversion rates (of customers who searched) by no less than 100%.9
The choice of image matters too, especially since the usual raison d’être of an image is to transmit a carefully chosen marketing message. Web hosting company Hawk Host was able to more than double their conversion rate by swapping a photo of a globe for a photo of a giant padlock. This change helped them to visually emphasise their company’s focus on security.10 Another example comes from Apple: They advertise their laptops with photos depicting them being slid into envelopes, showing us just how tiny their computers are.
Another effective use of image is to demonstrate how a product looks while being put to use. This helps the potential customer imagine how the product being advertised could deliver to them the benefit they desire.
Generally speaking, photos of human beings tend to convert better than photos of things. One company improved their conversion rates (of people who contacted their customer support) by 21% just by adding a photo of their customer support representative.11 (Actually, this company tried out photos of two different representatives; one of them only added 10% to their conversion rates. This goes to show that you should vary which people you display on your pages.) Basecamp (formerly 37signals) improved their conversion rates by 100% when they redesigned their home page to focus on a picture of a customer instead of information about their software products.12
Another advantage of using photos of people in your copywriting is that their facial expressions and postures communicate emotions. This can help you to conjure up feelings of comfort, happiness, and more in the hearts of your leads.
One final piece of advice: The pictures on your website should not be stock photos. Even the most amateur pic you yourself take is likely to substantially outperform stocks. An experiment run by a moving and storage company found that a crappy image of their truck beat by 45.05% a stock image of an attractive couple lifting a moving box.13 (A comparably amateur image of their team of workers wearing their blue overalls performed marginally better, giving a conversion lift of 45.45% over the stock image.)
The most successful conversion optimisation I ever instituted in my business came from a change that took all of half an hour to implement: I raised the prices by 20%. As I did this, I feared that enraged customers would throw a virtual brick through my shop front, refuse to pay the increased price, and decry our greed in every internet forum there is. In fact, nothing of the sort happened. The number of sales stayed exactly the same, causing profits to increase by the 20% that I had raised the prices by. I won’t deny that occasional customers (say five per year) contact us to say that our prices are too high. But these protestations aside, our prices don’t seem to inhibit actual buying behaviour. Customers continue to convert at practically the same rate at our higher price point. Indeed, some of the very same people who complained about our price increase nevertheless went on to purchase our products.
I cannot stress how big a difference raising your prices can make. This one tiny change might dwarf all the gains you achieve through every other business activity in the next year.
Increasing the amount you charge for a product is just one way of tweaking pricing. But there are other, more creative ways, to increase profitability by experimenting with new billing models. An acquaintance running a digital mapping service for cyclists used to charge his customers a monthly fee for access to all the maps in his system. He dramatically increased revenue by switching his billing model such that customers would instead buy maps of smaller regions. The customers would then “own” these maps forever instead of paying for them with a monthly fee. The typical customer likes the idea of buying to keep, and, moreover, most people have a deeply ingrained instinct to complete collections, with the consequence for my acquaintance that many of his customers bought every map in his system, regardless of whether they would ever use them.
All things equal, money now is better than the same money later. For that reason, it’s good business to give discounts to customers buying a year’s worth of service in advance (as compared to customers buying your service one month at a time). When you bill monthly, the customer may change their mind at any point and switch to a competitor. Had you billed yearly, the customer might have stuck around long enough to become so invested in and hooked on your product that they would no longer consider switching once the next billing cycle comes around. In addition, revenue booked now—as opposed to later—can be used to build new features or to woo new investors. Therefore, it’s worth experimenting with extended billing cycles as part of your conversion optimisation efforts.
I’m a fan of bundling, which is when you package a bunch of products together into a discounted collection. Although the margin on each bundled item drops because of the discounting, the average sale size increases thanks to customers now buying three items instead of just one. When done right, this increase in total revenue should more than offset the reduced margin, spelling increased overall profits. If you want to really boost bundle sales, try making the constituent products massively overpriced when bought separately. Although this will sabotage sales of the unbundled products, it will boost sales of the bundle by making it seem like comparatively great value.
