Where Does Website Speed Matter Most?
A slow website is a problem. A slow website when someone is trying to finish something can be a bigger one. Research on online retail suggests that customers do not react to website delays the same way at every point in the journey. In one peer-reviewed study, shoppers were particularly sensitive to slowdowns during checkout. That matters because website performance is often treated like one sitewide number. Fast. Slow. Pass. Fail. Customer behavior is not that simple.
The Question
Most performance conversations start with the website as a whole.
How fast is the site? What is the performance score? Are the Core Web Vitals good?
Those are useful questions, but they can hide a more important one:
Where is the customer waiting?
A one-second hesitation while casually browsing and a delay while someone is submitting payment, choosing an appointment, or completing a lead form do not necessarily interrupt the same decision.
Once someone has decided to act, the website has a different job. It needs to let them finish.
What the Research Examined
Santiago Gallino, Nil Karacaoglu, and Antonio Moreno studied the effect of what they call in-process delays in online retail. [1]
They combined retail and website-speed data to examine how slowdowns affected online sales and whether customer sensitivity to those delays changed depending on where someone was in the shopping journey. [1]
The researchers used threshold regression models rather than a randomized website experiment. Their analysis estimated sizable adverse effects from website slowdowns and examined how abandonment changed across different stages of the customer journey. [1]
This distinction matters. The study provides strong evidence from real retail behavior, but it is not the same thing as randomly assigning service-business prospects to different booking speeds. [1]
What the Research Found
Website slowdowns were associated with worse commercial outcomes, but the effect was not uniform throughout the journey. [1]
The researchers found that customers showed diminishing sensitivity as slowdowns increased, and that waiting affected abandonment differently depending on where the customer was in the process. [1]
Most importantly for this discussion, customers were more sensitive to slowdowns during checkout. The authors specifically identify checkout performance as an area with practical implications for website design. [1]
What This Does—and Doesn’t—Tell Us
That is the useful part.
Someone browsing products is still deciding whether to continue. Someone at checkout has already made a series of decisions and is trying to complete the transaction.
Introducing friction at that point gets in the way of an action already underway.
- Waiting Has Never Been a Simple “Faster Is Always Better” Problem
- Earlier research helps explain why waiting deserves more context than a single speed score.
Dellaert and Kahn studied waiting across four computer-based experiments and found that delays could negatively affect evaluations of websites, but not in every circumstance. They also found that managing the waiting experience could reduce some of those negative effects. [2]
Another peer-reviewed study exposed 25 participants to webpage delays of 1, 30, and 60 seconds while they completed information-finding tasks. Longer delays produced more frustration and lower task completion, although some of the other performance measures were less straightforward. [3]
That nuance is useful. Website performance affects people, but the task, the stage of the journey, and what the customer is waiting for all matter.
- What Happens During the Wait Matters Too
- Sometimes a delay cannot disappear completely. A payment has to process. A scheduling tool has to check availability. A quote calculator has to return a result. A large file or application has to finish an operation.
That does not mean the experience has to feel dead.
Research on progress indicators gives us another useful clue. Conrad and colleagues ran two experiments involving online questionnaires and found that the way progress was presented affected abandonment and subjective experience. When early progress appeared slow, participants were more likely to abandon and reported a worse experience. [4]
That does not mean adding a progress bar fixes a slow website. It means people react not only to elapsed time, but also to what the interface tells them is happening while they wait.
The evidence gives us good reason to treat performance as part of the customer journey rather than just a technical score. It supports the idea that delays can hurt customer behavior, that sensitivity to those delays can change depending on where someone is in the process, and that high-intent stages deserve particular attention. [1] [2] [3] [4]
What it does not give us is a universal load-time threshold for every website. It does not establish that every millisecond of improvement produces the same commercial return. It does not prove that a service-business booking form will behave exactly like an ecommerce checkout. And it does not mean the homepage can be slow simply because checkout appears to be more sensitive.
The useful question is:
Where can waiting interrupt something the customer is already trying to finish?
The closer someone is to finishing, the less room there is for the website to get in the way.
What This Means for Your Website
Performance problems are not always dramatic. Sometimes the whole website feels slow. That is obvious.
More often, one specific part of the experience creates the damage. The embedded calendar takes too long to appear. A form hangs after submission. The next step loads a heavy third-party script. A payment screen pauses without feedback. A mobile visitor taps a button and nothing seems to happen.
Those problems are especially worth finding when they sit directly between intent and completion.
A customer who is already trying to call, book, buy, request a quote, or submit information should not have to fight the website at the finish line.
