Performance Commerce: Faster Site, Lower Ad Costs

In Brief

Performance commerce treats your site's technical speed as a marketing cost line rather than an infrastructure detail. When a page loads slowly, the visitor bounces, the ad platform reads that click as a poor match, your cost per click rises, and the same budget buys fewer orders. A fast stack pulls conversion up and pushes ad cost down, and the two effects compound. Put simply, part of your ad budget may be spent by your server rather than your agency.

When cost per click climbs from 4 TRY to 5.20 TRY at a store spending 200,000 TRY a month on Meta, nobody opens the server logs. The ad copy gets rewritten first, then the creative, then the audience gets narrowed. Meanwhile, that same week, the mobile product page went from loading in 2.1 seconds to 3.4 seconds. That is where the money went.

Ad platforms measure what the person who clicked actually does on your site. A visitor who leaves before the page paints counts as a failed match. As those signals accumulate, your ad loses standing in the auction and you pay more for the same impressions. A slow site charges you twice: once in the lost sale, once in the higher bid.

The invisible link between code and ad cost

The moment someone taps your ad while scrolling Instagram, you pay. That click is already an expense. Everything after it belongs to you. If the largest visual element on the product page takes longer than 2.5 seconds to appear, the odds of losing that visitor before purchase rise sharply. Industry measurements put bounce rates roughly 38 percent higher on pages that cross this threshold.

The second layer gets discussed far less often. Google and Meta watch what happens on the landing page they sent the user to. Someone who backs out before the page renders drags down the quality signal attached to your ad, which in an auction means paying more than a competitor for the same audience. Once that signal degrades it also takes time to recover, because the platform writes new data over the old slowly.

Same budget, two outcomes

Picture two stores, each spending 100,000 TRY a month on Meta, each paying 4 TRY per click. Both receive 25,000 clicks. The fast store converts at 2.4 percent, the slow one at 1.3 percent. Average order value is 900 TRY. The fast store books 600 orders and 540,000 TRY in revenue, the slow store stops at 325 orders and 292,500 TRY. Nobody spent an extra lira to create that 247,500 TRY gap.

• To match the revenue, the slow store has to roughly double its budget.
• The gap widens over time, as a weak quality signal keeps pushing cost per click up.
• The fast store builds a larger retargeting pool, since more people actually saw the page.
• The fast store can run the same creative longer, because the campaign does not burn out early.

If you want the conversion side broken down lever by lever, our piece on practical ways to raise your conversion rate reads as a direct continuation of this arithmetic.

Reading Core Web Vitals in money terms

Core Web Vitals sound like developer jargon at first glance. All three tie straight to revenue. LCP measures when the largest element on the page becomes visible, INP measures how quickly the page answers a tap, and CLS measures whether content shifts under the reader's eyes. A button that moves on the payment screen is a customer who hits the wrong control and gives up.

Metric	Good threshold	What it means in practice	Budget impact
LCP	Under 2.5 seconds	Product image and price appear on screen	Keeps the paid click from being wasted
INP	Under 200 milliseconds	Size selection and add to cart respond instantly	Protects add-to-cart rate
CLS	Under 0.1	The page does not jump when the promo bar loads	Reduces misclicks during checkout
TTFB	Under 800 milliseconds	The server sends the first byte quickly	Keeps the site standing during campaign hours

Where to take the measurement

The Experience section inside Search Console reports field data, which means what really happens on a Turkish mobile network rather than in a lab. PageSpeed Insights tests the same page under lab conditions and lists every file with its weight. In GA4, build an Exploration report that puts device category next to conversion rate, and the gap between desktop and mobile becomes visible. If mobile converts at less than half the desktop rate, the creative is not your problem.

Do not measure a single page. The homepage flatters almost every store, because it gets the most attention from the design team. What matters is where your ads land. Measure the category and product pages your campaign links point to, list the worst three, and track them weekly.

Cart to payment: the most expensive forty seconds

From a budget perspective, your most valuable visitor is the one with something in the cart. You paid to bring them in, they showed interest, they reached the decision point. Every person lost here costs far more than a loss earlier in the funnel. Those forty seconds inside checkout carry the densest return in the whole marketing chain.

Details that break the flow

• Guest checkout switched off. Forcing a password breaks the momentum.
• Province, district and neighbourhood fields each making their own round trip to the server. Every wait creates a new exit point.
• Shipping cost appearing as a surprise at the last step. Once the price expectation breaks, the cart is abandoned.
• A 3D Secure screen that opens in a new tab and disappears on mobile.
• Card fields that block autofill and fail to bring up the numeric keypad.

Run the numbers. Say 4,000 people add to cart each month, 1,400 reach the payment step and 900 complete the order, which puts checkout completion at 64 percent. Collapse the address form into one screen and show shipping cost in the cart, and if completion moves to 71 percent you gain 98 extra orders a month. At a 900 TRY average order value that is 88,200 TRY in additional revenue with no change to ad spend.