Businesses which have corporations or public entities as part of their customer base may consider pricing schemes which capture exponential value from these large and wealthy entities. The typical software service provider has a linear pricing system, one that scales with usage (such as “$5 for each team member” or “$20 per server per month”). These service providers would therefore charge a corporation with 1000 team members the same amount of money as they would 1000 personal team-of-one customers. By contrast, a service provider billing with an exponential value pricing scheme would charge corporations exponentially more by creating special “enterprise” plans. These plans contain features that large corporations must have, be that for regulatory reasons, security reasons, tax efficiency reasons, data ownership reasons, branding reasons (e.g., white-labelling), or integration assistance reasons. Under such a scheme, the software provider would instead charge the corporation with 1000 team members five times as much as they would charge 1000 individuals. The higher price charged to corporations could even be used to subsidise the costs of individual accounts.
Sometimes, changing your price tag or your billing model is too daunting a task to undertake. In these situations, there’s always the possibility of opting for a merely cosmetic change, such as reframing the current price. Service providers who bill monthly could deemphasise their month charge (e.g., £30 per month) and instead draw attention to their lower amortised daily cost (e.g., £1 per day) or the average price per item (e.g., £0.33 per episode watched, assuming the average user streams three shows per night).
So far, we’ve shown how you can dress up your price in more attractive monetary terms; there’s also the possibility to frame your price in terms of non-monetary benefits, such as the number of hours of work saved or the number of server downtime incidents avoided.
Sprinkled through your codebase and business processes are variables determining durations of time. These represent things like how long your software waits before sending a discount to people who abandoned their carts, how many seconds it waits before displaying a pop-up to website visitors, how many days it gives away as part of a free demo before throwing up a paywall, how many times per month it emails the people on your mailing list, or on what date it switches on the holiday advertising campaigns. The online marketer concerned with conversion optimisation should be prepared to tweak these timings.
Time influences conversion optimisation in another respect: Sluggish website response times stifle conversion rates. Thus, a “quick win” comes to those who monitor their website logs (or New Relic/Google Analytics reports) for abnormally slow pages (e.g., those with response times > 200ms) and focus on making these pages snappy again.
H. Salience (includes graphic design)
Just because you, the author of your website, are intimately aware of every word on your website, doesn’t mean that a random visitor will be. The average person probably spends mere seconds on your website before they move on. It is the job of your visual design to hook passers-by into your world before these precious seconds are up.
The visual design of copy elements—such as their relative size, placement, spacing, or colour—matters. A headline in font size 48 at the top of the screen is going to receive dramatically more attention than your 6th bullet-point out of 10, rendered in font size 11 and submerged deep below the fold. This implies that moving different messages into and out of your design’s visual spotlight will affect your conversion rates. Rotating your existing pieces of copy to new positions on your website can potentially be as effective in improving conversion rates as coming up with brand new elements.
As well as an element’s font size or positioning above the webpage fold, there are various other visual cues and tricks that a marketer can employ in drawing attention. One is simple visual contrast. If you happen to have one paragraph of text on your webpage contained within a box with a coloured background, visitors will be more likely to see it and therefore to read its text, if only for the simple reason that it contrasts with all the other boxes on the page which are set against white backgrounds.
In fact, this brings us to a more general point: Movement is a sure-fire route to salience. One example of this idea in practice is how some e-commerce websites pop up a customer service representative chat bubble after 20 seconds of browsing. Another example would be “Please subscribe to our mailing list” bars that appear over a website’s header or footer after you move the mouse in a certain way. Yet another example is the updating of an on-site notification counter to indicate that a new message has arrived, as in Facebook or Twitter. Another example again would be the sudden appearance of a breaking news story at the top of a newspaper website’s homepage. In all these cases, the sudden change to the website visuals steals away your attention.
Salience doesn’t just happen by spotlighting the element in question; it can also happen by de-emphasizing non-important aspects of the page. That’s why some pages grey out their backgrounds when they want you to focus on converting.15
Another way to draw attention to your desired goals is by using directional cues, such as the cheesy-looking hand-drawn arrow depicted above. Classier marketers can achieve this same effect with subtlety, for example by inserting a picture of someone whose eyes are gazing directly at an important button, or by having a picture of an animal (e.g., a squirrel or a lizard) whose head is pointing towards the critical sales message.