Key Takeaway
Start With the Critical Path
A perfect performance score is not the business objective. A customer successfully completing the task is.
That means performance work should begin with the parts of the website tied most closely to real customer actions.
For a service business, that may include the landing page that receives paid traffic, the form used to request an estimate, the appointment scheduler, the mobile click-to-call experience, or any step between starting an enquiry and receiving confirmation that it actually went through.
Once those paths work well, keep improving the rest of the site.
But do not spend all of your attention shaving small amounts from low-intent pages while the booking experience is making customers wait.
Key Takeaway
Third-Party Tools Still Count as Your Experience
Businesses often treat embedded software as someone else’s problem. It is not.
If your scheduling system takes too long to load, the customer experiences your website as slow. If your chat widget blocks the page, that is part of the experience. If your form integration hesitates after someone presses Submit, they do not care which vendor caused it.
From the customer’s perspective, there is one journey. That is why performance reviews need to include the tools layered onto the website—not just the HTML and images you directly control.
Key Takeaway
Do Not Leave People Wondering Whether Anything Happened
When a process needs time, the interface should make that state clear. A button can change after it is pressed. A form can indicate that it is submitting. A booking tool can show that availability is loading. A longer process can show useful progress where appropriate.
The point is not animation for its own sake.
It is eliminating the moment where the customer thinks:
Did that work?
The research on progress feedback suggests that poorly handled feedback can make waiting feel worse, so this still requires judgment. [4]
Silence is not always neutral.
The Practical Takeaway
Do not stop at asking whether your website is fast.
Follow the customer through the action that matters most.
Open the site on a phone. Find the service. Tap the CTA. Load the form. Choose the appointment. Submit the enquiry. Wait for confirmation.
Then look at where the experience hesitates.
Which step takes too long? Which third-party tool adds friction? Where does the customer receive no feedback? Where are they closest to completing the action when the website makes them wait?
That is where performance becomes more than a technical issue.
It becomes part of conversion.
Where This Applies at CTG
Landing Pages
Getting someone through an ad and onto the page is only the beginning. We look at the complete conversion path, including mobile performance, forms, third-party tools, and the transition from interest to action.
Calls, Forms & Booking
High-intent interactions deserve particular attention. We look at what happens after someone decides to call, submit, or schedule—not just whether the CTA is visible.
Conversion Optimization
Performance can be a source of behavioral friction. We evaluate it alongside messaging, hierarchy, usability, analytics, and downstream outcomes rather than treating a technical score as the conversion result.
Website Redesigns
A redesign can introduce new scripts, integrations, media, and interaction costs. Performance needs to remain part of the design decision rather than becoming a cleanup task after the site is finished.
Sources
Santiago Gallino, Nil Karacaoglu, Antonio Moreno (2022). Need for Speed: The Impact of In-Process Delays on Customer Behavior in Online Retail Operations Research, 71(3), 876–894. DOI: 10.1287/opre.2022.2262.
Peer-reviewed journal
Measured outcomes: Customer behavior under website delays in online retail
Benedict G.C. Dellaert, Barbara E. Kahn (1999). How tolerable is delay?: Consumers’ evaluations of internet web sites after waiting Journal of Interactive Marketing, 13(1), 41–54. DOI: 10.1002/(SICI)1520-6653(199924)13:1<41::AID-DIR4>3.0.CO;2-S.
Peer-reviewed journal
Study design: Four computer-based experiments
Measured outcomes: Retrospective website evaluations under waiting; Effects of managing waiting experiences
Paula R. Selvidge, Barbara S. Chaparro, Gregory T. Bender (2002). The World Wide Wait: Effects of Delays on User Performance International Journal of Industrial Ergonomics, 29(1), 15–20. DOI: 10.1016/S0169-8141(01)00045-2.
Peer-reviewed journal
Study design: Information-location tasks with webpage delays of 1, 30 and 60 seconds
Sample size: 25 participants
Measured outcomes: Frustration; Task completion; Lostness
Frederick G. Conrad, Mick P. Couper, Roger Tourangeau, Andy Peytchev (2010). The impact of progress indicators on task completion Interacting with Computers, 22(5), 417–427. DOI: 10.1016/j.intcom.2010.03.001.
Peer-reviewed journal
Study design: Two experiments involving online questionnaires
Measured outcomes: Questionnaire abandonment; Subjective experience of progress feedback
About This Review
CTG reviews research to better understand how people interact with websites, search experiences, and digital decision environments. We separate published findings from our interpretation and do not treat one study as a universal rule.
Last reviewed: