Everyone loves to say, "Let's run an A/B test!"

But it's actually not the right move in many cases—particularly for startups. Make sure you can answer “yes” to these 7 questions before and after running an experiment:

Before

#1. Is your hypothesis testable?

Testable: “If we change our landing page header, our CTR will increase.”

Not testable: “If we change our landing page header, site visitors will like it better.”

That’s subjective and not measurable.

#2. Have stakeholders committed to abiding by the results?

We tend to reject outcomes that contradict our beliefs (called the Semmelweis Reflex, for gross reasons). That can be a problem if an experiment’s results are at odds with HIPPOs: highest-paid people’s opinions.

So ask: Are the HIPP actually willing to change it?

#3. Is the test doable?

Be realistic about the amount of time and resources it’ll take to reach statistical significance. If you don't have high volume, it probably isn't. Use this calculator to find out

#4. How can we make sure results are reliable?

Pay attention to Twyman's Law: Any figure that looks interesting or different is usually wrong.

Be very skeptical of the results, particularly if they're surprising.

After

#5. Are there clear cause and effect?

People shop more when it’s cold in the UK. That doesn’t mean they shop more because it’s cold. It probably has more to do with the holidays.

Correlation is not causation. Drill into the deeper reason.

#6. Have we gotten the most value out of it?

Your results could affect your market, product roadmap, future experiments, and entire growth strategy. One way to increase experiment value is to share your findings with your entire team.

#7. Are our experiments truly driving decision making at our company?

Are you just doing them to checking them off of a to-do list? Or is it actually something that's driving decision making? If not, they’re not worth the resources you’re putting into them.

And as a handy reference, here's a graphic!

‍

If you've never seen it, the Will It Blend? YouTube channel is one of the viral greats. For years they'd use their blenders to rip apart different everyday objects:

But what exactly makes a campaign like this spread and get 293,926,912 views? And for a boring old blender company, Blendtec, no less.

Jonah Berger (professor at the Wharton School) outlines in his book, Contagious, that we share content because of:

1. Social currency. Sharing it makes us look smart, cool, or interesting to our peers. For example: the ridiculousness of a blender tearing up a brand new iPad.
2. Triggers. A stimulus in our environment prompts us to talk about it. For example: a friend busts out their blender to make a smoothie—or mentions their breakfast.
3. Emotion. When we care, we share. Activating emotions like anger, excitement, amusement, and awe drive action more than happiness, sadness, and contentment. For example, the amusement and awe we get watching marbles get pulverized.
4. Public. Can others see that people are using or engaging with the product? It's hard to emulate the behavior of others if we can’t see it. For example, we see the millions of views on the videos.
5. Practical value. We believe it to be legitimately useful to someone—even better if it's niche and we can think of "just the right person this'll help." For example, knowing that this blender is more than enough to grind up your smoothie.
6. Story. What broader narrative can this be wrapped in? Embed ideas and products into stories that people want to tell. Make your message so integral to the narrative that they can’t tell the story without it. Like the blender that's able to rip apart a crowbar.

Contrast this instead to Blendtec making videos that simply talked about the tech specs (5 speeds, 1hp motor, etc) and showed it effortlessly make a mango smoothie.

BORING! Who cares, right?

If you want people to talk about your product, infuse your marketing campaigns with these elements to drastically increase the chance of it going viral (and driving sales).

YouTube is arguably the best social channel to grow a presence on. It's massive, it builds a ton of affinity, you generate ad revenue, and videos can get views for years (unlike a tweet).

If you don't know Ali Abdaal, he has over 4.5M subscribers on YouTube—and generated $4.6M in revenue and $2M in profit in 2022.

One of the main ways he's generated that revenue is by teaching people to grow on YouTube, here are his top 15 lessons after 6 years posting regularly:

We recently ran a market research survey for one of our clients.

We used Survey Monkey to pay for respondents from their network, who were selected based on specific demographic (age, location) and firmographic (job) criteria, given the client's limited existing audience.

Honestly, the answers were pretty bad—most were unusable.