I. Choice architecture
Owing to our human proneness to inertia and herd mentality, most of us tend to stick with whatever is presented to us as the default option. Professor of Psychology and Behavioural Economics Dan Ariely observed that in countries where being an organ donor is the default, the overwhelming majority of citizens “sign up”. For instance, 99.9% of Austrians are donors. By contrast, in countries where citizens have to explicitly opt in to organ donation, donor levels are dramatically lower: Only 12% of Germans are donors, despite their being culturally similar to Austrians.16
The study of how outcomes can be influenced by the way choices are presented is known as choice architecture. And, as you can guess, this field can be applied to conversion optimisation.
We’ve all seen those checkboxes at the bottom of web forms that read, “I wish to receive periodic updates about new offers”. As a webmaster, you can decide whether you leave this box ticked or unticked by default. Amazingly, only 18% of users ever change this default.17
There are other ways to leverage this idea in your business. You might suggest one subscription package as a default over the others; you might remind customers that it is customary to buy product X with product Y; or you might go full evil and include some opt-out “extra” (e.g., insurance) in the cart without the customer explicitly adding it. (See low-cost airlines for more on this shady practice.)
J. Available features
I feel a little cheeky suggesting new features as a conversion optimisation tactic. Conversion optimisation is supposed to be about quick wins, whereas building new features involves actual work. But even taking this effort into account, the jump in sales that could follow the addition of a feature highly valued by your customer base could make all the toil in the world worthwhile.
There is a low-cost strategy for web businesses unsure about investing in features of unknown contribution to future conversion rates. Instead of actually building the feature, these businesses could merely promise it within their website copy and observe how this promise influences conversion rates through A/B testing. If these tests demonstrate that the promised feature improves conversion rates, then the business could confidently proceed to build it. (Any customers promised the non-existent feature could be told before the payment step that the feature doesn’t currently exist, or they could be offered a refund.)
Deciding what new feature to promise needn’t require any creative genius—a change as banal as adding an additional payment option can suffice for a revenue gain that would be the envy of many companies. For example, some customers would never pay with Paypal because of a lack of support in their country or its unattractive currency exchange rates. At the same time, other customers, calmed by Paypal’s internationally recognised brand name and many years in business, would not trust any provider except Paypal. As such, a change as simple as wiring up another payment option can cure many a conversion optimisation ill. A/B tests enable you to discover these efficiencies without the hassle of actually building them.
This content is an excerpt from EntrepreNerd, my 337 page book on internet marketing. Buy it now, you won’t regret it!
Beginner’s Guide to Conversion Rate Optimization: deals with the pros and cons of multivariate testing, the statistical side I left out.
- https://articles.uie.com/three_hund_million_button ↩
- In-line validations brought a 22% conversion lift – http://www.getelastic.com/real-time-inline-validation ↩
- https://faculty.washington.edu/jdb/345/345%20Articles/Iyengar%20%26%20Lepper%20(2000).pdf ↩
- For example, one company changed from having two buttons to one, improving their conversion rates by 36.9% – https://blog.kissmetrics.com/100-conversion-optimization-case-studies/3/ ↩
- Brian Dean at Backlinko reluctantly installed a pop-up for collecting mailing list subscribers, only to discover that it more than doubled his daily sign-ups. He decided to keep the pop-up around, despite his misgivings – http://www.appsumo.com/sumome-double-conversion-rate/?rf=brws ↩
- Guide to programming some links to open in a new window/tab: http://stackoverflow.com/questions/15551779/open-link-in-new-tab ↩
- https://www.internetretailer.com/2011/05/26/rotating-images-generate-higher-conversion-duematernitycom ↩
- https://vwo.com/blog/multivariate-testing-case-study/ ↩
- https://www.internetretailer.com/2010/04/29/product-images-site-search-window-boosts-conversions ↩
- http://blog.hawkhost.com/2010/02/21/multivariate-testing-a-real-life-example/ ↩
- http://www.entrepreneur.com/article/220252 ↩
- https://signalvnoise.com/posts/2991-behind-the-scenes-ab-testing-part-3-final ↩
- https://vwo.com/blog/stock-photos-reduce-conversions/ ↩
- Read this page until you notice the jiggling ‘Download’ button: http://ghostinfluence.com/reddit-traffic-explosion-deconstructing-the-framework-and-tapping-its-potential/ ↩
- For an example of graying out the background, scroll around this page for about 30 seconds: http://kaiserthesage.com/email-list-building/ ↩
- http://danariely.com/2008/05/05/3-main-lessons-of-psychology/ ↩
- http://www.inc.com/articles/2002/10/24718.html ↩