Faith Forge LabsWebsite Speed Optimization
5 MIN READ

Why is the website slower after customers sign in?

Account pages may load different information and do more work than public pages. Test the tasks your customers actually use.

Describe the personalized task

Choose an action such as viewing an order, opening a lesson or checking an account-specific price. Record the starting state and use suitable test accounts. Don’t copy real customer information into an ordinary performance report.

For a fictional membership site, a lesson page may read progress records and load a video tool that anonymous pages never use. A homepage result can’t establish how that experience behaves.

Protect correctness and separation

Personal responses must not be reused for the wrong visitor. Caching decisions need to account for authentication, cookies and the information being returned. A broad rule chosen solely for speed can create a serious correctness problem.

The MDN caching documentation explains distinctions relevant to shared and private caches. Implementation should be reviewed against the actual application and its data-handling requirements.

Two buyers looking at one part number

Give two people from your team test accounts with different price lists. Ask them to open the same product address on separate devices. The photograph, dimensions and fitting instructions are identical. One account should see a carton price negotiated for a trade customer. The other pays the ordinary unit price. Those differences matter much more than whether a speed tool gives the product page a good score.

Make a short inventory of what somebody sees when they sign in. A price might appear on the product page, search suggestions and a list of recently ordered items. The number beside the basket icon belongs to that buyer. A downloadable specification sheet for the part might be public. A quotation with the customer's name and terms on it belongs to one account. All of these can appear under the same header. They don’t all need the same rules for keeping copies.

For the first round of testing I’d pick one distinctive, harmless difference in each account. Put four cartons of washers in one basket and leave the other empty. Use price lists with amounts that someone can recognize without consulting a spreadsheet. Open the same product first as one buyer and then as the other, and repeat the visits in the opposite order. If both people now see four cartons or the contract price follows the first visitor, stop and investigate which part of the response was reused. Saving a few tenths of a second isn't a reason to expose another company's details.

There are still useful places to look for speed. Two buyers often can share the same product photograph, public instructions and stylesheet. A system may also be able to reuse work behind the scenes while separately deciding which price an authorized account should receive. The right arrangement depends on where your application applies its access rules. Hiding the wrong price in the visible page doesn't help if the private value has already been sent to the wrong account.

The next person at the counter

Some wholesale sites run all day on a shared counter computer. During testing, sign out as the first buyer and use the same browser as the second. Try following a saved product link and returning to the last order screen. Refresh the page. Check what happens when a signed-out person follows a link from an old quote email. A page briefly showing the previous account's details before correcting itself is worth reporting. Write down the sequence and use test data so you can show the problem without copying a real customer’s order into a support message.

Then make a change your sales staff actually make. Move a test buyer onto a different price list or change the quantity break on a part. Where should the new price take effect, and what happens to an existing quote that's already been agreed? That's a business rule as much as a caching decision. Staff may need a signed quotation to retain its original amount while every new basket uses today's price. Treating both amounts as “the current price” leaves you with nothing reliable to test against.

If you want help with a slow customer area, send the public address and describe the first task that feels slow after sign-in. Let me know whether buyers have different terms and whether anybody shares a computer. We can arrange suitable test accounts separately and find improvements that respect how you sell.

Look behind the screen

The delay may involve database work, an external service or repeated application checks. It may also involve frontend scripts specific to the account area. Measure the relevant operation rather than assuming that public-page optimization will carry over.

After a change, test separate users and sessions with safe sample data. Confirm that permissions, prices, progress and logout behavior remain correct. Performance acceptance should include these outcomes, not just a shorter response.

Document which roles and tasks were covered. A successful test for one user type doesn’t establish every administrative or customer workflow. More complex load testing needs a safe, explicitly agreed plan.

See application and database performance for an investigation that includes the server-side task as well as the visible page.