Insights
7 min
read
If your old site was so bad, why was it making you more money?
If your old site was so bad, why was it making you more money?
Eight years of rebuilding Shopify sites has taught us that rebuilds fail in the research phase. This is the 3-pillar process we use to protect conversion.
Jordan Hill, Founder & CEO
A site rebuild is meant to lift revenue, yet many Shopify brands see conversion fall after launch. The old design looked dated and still made money. The new one looks better and makes less. If your old site was so bad, why was it making you more money?
Eight years of rebuilding Shopify sites has taught us that the answer sits in the research phase. This guide sets out the research that shows what the old site was doing right, so the rebuild can add to it.
What is research for a Shopify site rebuild?
Research for a Shopify site rebuild is the process of finding out which parts of an existing site drive revenue, so the redesign keeps what converts and only adds to it. It sets out what can change and what categorically cannot. A rebuild that skips it risks removing the copy, feature or element that was driving conversions, and a change that large leaves no way to trace what went wrong.
At That Works Agency, we use the 3-Pillar Rebuild Research to decide what a new site keeps, adds and tests. It covers heatmap analysis, funnel analysis and customer sentiment, and it follows one rule: add, never remove.
Why a rebuild can lower conversion on a site that was making money
We have seen many site rebuilds tank conversion rates. The cause is the quality of the research phase. That excludes discovery, the stage that decides how much the agency charges.
An old site that looks dated can still be the more profitable site. If it was making money, something on it was working. Every time an element is removed or adjusted, there is a risk of killing the exact copy, feature or element that drove those conversions. Remove or alter too much at once and the cause of a drop cannot be traced. That is where rebuilds fail.
Quick answer: the 3-Pillar Rebuild Research
Heatmap analysis - overlay financial data on a heatmap to see which elements drive conversion and which are only there for show.
Funnel analysis - check how traffic enters the site from the marketing funnel, because the landing page is often not the homepage.
Customer sentiment - collect the objections of the people who did not buy, then answer them on the site.
A site rebuild is meant to lift revenue, yet many Shopify brands see conversion fall after launch. The old design looked dated and still made money. The new one looks better and makes less. If your old site was so bad, why was it making you more money?
Eight years of rebuilding Shopify sites has taught us that the answer sits in the research phase. This guide sets out the research that shows what the old site was doing right, so the rebuild can add to it.
What is research for a Shopify site rebuild?
Research for a Shopify site rebuild is the process of finding out which parts of an existing site drive revenue, so the redesign keeps what converts and only adds to it. It sets out what can change and what categorically cannot. A rebuild that skips it risks removing the copy, feature or element that was driving conversions, and a change that large leaves no way to trace what went wrong.
At That Works Agency, we use the 3-Pillar Rebuild Research to decide what a new site keeps, adds and tests. It covers heatmap analysis, funnel analysis and customer sentiment, and it follows one rule: add, never remove.
Why a rebuild can lower conversion on a site that was making money
We have seen many site rebuilds tank conversion rates. The cause is the quality of the research phase. That excludes discovery, the stage that decides how much the agency charges.
An old site that looks dated can still be the more profitable site. If it was making money, something on it was working. Every time an element is removed or adjusted, there is a risk of killing the exact copy, feature or element that drove those conversions. Remove or alter too much at once and the cause of a drop cannot be traced. That is where rebuilds fail.
Quick answer: the 3-Pillar Rebuild Research
Heatmap analysis - overlay financial data on a heatmap to see which elements drive conversion and which are only there for show.
Funnel analysis - check how traffic enters the site from the marketing funnel, because the landing page is often not the homepage.
Customer sentiment - collect the objections of the people who did not buy, then answer them on the site.
Pillar 1: Heatmap analysis
Heatmap analysis is arguably the most important pillar. A tool such as Heatmap.com lets us overlay financial data on top of a heatmap, so we can see what on the site drives conversion and what is there for show. We categorise every element by its highest and lowest converting levers.
What heatmaps show
A heatmap tracks interactions, and that is how revenue gets assigned to an element. The result is a ranked list of levers. High converting levers stay where they are. Low converting levers become candidates for testing.
Real example: If a large share of traffic uses a sticky search bar and it converts well, its placement cannot change. Another agency might dislike how it looks in the header and remove it with a weak reason. That one change could cost seven figures.
What heatmaps cannot show
Heatmaps cannot detect the copy and messaging that convinced someone to add to basket. A headline's contribution to revenue never appears in a heatmap, because people do not click it. This is why the research follows the add, never remove rule. If the data cannot prove that an element is not working, it cannot be assumed that the element is not working.
Pillar 2: Funnel analysis
A rebuild needs an understanding of how traffic flows into the site from the marketing funnel. Without it, the design is built blind.
Many agencies treat the homepage as the front door. The data can show something different. For example, 50% of traffic can come from Meta and land straight on a collection page.
Example: Visitors from an ad never see the homepage. They arrive with a specific expectation, and the collection page does the job of a landing page without being built for it.
What changes once the traffic path is known
Every decision on that page changes. The sitewide offer may need explaining there. It has to be clear what the brand offers. The placement of social proof needs a decision. The site is the same, and the strategy is completely different, purely because of how the traffic enters.
Pillar 3: Customer sentiment
The site's job is to answer customer objections before they are asked. That requires an understanding of the customer and the objections they hold.
Where to find the real objections
Surveys are useful, but only people who already bought answer them, so the opinions are skewed. The most useful material sits in Facebook ad comments and Reddit threads. That is where the true sceptics live, the people who did not buy and will say why.
How objections become a build brief
Recurring objections turn into a brief for the site.
