Marketing Analytics

This post will show you how to use Google Analytics to understand where your traffic is coming from and what your conversion rate for each source of traffic is.

How Google Analytics Buckets Your Traffic#

To make sense of your analytics you must understand that there are 9 default types of traffic.


I prefer to think of these as six main groups:

  1. Organic is when someone finds your website on a search engine.
  2. Paid search is when you pay to show up at the top of search results
  3. Referral traffic is when someone comes to your site from a link on someone else’s website.
  4. Social traffic comes from a link shared on social media platforms.
  5. Email is when someone clicks on a link in an email you’ve sent out.
  6. Display ads, Affiliate marketing, and all other advertising count as paid advertising.

Direct traffic is when someone types your URL directly into the browser. This is more of a catch-all for when Google doesn’t know where the traffic is coming from.

These categories are just Google’s default buckets, you can create your own. Lots of people promote their products via public speaking, networking, trade shows, etc; you can set up custom categories so that traffic from these efforts doesn’t end up in another category.

For example, if I want to do a cold email outreach campaign then I could label this under the default Email category. I could also create an entirely new category called Outreach and have all my cold email traffic show up there.

How To Teach Google Where Your Traffic Is Coming From#

Google does a great job of figuring out where your traffic is coming from but sometimes it’s impossible for Google to know. The solution here is to use UTM links. A UTM link is like a regular website link except that it has a string of information appended to the end of it (UTM stands for Urchin Tracker module, Urchin was the name of the company before Google acquired it).

For example, the link to this article is

A UTM link would be

If you click on the first link it will just show up as any referral traffic in my Analytics, clicking on the second one will tell me that you clicked on the demo example link on my blog. That string of information at the end, after the ? tells Google the source of the traffic and how you want to categorize it.

Creating a UTM link is not complicated, you can just use a UTM link generator. I’ll link to the one I use in the footer. Most people also use link shorteners to mask these long ugly links.

All you have to do is add the original link to the first field in the generator and then enter the relevant information for the source of the traffic. You’re given 5 fields to fill out. I’m not going to explain what all the fields do because you only need to know two to get started: Source and Medium.

The source is where your traffic is coming from. If you parse through the example UTM link above you’ll see that it says utm_source=my-blog-demo-link in the string. This means my-blog-demo-link is the source and that is what will show up in my analytics dashboard when people click on this link.

The medium refers to the 9 default categories we were talking about earlier. In this case, it’s a link in a blog so I’m labeling it referral traffic. It is important to use lowercasing when defining a medium otherwise it won’t work.

Cleaning Up Fractured Traffic#

I like to spend most of my time in Acquisition > All Traffic > Source/Medium.


The first column shows me my top 10 sources of traffic sources and their medium. The rest of the columns then help me understand how that traffic is performing.

This table is a bit messy because there are 3 rows with the same sources. There’s, and This happens because facebook uses different servers for different views. This is unhelpful. What I want is a single row that just tells me how much traffic I’m getting from Facebook.

The way to clean this up is to use a filter.

Go to your test view before you create a new filter. If you don’t have one, set one up before you start adding filters. Filters can be dangerous, once data starts getting filtered out there is no way to retrieve it, so you want to set up a test view where you can test the filter works as expected before applying it to all your analytics data.


In your test view, create a new filter, give it a name and then click on the custom tab and then select the search and replace option.


In the filter field, you want to select campaign source, then in the search string you want to use regex to tell it to find any source that ends with ^.*$, and then in the replace string field you give it the new name you want it to use. I’ve linked to a neat guide on getting started with regex in the links in the footer.

Rather than having three different types of Facebook traffic in my sources panel, now it will all just show up under a single Facebook label. You can repeat these steps for any kind of fractured traffic by just replacing the domain name in the example above.

The next thing I did was create a UTM parameter for the links I share in Facebook groups so that I can distinguish it from when I share stuff with my friends on Facebook. This is why some Facebook traffic has started showing up as ‘facebook groups’ in the screenshot above.

The last thing we need to clean this up is a referral exclusion list. You can do this by going to the property settings. I’ve highlighted the button in red on the screenshot below.

referral exclusions.png

Referral exclusion means removing your own domains from your traffic. I have sources like joshpitzalis and learninglog showing up in my traffic. These are both my own sites. To prevent this from continuing to happen I’ve added both of these domains to the referral exclusion list.

Unfortunately, none of these changes will apply in retrospect so it’s important to make these changes as soon as you can. Now that the measures are in place, cleaner traffic will slowly drown out the messiness.

Now we know how much traffic we’re getting, where it’s coming from, and how it gets broken down. The next piece of the puzzle is to figure out which source of traffic is having the biggest impact on your business.

Tracking How Well Your Traffic Converts#

Conversion is just a measurement of how many people do something. If 1000 people visit your website and 20 of them sign up to your mailing list then your mailing list’s conversion rate is 2% (20/1000 X 100).