This outcome isn't entirely surprising, given that survey participants are typically paid per completed survey—they're incentivized to race through them as quickly as possible. Same goes for incentivizing people with $25 Amazon gift cards

To address the issue, we decided to do something unusual: We added a skill testing math question at the beginning of the survey. Nothing hard, just simple addition of two 2-digit numbers.

As we expected, this caused a huge drop-off at the start. But the quality of the remaining responses improved dramatically – they were well-thought-out and highly useful. 

The working theory is that the question:

1. Weeded out those only interested in answering as fast as possible, or
2. Forced participants to briefly pause and reflect, setting a more thoughtful tone for the rest of the survey.

Either way, it significantly improved the results of the survey.

If you run incentivized surveys, consider adding some mental friction to the process.

We're naturally attracted to and influenced by people who are similar to us, or a have a shared identity that we value.

For example, if I meet someone online who is from my hometown in Canada, or is from Ireland (because I have an Irish passport), or went through YC, or studied at the same university, or has an aussie shepherd—it immediately makes me feel closer to them.

This phenomenon is called the Unity Principle. Here are four ways to leverage it:

#1. Share personal stories to build affinity‍

The founder of a fitness startup can share a personal story of struggling with their weight.

The personal story creates a sense of shared identity (past struggle with weight). Other people with similar weight control issues can see themselves in the founder's story, feel a sense of unity, and are therefore more likely to trust the exercise program and sign up for it. 

Just make sure the story is true.

#2. Highlight a cause you care about

Patagonia talks a lot about their cultural values like environmental activism, transparency,  and community:

If you value those things too, you're more likely to purchase a Patagonia jacket instead of a North Face jacket because you feel a sense of kinship with the brand.

#3. Loyalty programs or "group name"

Raise up and label your best customers with loyalty programs and special perks. 

Even better if you give members a special name. Like how Lady Gaga fans are called Little Monsters, or how Ship30for30 cohort members are called "shippers." If you give people a label they'll feel more connected to each other—and to you.

#4. Lean into geographical/cultural unity

Ryan Reynolds goes by the handle VancityReynolds on social media. "Vancity" is the nickname for Vancouver. 

I'm confident that due to this, Vancouverites (and Canadians broadly) have a stronger affinity for him because he proudly wears his Vancouver-ness. An easy way to get millions of people on his side without upsetting anyone in the process. 

To learn more about the Unity Principle and more pre-suasive techniques, Katelyn Bourgoin created a great free, email course on how to Pre-sell with Pre-suasion.

Nearly six years ago, before Substack, we had a client at our agency who helped creators offer premium, members-only content on their WordPress sites. 

Their budget was limited, and the total market was relatively small: WordPress creators who monetized their audience with memberships. So ads and content were not a good fit.

We opted to grow their business using cold email.

But as you know, almost all cold emails are terrible and instantly deleted.

You know when they're not? When they:

1. Look like regular emails from people you know
2. Talk about solving a huge pain point you actually have

For #1, we wrote in a casual tone. For #2, we did our homework. 

We paid for scrapers and virtual assistants to go through lists of WordPress sites that sold memberships and label them based on the tool they were using. We did research into the top objections people had about each tool—likely their biggest headaches as creators.

In our emails, we called out those headaches and highlighted how our client's tool would relieve them. 

Another common objection was the headache of migrating. So we offered to do it for free.

Our response rate was nearly 80%. And we booked tons of sales demos.

To make cold emails work, do your homework, and speak to your prospects' biggest headaches that you know they have.

If you’re struggling to prioritize the content you create, you probably don’t have a well-defined content center of gravity. That's the core element that the rest of your content strategy is built around.

Defining your center of gravity (COG) helps inform the direction of your strategy.

Below are a few different COGs, plus companies built around them.

Keywords (SEO): Companies create content around the keywords their audience searches for. That typically includes blogs, white papers, and FAQs.

This is the most traditional COG—which also means a lot of people think of it as the default content strategy. But growth agency CEO Ethan Smith advises holding off on SEO unless you have 1,000+ visits a day from non-SEO sources and 1,000+ referring domains. Otherwise, reaching SEO success is a slow grind. 

Example: HubSpot has written content about nearly every possible keyword related to sales or marketing.

