Forget conversions... Until you have traffic.

[Brad Gosse]

Just a small observation.

Many people I talk to spend more time tweaking their sites for conversions than they do generating traffic. When I ask why they say "I want to make sure the traffic converts".

This is the equivalant in the offline world of locking your doors to re arrange your store for better traffic flow.

In my experience, the only thing you should worry about is traffic at first, even if your site design is ugly. When traffic is flowing in you can make minor changes and see the results quickly.

[Vaan]

Nice thread,

All warrior here is know that traffic is very important,
The problems of all marketer is how to drive it? massively..

[TLTheLiberator]

As far as I'm concerned Brad is right!

Become a master of traffic generation and it's gana be hard not to make good money.

After you conquer traffic generation you can simply test your way to success.

You can fix everything except lack of traffic. No traffic no money.

So fix it.

TL

[Kevin Williams]

I don't know about an ugly site. I'd rather have 50 visitors a day hitting a page that converts 10% than 100 visitors a day hitting a page that barely converts 1%. Maybe that's just me

[Brad Gosse]

Originally Posted by Kevin Williams

I don't know about an ugly site. I'd rather have 50 visitors a day hitting a page that converts 10% than 100 visitors a day hitting a page that barely converts 1%. Maybe that's just me

Usually the gap is not that wide

[JohnMcCabe]

Originally Posted by Kevin Williams

I don't know about an ugly site. I'd rather have 50 visitors a day hitting a page that converts 10% than 100 visitors a day hitting a page that barely converts 1%. Maybe that's just me

I'd rather have the site that gets 100 per day and converts 10%, but that's just me...

Seriously, like a lot of topics that come up here, this is not a pure either/or choice.

You can work on both traffic and conversion at the same time. Paul Hancox has an excellent guide on testing with low traffic.

To measure conversion, you need some traffic. Once you have that traffic, you can start tweaking and measuring. Continue driving traffic.

As your traffic grows, you can do more testing in less time.

[Ross Dalangin]

Most of the time I use exit popup if I'm not fully satisfied with my conversion to make those undecided to buy in the future. Traffic is still very important to me and conversion would be converting those undecided to be a buyer soon when the sales copy is fully modified. And also, outsourcing the traffic is important for me while tweaking my sales page.

[John Burton]

Agreed - traffic is everything when you first start out. If you don't have traffic, you won't make money, no matter how pretty your site is.

John Burton

[Traffic_Is_King]

Well for me in order to tweak to get higher conversions I need data. You get data by getting traffic. Doesn't make sense to me to tweak anything when you don't have enough data to support the tweaking.

[Lance K]

Originally Posted by Brad Gosse

Just a small observation.

Many people I talk to spend more time tweaking their sites for conversions than they do generating traffic. When I ask why they say "I want to make sure the traffic converts".

This is the equivalant in the offline world of locking your doors to re arrange your store for better traffic flow.

In my experience, the only thing you should worry about is traffic at first, even if your site design is ugly. When traffic is flowing in you can make minor changes and see the results quickly.

Good point, Brad.

It reminds me of my days as an estate planner. Some planners would stop prospecting when they landed a client. Work on that plan. Then when they were done, they would wonder why they had to work so hard to secure clients.

If you're not putting anything into the funnel, it will run dry.

If people automate (as much as possible) their conversion efforts, back ends, referral systems, etc. they can spend the bulk of their time making sure that the top of the funnel is full.

[achivement84]

I agree that you have 1st to care abut generating enough trffic to your site then you may start in the conversation, as the traffic is the core point here , specially when your site is new.

[Lance Johnson]

I feel like Sam hit in on the head.

Originally Posted by Traffic_Is_King

Well for me in order to tweak to get higher conversions I need data. You get data by getting traffic. Doesn't make sense to me to tweak anything when you don't have enough data to support the tweaking.

Testing with enough data to make it statistically significant, and analyzing it in a method that is practical is crucial.

Until you match up your "tweaking" with the actual effect it has on consumers, you're just guessing. Even if it's some axiom that other marketers were POSITIVE about in a controlled experiment (think eye tracking).

So (Tweaking) + (0Traffic) = HOPEFUL Conversion Rate (Not even, really theoretical).

-Lance

[4success53]

I agree with Brad.

I had someone ask me what my conversion rate was on my new site?

How can I tell them that when I have a new website.
First thing that I am thinking is traffic, traffic traffic.
I am not interested in conversion rates until I get enough traffic and then, I will worry about that.

[JohnMcCabe]

Originally Posted by 4success53

I agree with Brad.

I had someone ask me what my conversion rate was on my new site?

How can I tell them that when I have a new website.
First thing that I am thinking is traffic, traffic traffic.
I am not interested in conversion rates until I get enough traffic and then, I will worry about that.

The trick is, too many people don't have any idea what "enough" traffic is, so they constantly focus on driving more. They never get around to worrying about conversions.

If your site turns 1% of visitors into paying customers, and you can get just one more out of every hundred (making your conversion just 2%), you double your income. To get the same effect without testing for conversion, you would have to double your traffic.

As I hinted in my first reply, I want BOTH - as much traffic as I can get converting at as high a rate as I can achieve. That's how you maximize your ROI...

[Lance Johnson]

Originally Posted by JohnMcCabe

The trick is, too many people don't have any idea what "enough" traffic is, so they constantly focus on driving more. They never get around to worrying about conversions.

I think you bring up a good point John, but could hit it from another angle.

Robbin brings up a good point in her article. Read Here

Statistical significance of testing is based on having enough data in both a numerator and a denominator: The number on top of the ratio (conversions) and the one on the bottom (views of the test page.)

In most of our cases though, we can't get statistically significant data if we try to use a small amount of traffic - or pageviews from our last example - with a small change, which would result in a very small change in conversion, we'd be SOL.

-Lance

[darrencochran]

I'm right in the midst of trying everything to get traffic to my site, so as a "newbie" I have to conquer that simply getting the traffic is the most important thing.
It's a numbers game: more hits = more sales.

[JohnMcCabe]

Originally Posted by Lance Johnson

I think you bring up a good point John, but could hit it from another angle.

Robbin brings up a good point in her article. Read Here



In most of our cases though, we can't get statistically significant data if we try to use a small amount of traffic - or pageviews from our last example - with a small change, which would result in a very small change in conversion, we'd be SOL.

-Lance

Hard to argue with the article you linked - you do indeed need to have statistically significant numbers of both trials and results to gain from testing.

It is also possible to assign a number to each before starting the test. Pick a confidence level, and you can calculate how many results you need to be significant at that confidence level. When you reach that number of results, you need to look at the number of trials it took to reach that number. Is the test result likely to be repeatable, or was it just the fabled "lucky streak?"

And setting aside the "change one word and get 10x the results" headlines, your initial big gainers are going to come with fairly significant changes. For example, testing a half-price sale versus a buy-one-get-one sale. You are testing two totally different offers, and one is likely to pull significantly better than another.

We're not on opposing sides here.

My take would be, if you need x number of page views to establish a base to test against, concentrate on driving traffic until you establish that base. Once you have a base, set up a test, figure out how many trials you need to judge the test, and concentrate on driving traffic until the test yields significant results or you are confident that there is no significant difference in the variations.

Repeat the process. It sets up a classic feedback loop.

The process hones itself until you reach the point of diminishing returns, should you care to stick with it that long.

[bobsedge]

This post is dead on.

Too many people look to build the perfect mouse trap before they know if there are mice.

First, prove the market is there.
Second, prove the market is willing to buy.
Third, work on making your system perfect

[Lance Johnson]

Originally Posted by JohnMcCabe

It is also possible to assign a number to each before starting the test. Pick a confidence level, and you can calculate how many results you need to be significant at that confidence level. When you reach that number of results, you need to look at the number of trials it took to reach that number. Is the test result likely to be repeatable, or was it just the fabled "lucky streak?"

And setting aside the "change one word and get 10x the results" headlines, your initial big gainers are going to come with fairly significant changes. For example, testing a half-price sale versus a buy-one-get-one sale. You are testing two totally different offers, and one is likely to pull significantly better than another.

My take would be, if you need x number of page views to establish a base to test against, concentrate on driving traffic until you establish that base. Once you have a base, set up a test, figure out how many trials you need to judge the test, and concentrate on driving traffic until the test yields significant results or you are confident that there is no significant difference in the variations.

Repeat the process. It sets up a classic feedback loop.

Exactamundo John. You brought up the exact reason Taguchi testing is so nuts. Plus just plain cool.

Originally Posted by JohnMcCabe

We're not on opposing sides here.

Never thought we were . There's just so much information out there, and very few people that don't have an appreciation for statistics really pay attention to what's important.

It seems like there's a lot of courses that tell you to test for "50-100 pageviews" which is really bad advice if you've got a terrible conversion rate, or are testing a ton of variables.

Good to meet a fellow tracker

-Lance

[Anthony Tilley]

Build the site to the best of your(or your web designers) ability.
Follow the usual rules of selling.
Disturb the customer, create a need, fullfil the need.

Make 3 or 4 variations of headline, sales pitch and close.

You can only tweek once you have the traffic. If you use Google PPC you can put tracking code into the site that will rotate different combinations of headline, sales pitch, close etc.

If you have a good amount of traffic you can find out which combination works pretty quickly.

[JohnMcCabe]

Originally Posted by Lance Johnson

Exactamundo John. You brought up the exact reason Taguchi testing is so nuts. Plus just plain cool.



Never thought we were . There's just so much information out there, and very few people that don't have an appreciation for statistics really pay attention to what's important.

It seems like there's a lot of courses that tell you to test for "50-100 pageviews" which is really bad advice if you've got a terrible conversion rate, or are testing a ton of variables.

Good to meet a fellow tracker

-Lance

Taguchi has its place, and a low-volume site is not it. Better to stick with A/B testing.

I used to do a crude form of manual Taguchi testing when I used Adwords more. Depending on the available searches, I'd set up 2 headlines, 2 first lines, 2 second lines and 2 urls. Mix and match to cover the available combinations, and let them rip.

Letting a simple A/B test run until you have 50-100 results is much more reasonable. Better yet is to run until you have 50-100 results on the lowest option. Then check for significance.

I'm an engineer by education, who spent most of my engineering time in a factory environment. So I gained an appreciation for the Deming method, control charts, kaizen, etc. Much of it can be applied to marketing, too.

[Michael Silvester]

And you can forget about traffic until
you have impressions

Take Care,

Michael Silvester

[Lance Johnson]

Originally Posted by JohnMcCabe

Taguchi has its place, and a low-volume site is not it. Better to stick with A/B testing.

Quite right John, but some of the huge sites, Monster.com, CNN, Amazon. Just think of the data they can pull in a day using Taguchi!

Experimental design for A/B is also much easier. You sure don't want to get stuck there, and never optimize anything!

Sidebar: You seem like the engineer kind of dude (that's a good thing). I started off in biomolecular engineering, but switched schools and am a math/physics dude now. I want to do the entrepreneur gig, but math and science is just too dang cool to ignore! Obviously business speaks math quite well too so they overlap.

-Lance