Completion rate is not the only thing worth logging on the payment page. Record which bank, which instalment option and which device produced the error. In most stores a single virtual POS setting or one browser incompatibility explains a large share of monthly losses.

What your stack does when traffic spikes

You have signed a creator whose post goes live at 21:00. Or you opened your discounts on the night of 11 November. Thousands of people arriving at once is the scenario every ecommerce manager wants. On shared hosting or an ageing monolith, that scenario inverts within minutes: carts throw errors, the payment page times out, and your campaign keeps pushing traffic to a site that cannot serve it.

The damage outlives the evening. The ad platform sees no conversions during the busiest hour of the campaign and restarts its learning phase. Even after the site recovers the next morning, the campaign takes several days to return to its earlier efficiency.

What elastic infrastructure actually buys

Auto-scaling means server capacity grows on its own as traffic climbs. The store behaves the same way with ten visitors as it does when 100,000 people hit the buy button at the same moment. Serving content through a global delivery network lets a shopper in Izmir pull images from the nearest node. Trimming database queries and adding a cache layer stops every single page request from taxing the origin server during peak hours.

The pre-campaign checklist is short. Publish the campaign page a day early to warm the cache, convert images to webp, temporarily reduce third-party scripts on the homepage, and confirm your payment provider's transaction limits in advance. For the decisions that create this problem in the first place, our article on costly mistakes in platform selection doubles as a pre-campaign checklist.

Your dashboard is a command center, not a catalog tool

Traditional ecommerce packages design the admin panel as a product entry form. What a merchant needs is different: a screen that shows the financial pulse of the business. You already know how many units you sold. What you do not know is exactly where the funnel is bleeding. Are 60 percent of customers stalling at the address step? What share hit an error on the card screen? Which marketplace actually leaves a profit once commission is counted? Those answers live in a report screen, not a product list.

Four screens worth demanding

• A step-by-step funnel report. Category, product, cart, payment and thank-you page transitions in one table.
• Channel breakdown. Conversion rate read separately for Google, Meta, email, organic search and marketplace traffic.
• A payment error log. Which bank, which amount, which error code.
• Profit by product. Not revenue, but what survives shipping and returns.
• Stock cover and sell-through. Knowing how many days of stock a product has left should come before you point budget at it.

Put those together and marketing decisions stop being guesswork. Traffic from Trendyol may look cheap on a cost-per-order basis until commission and return rates enter the calculation and your own site turns out to be more profitable. On Hepsiburada, volume can rise on campaign days while average order value falls. Allocating budget without seeing both channels in one table is shooting in the dark.

It also matters that the panel does not depend on a stack of third-party tools. If reading the data requires a separate analytics subscription, a separate reporting screen and a separate user management setup, the team will not open it even once a day. Because the Tecof ecommerce platform keeps these reports inside store management, the distance between seeing a number and acting on it collapses to one click.

Splitting ad budget using infrastructure data

Most brands allocate budget by looking at last month's table. A healthier method reads channel performance alongside the technical state of the site. Raising Meta spend on a site that is slow on mobile only widens the hole, since the overwhelming majority of Meta traffic is mobile. Search behaves a little differently. Someone typing a query already has a need in mind and tolerates a few hundred milliseconds more than a shopper drifting through a feed. That tolerance is not unlimited. It only buys you a bit of time.

Situation	Mobile LCP	Recommended move	Expected effect
Fast site, ROAS above 4	Under 2 seconds	Scale budget gradually	Volume grows, efficiency holds
Fast site, ROAS near 2	Under 2 seconds	Test creative and audience	The issue is the message, not the page
Slow site, ROAS falling	Over 3 seconds	Freeze budget and fix the page	Cost drops within two weeks
Slow site, search performing	Over 3 seconds	Shift weight toward search	High-intent traffic limits the loss

Using this table requires a ROAS calculation you can trust. A ROAS that ignores return rates and shipping costs manufactures campaigns that look healthy on paper and lose money in the bank. The formula and the usual mistakes are laid out step by step in our guide to calculating return on ad spend.

There is no single correct channel weighting. Products with a high average order value and a long decision cycle usually behave more predictably on search. Products sold on visuals and bought on impulse shift the weight toward social. Move budget in increments of about 15 percent rather than all at once, and wait two weeks after each step, because the learning phase makes early results look worse than they are. We compared how the two platforms differ under 2026 conditions in our Google Ads and Meta Ads budget split guide.

The local side of performance in Turkey

International performance guides skip a few realities of the Turkish market. Mobile carries a large share of traffic here, and network conditions vary considerably from city to city. One shopper opens your campaign over fiber in Istanbul while another tries to load the same page on a weak mobile signal in a smaller district. Once total page weight passes 3 MB, you will most likely lose the second one.

Compliance is a performance topic too

Under KVKK, explicit consent notices have to appear without blocking the checkout flow. Showing consent at the relevant step, rather than stacking every checkbox onto one screen, keeps you compliant and lowers abandonment. If you plan to send commercial electronic messages, IYS registration is mandatory, which makes correct labelling of the opt-in field both a legal and a marketing asset. Every address you collect with permission is money you do not pay to reach that person through ads next month.

Payments and shipping

With local payment providers, requesting the instalment table after the card number is entered rather than on page load noticeably shortens first render. On the logistics side, integrations with carriers such as Yurtici, Aras and MNG belong in a background job rather than in the order request itself. Otherwise order completion becomes hostage to a carrier's response time, and the day one provider slows down, your orders slow down with it.

The balance between marketplaces and your own store is technical as well. If stock synchronisation lags, a product already sold on Trendyol stays available on your site and your cancellation rate climbs. You can see how growing brands hold that balance, with examples, on our solutions page for ecommerce brands.

A thirty-day performance plan

You can produce a measurable difference without launching a full replatforming project. The plan below is sized for a marketing team of one.

Week one: measure

List the five pages that receive the most paid traffic. Record the mobile LCP for each. Build a conversion report in GA4 broken down by device. Note the abandonment rate at the payment step. Those four numbers become your baseline.

Week two: easy wins

Convert images to webp and add explicit dimensions to product images. Remove unused third-party scripts, since live chat and heatmap tools do not all need to load at once. Serve fonts from your own domain. Those three moves alone are usually worth close to half a second.

Week three: the checkout flow

Turn on guest checkout, collapse the address form into one screen, show shipping cost on the cart page. Start logging payment errors. Do not push the change to all traffic at once, run it on half your traffic for a week and compare completion rates between the two groups.

Week four: reallocate

Measure the same four numbers you recorded in week one. If mobile LCP dropped and checkout completion rose, raise Meta budget by 20 percent and watch it for two weeks. If nothing moved, hold budget flat, because the page is still the constraint. Here is the first thing to do tomorrow morning: open the landing pages of your three highest-spending campaigns, switch your phone to mobile data, and time them with a stopwatch. Every page that crosses three seconds is evidence of where this month's budget is going.

Frequently Asked Questions

Does site speed really affect my ad costs?

Yes, through two paths. Directly, a slow page converts worse and raises cost per order. Indirectly, ad platforms treat landing page experience as a quality signal, and a weak signal means a higher auction price. Together they produce visibly fewer sales from the same budget.

What speed is good enough?

Having the largest content element appear in under 2.5 seconds on mobile is the widely accepted threshold. On pages that receive paid traffic, aiming under 2 seconds is safer, because those visitors have no relationship with your brand and very little patience. A fast homepage is not enough, so measure the campaign pages themselves.

Can I gain speed without changing my platform?

Up to a point, yes. Image optimisation, removing unnecessary scripts, font handling and cache settings typically win between half a second and a full second. Delays caused by server response time and database queries cannot be solved at the theme level. That is where the decision moves to infrastructure.

My site crashes during campaigns, what should I do?

First locate the failure. If the homepage loads but payment does not work, the culprit is usually the payment provider or database contention. If the site stops answering entirely, server capacity is short. Auto-scaling infrastructure solves the second case on its own, while the first calls for a limits conversation with your provider before the campaign starts.

My conversion rate is low, should I fix the ads or the site first?

Look at post-click behaviour. A high share of visitors leaving before seeing the page points to speed. People browsing but never adding to cart points to the offer, the price or the product description. People adding to cart and vanishing at payment points to the flow. The three have different fixes, so identify where the loss happens before spending anything.

Would a mobile app solve my speed problem?

An app helps loyal customers, but paid traffic almost always lands on the web. The download step introduces new friction and costs you the first-time visitor. Speeding up the web experience first, then evaluating an app for repeat buyers, is the more efficient order.

I sell on marketplaces, does this apply to my own store?

On a marketplace, speed is not yours to control and competition runs on price and delivery. On your own site, both the speed and the customer data belong to you. That is the biggest return on steering budget to your own store: no commission, and the customer list stays with you.

How do I measure the impact of a speed improvement?

Compare the four weeks before and after the change using the same channel breakdown. Put conversion rate, cost per order, checkout completion and mobile LCP side by side. Where seasonality matters, check the same period last year too. Avoid judging on a single day, because campaign days distort the picture.

Do KVKK and IYS compliance slow the site down?

They do when implemented carelessly. Heavy consent banners and third-party permission tools loading on first paint stretch render time. Showing the consent text at the relevant step and collecting IYS approval on the post-order screen keeps you compliant while keeping the page light.

Should I outsource this to an agency?

Your own team can handle the measurement and prioritisation, which needs five pages and four metrics. Theme code and server-side work call for engineers. If you already work with an agency, ask them to add mobile LCP and checkout completion to the monthly report so the conversation stops revolving around ad copy alone.