Use-case content: Think customer stories, webinars, templates, and other formats homing in on user pain points.

The goal is to provide solutions to those pain points with your content. This strategy is often informed by speaking to customers and determining common problems.

Example: Besides creating guides and templates, the productivity software company Scribe also runs a Slack group to understand user needs. 

Podcast: Repurposing content is the name of the game with a podcast as your COG.

Using this strategy, you might turn interview transcripts into blog posts or newsletters, and publish short audio/video clips on social media. You leverage smart things said on the pod. 

Example: The media company Testimonial Hero doesn’t rely on full-time writers to create content. It repurposes snippets from its podcast as YouTube videos and LinkedIn ads.

–––

These aren’t the only possible COGs out there. Find a core format that's ideal for your product/industry, and use that to inform how you build and distribute your content.

Runners end races with a sprint finish, getting faster as they approach the finish line. The chance of them quitting plummets.

This is an example of the goal gradient effect. We become more invested in completing a task when we think the end is getting closer.

It’s yet another reason why reducing uncertainty during the checkout process is so important. If a shopper doesn’t know how long it’ll take to complete a purchase, they’ll miss that final sprint-to-the-finish-line burst of investment. 

And be more likely to abandon their shopping cart.

The easiest way to leverage this effect is to display a progress indicator during your signup or checkout process, such as the percentage of steps completed, the number of steps left, or a checklist. That way, shoppers know exactly how long until they can sit back and celebrate their new purchase.

Here are two examples, from Adobe and HelloFresh (I think HelloFresh's specificity is better):

You can also use encouraging copy before and during the signup/checkout process to let users know how close they are to finishing. For instance:

• “Signing up will only take three minutes.”
• “You’re halfway done.”
• “Almost there! Just one more question…”
• “Last step: Where should we send X?”

If you're on Shopify, you can use an app like iCart for this.

There’s a reason Skittles wants us to taste the rainbow instead of seeing it.

Skittles’ famous slogan “Taste the rainbow” is a great example of using a sensory experience to make you want some sugary treats right away.

Imagine instead of encouraging you to taste the rainbow, Skittles said “See the rainbow.” 

You probably wouldn’t be as tempted to impulse-buy a bag.

The reason? In ads, some senses are more likely than others to drive faster conversions. That’s according to a study led by researchers at Brigham Young University and the University of Washington.

• Proximal senses like touch and taste lead to earlier purchases.
• Distal senses like sight and sound lead to more delayed purchases.

In one of the study’s experiments, participants saw fictional restaurant reviews that emphasized either: 1) taste/touch, or 2) sound/vision. The taste/touch participants were significantly more likely to book reservations closer to the present date.

In another experiment, participants read ad copy for a festival that highlighted either taste (“You will taste the amazing flavors…”) or sound (“You will listen to the amazing sounds…”).

People who read the taste version had higher interest in attending the festival that weekend, whereas sound folks were more likely to be interested in going to the festival…next year.

Takeaway: If you want your ads to drive immediate sales, create a touch- or taste-based sensory world in them. Either with words or imagery.

If you've ever used Instagram, you've probably seen people add 30 hashtags to their posts and nothing else for their caption.

A recent study that analyzed 65,000 Reels and 180,000 posts on Instagram has two big findings on what helps content perform better:

#1. Hashtags help.

A Reel with >1 hashtag has on average 30% more reach (and about 18% more engagement) than one that has no hashtags.

The optimal number of hashtags falls between 2 and 4.

So use some, but don't do 30. 

And of course, make them relevant to the content.

#2. Posts benefit from long captions. Reels do not.

I was both surprised and not surprised by the results below.

I wasn't surprised that short captions dominate on Reels. Videos are fairly self-explanatory, and Reels are a quick dopamine hit and move on. 

But I was surprised by the effectiveness of 1,000+ character captions across posts and Reels. That's starting to get to LinkedIn post length—something you don't often see on Instagram.

So go ahead and tell a story and add context along with your posts.

People often talk about sales copywriting frameworks. A very popular one is PAS: Problem → Agitation → Solution.

For example:

• Problem: "Tens of millions of people have been laid off in 2023."
• Agitation: "And the recession still isn't even over—more are coming."
• Solution: "Here's how to make yourself invaluable at your company:"

You make readers think about a problem. You make the problem seem worse. Then offer a solution to that problem now that it's top of mind.

Now that you know the PAS framework, you'll see it being used a lot. 

To extend PAS and apply it to a multi-email sales campaign, Jon Brosio recommends:

• Email 1: Establish a problem (Problem)
• Email 2: Amplify consequences (Agitation)
• Email 3: Share a transformation story (Tease the Solution)
• Email 4: Back up your offer with social proof. (Prove efficacy of the Solution)
• Email 5: Finally reveal the offer and show how it works. (Solution)
• Email 6: Create urgency by limiting the time or added bonuses. 

The next time you're setting up a sales email campaign, try this framework out.

Threads (Instagram's Twitter clone) is officially the fastest growing app of all time—hitting 100M users in just 5 days. For context, Twitter has 250M users total.

But... it's completely unclear if this is going to be the next TikTok or Clubhouse. 

Every creator, founder, and social media manager is asking themselves, "should I be posting on Threads right now??"

The benefits of doing so could be:

1. Land grab. Typically when a new social platform launches, it's easier to grow on it due to reduced competition.
2. New audience. Anecdotally, only my weird tech friends use Twitter. Yet everyone I know uses Instagram. Threads could give access to a new group of people.

But social expert, Jack Appleby, argues that the best strategy right now is actually to do almost nothing. 

Because frankly we have no idea if this could die out in a few months. It's not like Threads offers any unique value over Twitter. In fact, it currently offers less.

So at the moment, the risk-reward ratio tips towards more the risk category of wasting your time and resources. That being said, Jack does recommend doing the following:

1. Create an account, update your bio, and add a link to your site.
2. Do one post as a placeholder so the account isn't blank.

That way if someone searches for you, they find you, and you avoid a negative experience.

But then just wait a few months and see what happens. If it proves itself to have a ton of organic usage and a unique value prop, then great, invest then. 

Hi, it’s Joyce 👋—here to break down another piece of marketing spotted in the wild. This time, direct mail from my local Chevrolet dealership.

This mailer set off my market-y senses for a few reasons:

• The personalized gift card. That’s an actual gift card—and it even has my name. The personalization leverages the endowment effect—when a sense of ownership enhances value. In this case, a gift card with my name makes me value it more than if it were just a generic “save $35” coupon.
  
  
• The QR codes. They help make it easier to take action—no need to type in a phone number a URL yourself, you just scan a code to book an appointment. Going one step further, the dealership includes instructions in big bold font so that the less tech-savvy can still take advantage—a great call since people of all ages own Chevies.
  
  
• Who can redeem this offer. The offer isn’t just for existing customers: “all Chevrolet owners we have yet to assist” are also welcome. It’s the perfect way of encouraging recipients to share the deal with Chevy-owning friends.

All that good stuff aside, the copy is... well, not great. Here’s how I’d clean it up.

Here's how we improved it:

1. Straight to the point. No flowery and meaningless copy.
2. Shorter paragraphs and lists. Making it way easier to scan and digest.
3. Highlighting the most important parts—expiration dates, phone numbers, and offer.

People have short "consideration spans"—particularly for something they randomly got in the mail and are trying to decide whether to huck into the junk mail bin (or keep scrolling). 

Make it as easy as possible for people to get the point.

I still wouldn’t trust AI to generate or write these insights. Or our playbooks. Or our course material. Or our teardowns. (Maybe for outlines and edits, but not core writing.)

BUT, it could be good for more mechanical SEO content and not “thought leader” content. 

A glossary that goes deep on industry/niche specific words can be an SEO treasure trove. If you’ve ever searched for anything related to investing, you’ve probably landed on Investopedia's glossary. It’s massive. Every term is defined in great detail.

According to SEO expert Jake Ward, most brands barely invest in glossaries due to the cost-benefit of paying someone to write them. That means they’re generally easy to outrank.

Assuming $0.05 per word (the low end for writers), 1,200 words per article, and 300 glossary items, it’s a $18k investment minimum ($600 per article). And a ton of time. 

With AI tools you can generate them for about $1k total ($3.33 per article).

It's a fast way to generate a ton content that's underserved and targeted to your niche.‍

BUT here's the problem with this strategy.‍

This is 100% a growth hack. As we shared last week, SEO is fundamentally going to change for top-of-funnel queries once Google bakes in generative AI responses. Glossary items are definitely going to be one of the first to be answered adequately by AI. 

So sure, you can use it quickly scale your SEO today. But it will not last as AI quickly develops and gets integrated into search. It could still be worth it given the low investment cost, but don't expect it to be a long-term growth solution.

To hook people, you need to tap into their emotions.

One of the best ways to do that is to eloquently put a voice to an unvoiced concern, frustration, or opinion.

You want people to say:

1. "Finally, someone said what I've been thinking!"
2. "That’s so damn true."

Finally, someone said it

Many of the comments on the following post were people saying: "Yes, finally somebody said this thing that I've been thinking."

Chris clearly hit an emotional nerve. People are tired of companies simply virtue signaling, rather than fixing a fundamental problem.

Because this post resonated, it got thousands of responses.

That’s so damn true

Here, Andrew tapped into a feeling that every entrepreneur has had:

We all know that dumb luck and great timing play a huge part in success. We just don’t always say it.

So when someone else does, we go, "That’s so damn true."

For eloquent hooks, you need to be able to:

• Identify a concern that people haven't adequately voiced, which means keeping a pulse on trends like shared frustrations
• Give voice to that concern in a way that rings true

If you can nail it, it really works.

You could be quoted for years to come as various accounts share your post. 

We recently launched a free email course on crafting un-ignorable hooks. Join 860 others learning how top creators hook our attention—with tons of examples. This was a sneak peek 👀

Many ecomm founders think of product images as a one-and-done investment.

They schedule professional photo shoots, then use the pics from them for years.

But product photos have a major impact on purchasing decisions, so it's worth experimenting with them. Here are five ideas. 

#1. Use something for size comparison (show your product in context).

Even if your product description includes dimensions, people will understand your product's size better if you show it near something familiar. For example, if you’re selling a picture frame, show it over a couch in the living room.

Don't make people think—uncertainty is a conversion killer.

#2. Test photos without smiling models.

Product images with smiling models are overdone. So test photos in a natural scenario. Consider how hotels show photos of rooms without anybody in them, making it easier for you to picture yourself in those rooms. 

#3. Try different image sizes.

Size can influence our perception of value. In one study, a dress shirt seemed more valuable when shown in a small image, whereas a hard drive's perceived value increased with a bigger image. It’s worth testing different image sizes and the amount of white space in your photos.

#4. Include complementary products.

If you plan to capture your product in a natural scenario, why not include other relevant products to show how they benefit each other? This is a major upsell opportunity—just remember to link those other products shown.

#5. Have fun with it—it's just a test.

Use Midjourney, Photoshop, and other AI and design tools to throw your product into strange scenarios. Like a watch on Honest Abe's wrist.

For that matter, if you can't afford actual product photos for initial or test images, play around with AI. If something does really well, consider staging an actual photo.

Although ChatGPT can already answer a lot of search queries, most people still don't use AI as their default search tool.

That's soon to change.  

Google is rolling out generative AI answers at the top of its regular search results, which could kill the need to click to a website for most queries.

Here are SEO expert Eli Schwartz's recommendations for a new strategic approach to SEO.‍

#1. Focus on mid- and bottom-funnel keywords 

AI will adequately answer basic and broad questions.

Examples: "What is a demand curve?" "How long are Instagram stories?"

Searches like "hotels Vegas Strip" will spit out lists of hotels. On the other hand, "Usher Las Vegas" is further down the funnel—the person searching for that is either looking for reviews of the show or interested in booking it. They're more likely to click on a result, rather than read an AI response.

#2. Focus on revenue from SEO, not volume

Because a lot of "tire kicker" searches will be handled by AI, traffic and clicks will go down. So revenue is a better metric to track SEO success.