Before you can track any kind of conversion you have to figure out what it is you want people to do. In the example above we want people to sign up for our mailing list. For most Saas businesses the goal of the landing page is to get people to sign up.

You’re allowed to track 20 goals in google analytics. To keep things simple I recommend just starting with your single most important goal.

To set up a goal in Analytics you must go to the admin section, and then click on the goals in the view column, and then click on the red ‘create goal’ button. This will take you to the following page:


If you want to track people filling out a registration form then you will need to set up an event goal so that you can track the registration form submission event. You will need to go into the form in your landing page source code and trigger the event on submit. Depending on how you have analytics setup, you will need to do this with Google Tag Manager or the Google Analytics event trigger (or however else your project tracks events).

event goal.png

When you fire the event you will need to define an event category and an event action. The category is the broad grouping of the event, so ‘sign-up’ or ‘registration’ or something like that. The action is the specific behavior you are tracking, for example ‘form submission’. Labels and values are optional. If you have multiple signup forms you can use a label to add info about which form fired the event. If the signup is paid you can also allocate a dollar value to the action. These last two are optional.

That was a bit complicated but you must set up a goal in order to track your conversion rate. If event goals sound too complicated, then a simpler alternative is to set up a destination goal. This tracks when people land on a specific page. To do this you must create a page that people can only reach after completing your goal. You simply add the URL of the destination page to the goal and you are done. This is often why people direct you to a thank you page when you download something online. You could also have a welcome page that people only visit once after they sign up. However, you must make sure people don’t see the welcome page every time they log in otherwise it will skew your metrics. This is why, for signups, event goals make more sense. There are two other types of goals but those are more suited for blogs and content-focused websites so I won’t go into those.

Once you have set up your goal, the source/medium page from before will have a whole new section appended to the end of it called Conversion. In this case, my goal for my website is to get people to book in an initial conversation.


This page now has three groups of columns: acquisition, behavior, and conversion ( I have outlined the conversion column in orange above).

The acquisition column tells you how much traffic is coming from each of your sources. This is what almost everyone uses Analytics for.

The behavior column tells you how engaged a source of traffic is. There are three sub-columns, you have the bounce rate (the percentage of people that view a page once, do nothing else, and then leave), pages per session, and the average session duration. These three columns help me assess the quality of my traffic.

If I just looked at the acquisition column (and tally up all my fractured facebook sources) I would see that I am getting most of my traffic from Facebook. However, if you look at the behavior column you can see that my Facebook traffic has a much higher bounce rate than anything else. It also shows me that people who come from Twitter spend a lot longer on the site and view more pages per visit. So thanks to the behavior columns I can tell that Facebook might be giving me more traffic, but Twitter is clearly giving me better traffic.

The final column is Conversion, this shows you the result of all this traffic and engagement. In the screenshot above we can see that one person booked a consultation with me (this is the goal I have set up on my website). Unfortunately, this person came from the direct medium so analytics can’t tell me how this person found my website. Had I started using UTM links sooner I’d see what source of traffic this conversion came from. As more conversions start coming in with cleaner traffic data I will be able to tell where my highest converting traffic is coming from.

There are lots of other things you can do with the conversion section. You can set up multiple goals, you can have multi-channel goals, you can even set up neat visualizations to see where people drop out of your sales funnel but I don’t recommend diving into all that until you need to. Trying to learn analytics in its entirety is a waste of time. A much better approach is to ask an important question, and then just learn enough to answer that question. A good baseline is to understand how much traffic you get, where it is coming from, and how well it is converting. Now you know how to answer all three of those questions.

Product Messaging

Let’s start with our landing page. Instead of judging a page on opinion and preference let’s based the teardown on proven persuasion principles. I’m going to use these principles like a gap analysis tool. These are critical components to building a persuasive argument. All I have to do is go over the copy and ask whether or not I’m including those elements on not.

A Teardown In 20 Questions#

What does the product do?#

Why should I care?#

How do I believe you?#

How do I get started?#

For each question, I assign a score of zero if the answer is no, 2 if the answer is yes, and 1 if the answer is somewhat, but could be improved. Then I tally up my score for each section so that I can prioritize which aspect of my messaging needs attention first.

This is Chirr App’s existing pricing page. I ran through each section and got a total score of 13 points. So the page is about 40% as effective as it could be.


I gave the What does the product do section a score of 0, the Why should I care section got 5 points, and the other two sections got 4 points each.

There’s a lot of leeway in how you interpret each of the points, and there could be hundreds of different points in each section. You can spend days getting lost in minutiae here, but I limited this exercise to my top 4 four things in each category so I could identify what’s broken as quickly as possible and then fix it to see if you’re on the right track.

Feel free to use these 16 questions as a starting point and then modify it to use whatever framework or set of principles you’re comfortable with. What’s important is that you have a clear, shared, repeatable framework for assessing your messaging that everyone on your team agrees on.