Example: If the recurring complaint is that the product is too expensive, the site has to do the convincing. That means comparison tables against cheaper brands, the science behind why the product works better and the cost-per-use maths. The price becomes a story to tell.
Everything already on the site is there for the people who convert. Everything added is for the sceptics who do not. That is the add, never remove principle in practice.
Turn the research into a design presentation
Data analysts run the research. The next step is making it visual. A real skill goes into translating 30 pages of reports into a before and after presentation, with data points behind every design choice.
The presentation educates the client as much as it presents a design. When a brand understands why every decision was made and sees the data behind it, approval is fast. Nothing rests on a designer's opinion or on something someone once read on Twitter.
Test removals after launch
If something looks like it should go, test removing it after launch. This protects the integrity of the investment, because a removal is isolated and its effect can be measured.
Pillar 1: Heatmap analysis
Heatmap analysis is arguably the most important pillar. A tool such as Heatmap.com lets us overlay financial data on top of a heatmap, so we can see what on the site drives conversion and what is there for show. We categorise every element by its highest and lowest converting levers.
What heatmaps show
A heatmap tracks interactions, and that is how revenue gets assigned to an element. The result is a ranked list of levers. High converting levers stay where they are. Low converting levers become candidates for testing.
Real example: If a large share of traffic uses a sticky search bar and it converts well, its placement cannot change. Another agency might dislike how it looks in the header and remove it with a weak reason. That one change could cost seven figures.
What heatmaps cannot show
Heatmaps cannot detect the copy and messaging that convinced someone to add to basket. A headline's contribution to revenue never appears in a heatmap, because people do not click it. This is why the research follows the add, never remove rule. If the data cannot prove that an element is not working, it cannot be assumed that the element is not working.
Pillar 2: Funnel analysis
A rebuild needs an understanding of how traffic flows into the site from the marketing funnel. Without it, the design is built blind.
Many agencies treat the homepage as the front door. The data can show something different. For example, 50% of traffic can come from Meta and land straight on a collection page.
Example: Visitors from an ad never see the homepage. They arrive with a specific expectation, and the collection page does the job of a landing page without being built for it.
What changes once the traffic path is known
Every decision on that page changes. The sitewide offer may need explaining there. It has to be clear what the brand offers. The placement of social proof needs a decision. The site is the same, and the strategy is completely different, purely because of how the traffic enters.
Pillar 3: Customer sentiment
The site's job is to answer customer objections before they are asked. That requires an understanding of the customer and the objections they hold.
Where to find the real objections
Surveys are useful, but only people who already bought answer them, so the opinions are skewed. The most useful material sits in Facebook ad comments and Reddit threads. That is where the true sceptics live, the people who did not buy and will say why.
How objections become a build brief
Recurring objections turn into a brief for the site.
Example: If the recurring complaint is that the product is too expensive, the site has to do the convincing. That means comparison tables against cheaper brands, the science behind why the product works better and the cost-per-use maths. The price becomes a story to tell.
Everything already on the site is there for the people who convert. Everything added is for the sceptics who do not. That is the add, never remove principle in practice.
Turn the research into a design presentation
Data analysts run the research. The next step is making it visual. A real skill goes into translating 30 pages of reports into a before and after presentation, with data points behind every design choice.
The presentation educates the client as much as it presents a design. When a brand understands why every decision was made and sees the data behind it, approval is fast. Nothing rests on a designer's opinion or on something someone once read on Twitter.
Test removals after launch
If something looks like it should go, test removing it after launch. This protects the integrity of the investment, because a removal is isolated and its effect can be measured.
The bottom line
A rebuild protects growth when the research is thorough enough to show what the old site was doing right. Use the 3-Pillar Rebuild Research to rank the levers that convert, understand how traffic enters, collect the objections of non-buyers and add to the site on that evidence. Test removals after launch.
Go deeper into related reading:
The Hierarchy of CRO - the framework for ordering CRO work
Why your conversion rate drops when you scale ads, and how to fix it in 3 steps - what to check when ad traffic stops converting
The bottom line
A rebuild protects growth when the research is thorough enough to show what the old site was doing right. Use the 3-Pillar Rebuild Research to rank the levers that convert, understand how traffic enters, collect the objections of non-buyers and add to the site on that evidence. Test removals after launch.
Go deeper into related reading:
The Hierarchy of CRO - the framework for ordering CRO work
Why your conversion rate drops when you scale ads, and how to fix it in 3 steps - what to check when ad traffic stops converting
Frequently asked questions
Why do site rebuilds lower conversion rate?
What does add, never remove mean in a site rebuild?
What should I research before rebuilding a Shopify site?
Where do I find customer objections for a rebuild brief?
How does rebuild research connect to the That Works Hierarchy of CRO?
Insights
8 min
How to choose a Shopify CRO agency in the UK
What to check before you hire a Shopify CRO agency in the UK, from a Shopify Premier Partner that has run tests across hundreds of Shopify stores.
Copy
9 min
How to build customer personas from ecom data to win more conversions
The exact 8-step process we use to extract buyer personas from ecom data, validate them against real behaviour, and test messaging that lifts conversion.
Insights
5 min
The anatomy of the perfect offer: 6 fundamentals to scale your brand
A practical breakdown of every component that determines whether an offer converts, retains, and scales. Built from testing across Shopify brands
Insights
8 min
How to choose a Shopify CRO agency in the UK
What to check before you hire a Shopify CRO agency in the UK, from a Shopify Premier Partner that has run tests across hundreds of Shopify stores.
Copy
9 min
How to build customer personas from ecom data to win more conversions
The exact 8-step process we use to extract buyer personas from ecom data, validate them against real behaviour, and test messaging that lifts conversion.