#3. Use "People also ask" for content ideas rather than keywords

"People also ask" is a window into the types of queries people are trying to get answers to, and they're generally longer-tail (and more specific). It'll be useful for spotting mid- and bottom-funnel queries that can help generate content ideas.

#4. Prioritize "domain authority" less and relevance and topical authority more

Backlinks from sites with high domain authority were once useful for the most competitive search terms. That might not be the case anymore, since those competitive broad terms will be addressed by AI.

So instead of fighting (or paying) for backlinks from Forbes and TechCrunch, focus on backlinks from less popular sites that are highly relevant and specific to your niche.

Us marketers are nerds. We love judging the ads in our feeds.

Like this Facebook ad from Nood, the makers of hair removal products. Let’s break down what makes it so effective.

This stopped me (Joyce) in my tracks for a few reasons:

First, the creative is a black and white hand-drawn image—a far cry from every other sleekly designed ad on Facebook and a good example of a pattern interrupt. It also made me wonder, "What exactly is going on here?"

That brings us to the copy, which takes a “story first, product second” approach. Take a look:

“Well it happened… our designers went on vacation and now the marketing team is on their own creating ads.”

Nothing about the actual product until the next line.

“But we don’t need fancy graphic design to get our message across: The Flasher 2.0 provides Permanent Hair Reduction in the comfort of your home.”

This is a clever and entertaining way to sell your product, no flashy visuals needed.

And even though the ad copy is long, people actually read it all. You can tell because it ends with Nood inviting readers to draw a better ad for a chance to win a prize. And plenty of people actually shared their drawings in the comments!

Taking full advantage, Nood repurposed one for a retargeting campaign.

Ricky from Trailer Park Boys would say that’s "4 birds getting stoned at once." The ad:

• Sells Nood’s product
• Incentivizes people to engage
• Generates creative for future ad campaigns
• Personalizes the brand to make it more fun

Profitwell has collected over 10 million pricing data points for SaaS products.

One thing they've found is that companies that offer "add-ons" (aka upsell) can boost revenue by 12-22% relative to when they just include them into base tiers. And boost LTV by 18-54% by increasing average order value and retention rates. 

Ironically, this happens even when that feature was previously included for free in a tier.

But how do you choose which features to include and which to make add-ons?

Here Patrick recommends the Rule of 40%:

1. List each feature in a spreadsheet. 
2. Assign a % to each feature based on the % of users who use that feature monthly.
3. If it's used by less than 40% of users, consider turning it into an add-on.
4. Often this could be integrations, analytics, or priority support—supplemental things that aren't core to the experience for most users.

Just make sure you don't do this with a "core" feature that most people use—or else people will likely be upset or just not sign up in the first place.

There are a few reasons why this add-on strategy works;

1. It makes the initial price lower, making them more likely to continue down the funnel.
2. Once they've already said "yes" to the base plan, a quick add-on that increases the price by 10-20% is a quick "sure, why not." It's the equivalent of the grocery store impulse buy candy bar. 
3. It means they get the features they want without needing to upgrade to the higher tier and feeling like they're paying for a bunch of stuff they don't need.

Try this out; it could be a meaningful boost in revenue.

Mutiny, a website personalization tool, doubled their conversion rate on their homepage with a single word change.

Absurd, I know.

Normally we wouldn't even recommend an A/B test so small unless you have tons of traffic.

What did they do? They changed the button text on their homepage hero from "See it on your site" to "See it on your homepage." That's it.

The button links through to their live website demo. You enter your URL, and they let you play around with personalizing your website for different audience segments. 

This demo has two benefits: 

1. It lets you "try before you buy," alleviating a lot of objections or concerns, and
2. Nothing explains your product better than just seeing it in action. 

Now, you'd think the difference between "site" and "homepage" is minimal, but the word "homepage" is a lot more tangible and specific. Even our website has hundreds of pages. I can't clearly picture our "website" in my head. 

However, I do know what our homepage looks like. I know how valuable our homepage is.

And when I do hit the demo, I'm already primed to enter the URL of our homepage, rather than stopping to think which page would be best to test.

Uncertainty is a conversion killer. So remove as much of it as you can.

The lesson here? Words are powerful. Use ones that prime and remove uncertainty.