Messaging Hierarchy#

People of the internet have been building landing pages for a while now and have established a sequence to the sections on a landing page that works.

Screenshot 2021-09-12 at 3.12.21 PM.png

The Hero Section#

A Helpful Headline – Imagine a five-year-old finds your product and asks you what it is. Your response should help the child understand what your product does and who it’s for. Prioritize clarity and brevity.
People don’t read much online. Typically, they just want information quickly. They look at stuff that is either new, unusual or helpful. New gets old. Unusual can be good if it works. Helpful is a solid bet. Use your headline to tell people how your product helps them.
Defining who your product is for in your headline is also a good way to filter out people you cannot help. You only want interested people to continue reading. Avoid getting people to read stuff only to discover that it doesn’t apply to them at the end.

A Supporting Byline – Our hypothetical five-year-old understands what your product does, now explain how it does it in ten words or less.

** A Preview** – This could be an image, a screenshot, a video, a demo. The idea is to reinforce what you are talking about with a clear visual depiction of the thing so that there are no misunderstandings.

Your call-to-action – There is a whole section dedicated to your CTA below but you always want to include one above the fold to let the people who don’t need persuasion proceed right away.

Features & Benefits#

Imagine the five-year-old challenges you and asks you why someone would need your product. You must highlight the core problems it solves.

Our brains are fine-tuned to detect problems. A giant piece of cake on the sidewalk might get your attention but a tiger will stop you dead in your tracks.

When you have a specific problem, a specific group of people and a solution you can construct a trigger.

People: 5-year-old children
Problem: Being scared of the dark
Solution: A bedside lamp that projects a faint night sky onto their ceiling.
Trigger: We help kids enjoy going to bed by letting them explore the universe on their ceiling before they fall asleep.

When deployed well, a trigger will prompt someone to ask “What do you mean by that?” which actually means, “Tell me more”.

This idea of a trigger is taken from a book called The Brain Audit by Sean D’Souza. He says that the ‘kiss of death’ is when someone says ‘oh! that’s interesting’. What they are actually saying is ‘No thank you, I want to escape this awkward interaction now’.

If our five-year-old thinks your product is “interesting” then it’s game over. You’ve lost them. You need to start from the top and rework your trigger.
On the other hand, if her response is akin to “What do you mean by that?”, then it’s time to explain what your product can do for her.

Another way to communicate the benefits of your product is to list out everything people resort to when they don’t have your product. Then outline all the problems with each of these alternatives. Finally, explain how and why your product is better. That’s what your product can do for me. Now pick your top three.

If you are still left with a bunch of features that describe your product and not its benefits, another solution is to add “which means that…” to the end of your feature sentences. For example: “Our products are only made with organic ingredients which means that they are good for you and they taste delicious”.


Imagine the child’s parents just showed up. They are skeptical and want to know what’s going on here. You need to back up your claims and disarm the most common objections with facts and specific data.

Objections are good by the way. They are an indicator of interest.

Disinterested customers won’t object, they won’t ask questions, they just walk away. When someone engages with your product, that’s when they start asking questions. Objections mean engagement.

If the objection is valid and you can’t address it then they are not the right person for your product. Work on eliminating them well before they get to this stage. Be clearer about who your product is for in your headline.
Brainstorm all the possible objections to your thing and then address them one by one. If you can get someone else to address the objections for you, even better.

Testimonials – There always come sugar-coated. People can taste sugar. A good testimonial starts with skepticism. They describe the fear and uncertainty going through people’s heads when they first considered your product.

A reverse testimonial works because it speaks to us, in the way we speak to each other. When we’re recommending a restaurant, we intrinsically lace our recommendations with doubt.

The five questions you need to ask to get a powerful testimonial are:

  1. What was the obstacle that would have prevented you from buying this product initially?
  2. What happened as a result of buying this?
  3. What did you like most about the product?
  4. Would you recommend it? If so, why?
  5. Is there anything you’d like to add?

Once you link each objection to a testimonial, you can bring it home with a guarantee.

A Guarantee – If you’re getting lots of complaints, it means your product isn’t doing what it’s supposed to. Complaints are valuable feedback. Listen to them so that you can fix the problem.
Generally, someone who complains wants you to improve. People who don’t care won’t complain but will leave anyway.
A guarantee lets people know that they can complain and something will happen. Refunds are an early warning system. Use them.
That’s it.

The Call To Action#

Now the child is excited. Now they want to get involved. What is the next step?

Your call to action button text should start with a verb and describe what will happen next ( Start trial, See pricing, Join waiting list). The button should be obvious twenty steps away from the screen.

Putting it all together#

Rather than trying to fix everything at the same time, the teardown at the beginning of this post let me identify that the What does the product do aspect of the current pricing page was the bit that needed the most attention.


Our headline doesn’t really explain what the product does, nor does the byline outline how it does it and there wasn’t a preview or screenshot of the product anywhere on the page.

Without paying too much attention to all the other aspects of the teardown I just focused on fixing these three things first.

new (1).png

I went through the audit again on the new page and this time I scored 18 points. I added 6 points to the What does the product do (and I lost a point because we n longer have CTA above the fold).

This version of the landing page is currently being AB tested and I want to make sure that these changes have an impact on the conversion rate before I spend more time fixing all the other details.

How to run a good survey and get valuable insights about how people use your product


Surveys are a powerful tool for building up a clear picture of your audience but they are often used in the wrong way. I want to show you how to avoid the common mistakes of running an in-product survey.

Have a clear idea of what you want to know going in#

I often come across people that run surveys because they want to “know more about their users”. They just ask a bunch of different questions. Before you know it, you end up with a big, bloated survey. This is terrible because nobody is going to complete a survey that long, and if they do, you just end up with a lot of data that you don’t know what to do with.

If you have a clear idea of what you want to know then you should probably just be asking one question. Answering that question should allow you to do something specific. Don’t just have a question that you want to answer out of curiosity, ask a question that lets you achieve something specific with the answer.

If you don’t have a specific question in mind, then you probably shouldn’t be running a survey. It’s not the right time to do it. Customer interviews are a much better fit for more exploratory qualitative research. On the other hand, if you already know the answer and you’re just looking for confirmation then experiments and AB testing be a better approach.

A good example of a one-question survey is to ask people is “if you decided not to try our product today, which of the following reasons stopped them?” And then provide a list of reasons, and at the bottom, I put an option for ‘Other’.

It’s a good survey question because the answer will help inform a clear and specific. The other reason a survey is a good fit here is that the reason could’ve been any number of things. It could be price, it could be that they didn’t know what the product does, maybe they just don’t know how to use it. You’re not looking for reasons, you have a pretty good understanding of why people don’t try the product, you just need help understanding which option to focus on.

Beyond getting people to identify basic segmentation data about themselves, don’t focus on who your audience is. Instead, focus on what they want to do. What tasks do they want to complete? What questions do they want to answer? What are their goals? What are their pain points? What are the objections that stop them from acting?

So, how can we encourage more people to complete a survey?

Pick your moment to ask#

You arrive on a new website for the first time and immediately a survey pops up and asks you to answer a few questions.

No thank you.

People typically come to a product with a task they want to complete. Let people complete the task they came to do and then ask them to complete a survey. At the very least give them a few moments to get settled in, maybe let them visit one or two pages first.

The best place for a survey is on exit intent. It’s at that moment when they’re about to leave the website. This might not make sense depending on the content of the survey, but you certainly shouldn’t jump people the minute they arrive.

Keep it short.#

One question is usually enough.

If you don’t have a single question, make sure you keep it short and make sure it’s focused on a single subject. The more different, random questions you throw in, the more people are likely just to give up.

If you do have more than one question then always start with the easiest one to answer. One of the first things people will ask you on a telesales survey is what your name is. The reason they ask that is that it’s so easy to answer. If you can get them to answer one question, the chance of them answering the second one goes up significantly.

Use closed Questions#

They tend to have a higher response rate than open questions. People don’t want to have to think too much when completing surveys. Offer a set of predefined options for people to select. If you’re planning lots of open questions, maybe you should be conducting interviews rather than running a survey.

Explain why you’re asking.#

Why do you want to know this information? What are you going to do with it? If you’re honest about this and you tell people they will be more likely to help you out.

Don’t ask for personal information.#

People sometimes use surveys as an opportunity to start collecting email addresses. If you want people to complete the survey, you shouldn’t be asking for stuff like that. It’s just going to put them off.


Incentives are one of those things I often struggle with. Offering an incentive biases the type of people that will complete it. That said, not offering an incentive also introduces bias. As I said, it’s something I struggle with.

An interesting approach you can use with incentives is to only tell them about the gift once they’ve completed the survey, rather than saying upfront that you’re going to give them a gift for completing the survey.

We’re giving people an incentive to motivate them to complete the survey. If you don’t tell them about it, how are they going to know, and therefore, how will it motivate them to complete the survey? I know this sounds back-to-front but bear with me for a moment.

By offering them a gift at the end to thank them for completing the survey, you trigger a reciprocal response because it’s an unconditional gift. They weren’t expecting it. They didn’t factor it into our decision to complete the survey.

Because it was a surprise, and a delightful one, in those kinds of situations, people sometimes feel the need to give back. What some people will often do in this situation is tell other people about it. “Oh, I just did a survey and they gave me this great gift at the end. Isn’t that nice?” Because that’s their way of saying thank you.

There is also the idea of it being a little insider trick that you can give to other people. “Oh, did you know if you complete this survey, you get this at the end of it?” So, sometimes not talking about the incentive is an effective way of getting more people to spread the news about it and encourage other people to complete it too. At the very least, it’s worth trying out.

Surveys are an invaluable tool for understanding how people use your product, but they’re often a wasted opportunity. To avoid the most common mistakes, be very clear about what you want to learn from your survey and focus exclusively on that. Secondly, pick the right moment to ask people to complete your survey. Third, explain why you’re asking, and if possible, offer some kind of incentive to encourage people to act. Most importantly, keep that survey short and focused, and easy to complete. In my opinion, when it comes to user research, often, one question is more than enough to point you in the right direction.

Avoid wasting time testing inconsequential stuff when growing a product

Artwork by Justine

Traditionally, marketing focuses on acquisition as the main growth lever. The problem is if you don’t activate new customers, they’re unlikely to retain long-term. In SaaS businesses especially, if you don’t retain customers then you’re really just replacing a customer each month. You might be able to do it effectively for a few months. But over time, it’s an unsustainable way to grow a business.

Begin with a north star

The best way to start thinking about this process is to begin with your core action, this is the thing people do that drives value for them. This is sometimes called a north star metric. The north star metric is one of the breakthroughs that came out of Facebook and now all the effective growth companies have latched onto the concept.

All that matters is that your north star metric reflects the value that is delivered to users. You’re trying to track the aggregation of customer value over time. The idea is that customer value is what drives retention and retention is what drives sustainable revenue.

A couple of examples here would be Airbnb’s north star is ‘nights booked’. So whether you’re a host or a guest, when a night gets booked, everyone is getting what they came to the platform for. The more nights booked over time, the more value is being created in the system, and that value is what drives sustainable growth.

Then find high leverage opportunities

Once you’ve narrowed in on a north start and began tracking your core action, the next step is to understand the relationships between the variables that move your north star. The only way to do this is to test what actually drives growth.

There are two different types of testing. The first is testing to discover and the second. is testing to optimize. The best way to think about this is like a game of battleship.

If you’ve never played battleships then here’s a 2-minute walkthrough of how the game works:

The point is that when if you get the coordinates wrong, you still learn something. Over time, you start to get an idea of where those opportunities are. This kind of exploratory ping is a discovery test. When you do hit upon a growth opportunity, it’s safe to assume there’s a better way to do things. The only way to find a better way is to start testing variations. These kind of tests are optimization tests.

I first heard this Battleship example in a talk by Sean Ellis. He claimed to have stolen it from Brian Balfour so I think it’s safe to steal once again because it’s such a good analogy.

The general rule here is that the more testing you do, the faster you tend to grow. If you test 10 things, the chances of finding a growth opportunity are higher than if you take one or two shots. However, that’s not an excuse to test pointless details. Tiny changes like the shade of a button are a waste of time because you could be using your limited resources to test more consequential stuff.

The way that you find leverage is that you look at the data, and study where you’re losing people. For example, if in a funnel you find that you have a 90% drop-off somewhere, running tests around how to fix that are going to be more consequential to growth than testing different colors of your sign-up button.

Then use qualitative data to brainstorm why

Quantitative data helps you narrow in on high leverage opportunities. Once you’ve found a focus the next step is to figure out what the problem is. Qualitative data is much more useful here. Dress it up however you want but fixing product problems is a guessing game. Surveys, session recordings, competitive research, industry best practices, and (my favorite) customer interviews make it an educated guessing game.

If you want to rely on intuition alone then you look at the problem and rely on common sense to figure out what the most likely cause of the problem is. A more effective approach is to get as many relevant people in the room as you can and leverage as many different viewpoints as possible. You want diversity, so rather than getting the whole product team into a room, aim for one person from sales, one from customer support, an analyst, a developer, and someone from the customer research team.

These kinds of group brainstorming sessions can fall apart easily. It’s important to make sure that everyone understands the opportunity you are focused on. This is not a general-product-feedback session.

Typically, people like to start by throwing solutions on the board. This makes it really hard to compare and prioritize different solutions. An easier approach is for everyone to come up with ideas around why the problem exists first.

Once you have a shortlist of the most likely problems then you want to come up with the simplest solution to test if the problem exists. Don’t spend 3 months building a possible solution to something that might be a problem. Spend a month building 10 simple solutions to the most likely problems. Then spend the remaining 2 months building better solutions to the problems that had a measurable impact on your north star metric.

Identify and focus on the highest leverage opportunity you have. Then run as many discovery tests as your resources allow using the best guesses you can source.

As soon as you hit a battleship, switch to optimization tests and figure out the best way to execute that solution. Measure success in terms of improvements to your north star metric. That’s why it’s so important to make sure this number corresponds to actual user value when you begin.

After a while, you’ll start hitting a ceiling on the optimizations you can make. This is when it’s time to switch focus and repeat the process with the next high leverage opportunity.

Testing takes a lot of expertise to set up properly and a lot of time to run effectively. This is why it’s important to avoid wasting time testing inconsequential stuff when growing a product. You want to focus on high leverage opportunities and systematically test your best assumptions before you narrow in on optimization tests. The more tests you run, the more you will learn, and the more likely you are to drive a successful growth outcome.

Email marketing laws

Good email practice boils down to setting clear expectations. If someone is on your list they should know exactly what they’re going to receive and why, who it’s coming from, and how often to expect it.

If you send someone an email, and they don’t know why they got it, they can hit the spam button. If they do Gmail, Outlook and Yahoo pay attention. They will listen to a user over a marketer any day of the week.

Legality and permissions aside, we’re talking about deliverability here. Just because you send an email does not mean that it will land in someone’s inbox. Deliverability is not the same as a bounce rate. If Gmail thinks your email is spam, they might just not deliver it. They don’t report back to you either, there’s no bounce back, they just don’t deliver it.

According to Jessica Best (I’ll link in her amazing course on Email Marketing at the end), in the US, what we see on average deliverability is about 85%. About 5% or 6% go to a spam folder, and the rest of it just goes missing entirely. So whether or not the following laws apply to your it is in your interest to follow them

Let’s say you’re setting up a new newsletter for your SaaS business and you fully intend to follow the law, here’s what you have to do to make sure everything is above board:

  1. You must accurately identify yourself and who the email is from.
  2. You must have a functional opt-out mechanism.
  3. You must add a physical address to your emails. The reason being that people need to be able to opt out without clicking on a link. If you get dangerous-looking spam and you don’t want to click on any links you always have the option to write a letter to opt-out of any email marketing.
  4. Your message and subject line must not be deceptive in any way. This means that the subject line must accurately reflect the content of the message. If your email is an advertisement you must disclose this clearly and it all has to be in plain language. If you promise something in the email and don’t deliver that is just fraud. For example, if you say you’ll compensate or reward people for taking some action and have no plan to follow through. There is zero grey area here, this is straight-up illegal.

These are all US rules based on the CAN-SPAM act. The last one is actually from the Federal Trade Commission. I will include links to all of these sites at the end of the post.

The US has some of the most relaxed laws when it comes to email marketing. The law is literally called CAN-SPAM. You can spam anybody you want until they tell you to stop.

In the US, the only people who can take action are inbox providers like Gmail. They take action on mass on behalf of the folks that use their service for their email. On the other hand, Canadian anti-spam legislation allows individual citizens to take action.

With CASL you must explicit permission to send an email to someone on your list. No pre-check checkboxes and no automatic opt-ins. You have to have affirmative consent to email somebody.

The kicker is that anyone in Canada can invoke CASL. Not just Canadian citizens. Anybody who’s in Canada at the time of receiving your email. If someone reads your email in Canada, technically, you are required to follow the laws of CASL.

Similarly, GDPR is for people in the European Union. Again, this is about where your readers are, not where your company is. If your company sends emails to people in Europe you are subject to GDPR law.

If you want to comply with all of these laws the trick is to explain how someone’s email is going to be used when they signup. This means everything should be clear right before someone presses the submit button where they give you their email address.

You have to use plain, concise, intelligible, transparent language that is easily accessible. No sneaky jargon and you can’t obscure important details on a hidden privacy policy page.

The scary bit for businesses is that under GDPR law an individual subscriber can request all of the data that you have on them and ask to be forgotten. This can be
tough because you don’t always know where all your consumer data is. If somebody asks to be forgotten, you must be able to comply in 30 days.

Fortunately requesting to be forgotten is not the same as unsubscribing. Unsubscribing just means they don’t want to receive marketing emails. You can still keep someone’s details on your database after they unsubscribe, provided they are marked as ‘do not mail/email/market’. In most cases, you need to keep someone’s details so that you don’t accidentally add them back to your through another channel.

The last thing I think I’ll cover in this primer is that you can’t deny people goods or services or charge them different prices, based on the profile data you collect. For example, if you live in a higher income zip code, you can’t charge people more based on the zip code data. Or if people live in other zip codes, you can’t deny a sale just because they live in lower-income zip codes.

I think I’ve now covered the most important fundamentals here. I’m still learning about this stuff so if I’ve got any of it wrong please let me know.

If you’re interested in learning more, here are all of my primary legal resources around permissions and data:

Landing page optimization

When I begin working on improving conversion for a landing page the starting point is always the value proposition.

A company called Marketing Experiments put together a heuristic for calculating the probability of conversion:

Conversion = 4 motivation + 3 value prop + 2 (incentive - friction) - 2 anxiety

The single biggest factor here is motivation. I can’t control why people want my product but you can understand it. Clearly understanding what brings someone to my products makes it easier to craft a value proposition that they can relate to. If I can craft an effective value proposition that speaks to people’s core motivation around my product then that’s more than half of the battle won.

If I were starting a restaurant I’d pick a hungry crowd over a fancy chef any day. I like to spend a disproportionate amount of time trying to figure out who my crowd is and what they are hungry for. I’m talking about researching competitors, listening for relevant conversations in related forums and communities, understanding what the customer support team already knows, conducting surveys, and most importantly, setting up customer interviews.

I’ve managed to conduct 15 customer interviews in the two months since joining Chirr App and we’ve surveyed exactly 199 people to date. If you’re one of the people I interviewed or if you filled out our survey, I am grateful for your time and feedback. I’m also now handling all of the customer support, which is basically free customer research.

This is painstakingly slow work, but it’s worth it. If you get it right it’s all downhill from here. If you get it wrong, nothing else really matters.

There are loads of ways to establish your value proposition. I like to take a volume-based approach. I list out every single problem I can think of or find a reference to, around the problem my product solves. Then whittle it down to the best ones.

By the end of it, I’ll either have a few different value propositions that I want to try out or I’m working on optimizing an established value proposition.

Either way, once I’ve narrowed in on a value proposition, the first step is always the headline. I work on figuring out the words first. Once the messaging is down for the whole page then I work on the design.

Start with the headline#

The headline is arguable the most important part of a landing page.
If someone read nothing else, would they understand what my product does?

I like to pretend the person I’m designing for has zero context. Let’s say they walk into the Apple store and this is the default page on a computer screen. I also like to assume they are not my target audience. I’m not trying to get them to buy anything. The litmus test is whether or not they understand the product clearly enough to recommend it to the right person when they’re having drinks with friends later in the day.

For example: “Our products are only made with organic ingredients which means that they are good for you and they taste delicious”

I don’t formulaically include “which means that…” in my headlines, it’s just an easy way to start thinking in the right direction.

A mistake I’ve made with headlines in the past is to try and get fancy, now I just aim to be obvious. People don’t read much online. Typically, they just want information quickly.

On the Chirr app pricing page at the moment our headline is “Save your time and nerves”.

Screenshot 2021-08-13 at 8.12.08 AM.png

This makes sense because our pricing page isn’t really a landing page. Most people who get to our pricing page have already used the product.

However, I would like to drive some ad traffic to our landing page in the future. This means we have to make sure the headline makes sense with zero pre-existing context.

The current headline does not pass the litmus test of people being to understand what it does if they read nothing else, nor could they recommend it to the right person in casual conversation.

  1. To improve things, a feature I’m going to focus on to capture the product’s purpose is that it lets you turn long blog posts into Twitter threads.
  2. Which means that you can create Twitter threads easily and quickly.
  3. In less than 10 words I’m going to go with “Make Twitter Threads
    quickly and easily.”
Screenshot 2021-08-13 at 8.13.53 AM.png

It’s not sexy, but it’s obvious. I’ll probably tweak this some more before any of these changes go live but that’s my thinking so far.

The subheadline#

If a good headline tells your audience what you do and why it makes their life better. Your subheader explains how you do it, again in 10 words or less and with zero fluff or jargon.

Feature sections#

I like to use my feature sections to handle common objections to using the product.

The idea is to communicate the core benefits of the product, explain the biggest problem our product solves, and address your main concerns and objections all in one go.

I aim for 3-5 features. The more expensive or unintuitive the product, the more objections I need to address.

A good feature description is made of three parts: a title, a description, and an image.

A title lets people decide if something is relevant#

A title describing your feature or its value in 5 words or less

I’m just trying to describe what it is so that people can decide if it’s relevant to them.

I have a group features, benefits, and objections that I want to focus on:

For the first value prop, I came up with as many headlines as I can think of

Then I try and trim the best ones down to 5 words or less

I’m going to go with ‘Preview threads as you write’ because I think that will make the most specific to anyone who has tried to create a long thread on Twitter and is then searching for a solution.

Then I repeated the same process of laying out options for the other two value props and then trimming then and picking the best one.

The description twists the knife#

If the title lets people skip over bits they don’t care about, then I’m going to assume anyone reading this description stopped here because they can relate to the value prop in the title.

Now I want to speak directly to the person (not in the third person) and go straight to their problem. If it goes on for more than 2 sentences I resort to bullet points. I try and end each line with a clear benefit.

I’m trying to point out what people are doing wrong and how I make things better. The ideal outcome is to reveal the larger implication of the issue and make them viscerally feel the pain of their problem in the headline.

Preview threads as you write
Making threads in Twitter means copy-and-pasting stuff into a notebook, counting characters to figure out where the breaks go, and moving images around as you edit 🤯 Start sharing great threads on a regular basis without all the hassle.
✨ Know exactly how many characters fit into each tweet.
✨ Easily split text into separate tweets.
✨ Add as many images, videos, and gifs as you want.

Then demonstrate the feature in action#

Your feature’s image should visualize the feature in a way that puts everything into context. Even better, use an auto-playing GIF or SVG animation.

Here is everything put together.


This is all work in progress, I’ll continue to write about this as I improve our landing page, then I’ll AB test the changes and share them.

User Retention Audit

Retention is a measure of how often someone comes back and uses your app after they first sign up.

Here is a quick framework to measure and improve user retention. I’ve distilled everything I know about improving retention into 9 YES/NO questions.


Customer discovery in the context of existing products

Customer discovery is the process of speaking to people to understand their problems so that you can uncover opportunities to improve your product.


Writing Design

Most of what you see on a screen are words. Words make up the content, the navigation, it’s on the buttons, in the headings, it’s everywhere.

Everyone knows what the product is supposed to say. Yet, at the last minute, someone on the team is tasked with filling in the final wording. If you have ended up being that person, this article is for you.


Getting Press

For smaller companies that don’t always have the budget for advertising or building their own content team, focusing on public relations offers the biggest bang for your buck.

Even when teams do have the resources to produce original content, PR provides a type of third-party validation that no other form of marketing can.

Advertising is only becoming more and more expensive. Native advertising and editorials do work but the savvier the customer is, the less effective they are. People are just getting better at distinguishing between an ad and a real story.

For larger teams that can produce original content scale and place advertisements, PR complements the overall marketing effort by validating your message. Whether it’s an interview or an opinion piece or a guest post on a popular blog, having other people talk about your product always hold more weight and credibility than investing in ads or promoting yourself.

What PR Is and Isn’t#

One of the biggest misconceptions is that PR is about advertising. People think you create ads and work with the media outlets. That’s advertising. PR is earned media.

Earned media is working with reporters to pitch stories so that they either write about you or interview you. That interview on TV, a PR person probably pitched that and set the whole thing up. The articles you read in newspapers are the result of PR efforts, not advertising.

Another misconception is that PR people can guarantee results. You can always guarantee that you’ll put in the work, but you can’t control when the results show up or how big they’ll be. If anyone guarantees that stuff, then you probably shouldn’t hire them as your PR person.

A good PR effort is about coming in and figuring out your messaging, understanding who your target audiences are, working out what the right messages for each audience are, and putting the two together.

There are no special rules for talking to the press and getting great stories out there. There’s no fiddling around with email subject lines or blasting lists of reporters with story ideas. PR is about developing relationships with people and creating a fair exchange of value.

Define Your Audience#

When it comes to your public messaging, your message needs to be laser-focused on a target audience. Robust messaging is always tightly focused on a specific segment. Unless you have a product like Coca-Cola where everybody is your potential audience, speaking in very broad terms is not a good sign.

Once you have figured out who your target audience is then you can work backwards and ask:

Your PR effort is about making connections and figuring out how to get placements in those areas so that you can get your messages out through those mediums and reach that audience.

Generate News#

The most important thing is to generate news. News is timely. It’s the stuff that makes more sense to talk about today than tomorrow. It’s what reporters want. Product launches, new features, product milestones, huge sales, significant deals or customers, and fundraising achievements all count.

Sometimes things will resonate, other times they don’t. When a story doesn’t get traction, you move. This happens a lot in marketing, and it certainly happens in PR. If something isn’t working and the messaging doesn’t resonate, you get to work on the next story.

Rely On Warm Introductions#

When pitching reporters, always do your research. Who are you pitching? What have they been writing about? How have they been covering it? Follow them on Twitter, comment on their stories, focus on building a relationship with them first.

Reporters want to talk to CEOs and co-founders–not PR people. As an early-stage founder, you should be building relationships with reporters and making pitches.

The best way to meet a reporter is through a warm introduction. If a reporter you want to meet has written about someone you know, have them introduce you.

In your intro, you should be able to describe your news in one sentence. Let the reporter know you’re offering them an exclusive. Always offer your story as an exclusive because your audience usually all exists in just a few outlets. The goal is a twenty-minute phone call.

When you get the reporter on the phone, be ready with three to five clear reasons why your news and company is important. At the end of your call, offer your notes and helpful collateral like screenshots, graphs, logos, pictures, etc.

Measuring The Impact#

Measuring results with PR is complicated. If I get you on TV a few times, people start to see you, they start buying your stuff and that’s good PR. But you can’t say that because you were on TV three times that translates into $100K worth of sales.

There is a correlation, but you can’t always trace it. With the trend towards digital, there are more analytics to work with. You can track mentions, you can track sentiment, you can see where people are coming from and how many people click through to your website.

However, “We got a story in the New York Times, and one in the L.A. Times,” does magic for your image even if there are no backlinks involved. It’s the quality of the mention, the quality of the article that matters, and that stuff is harder to quantify.

A front-page story might cost tens of thousands of dollars if you’re advertising but it doesn’t cost a PR agency anything. That’s not to say it’s free. It’s just that the value is in the relationships and understanding how and when to pitch stories.

The benefits of PR are hard to measure but they do exist. To see real PR results you’re better off getting one story every month versus one huge story every year. Don’t combine all of your news into one story, the idea is to spread it out over time.

After about six to twelve months of sharing real news, you should be relatively friendly with two to five reporters. In other words, they’ve written about your company enough that you’ve probably met them in person and certainly they’ll reply to most of your emails.