Why Good Enough Performance Is No Longer Enough

Why Good Enough Performance Is No Longer Enough ![rw-book-cover](https://img.youtube.com/vi/1Vg7CycTdWU/maxresdefault.jpg) 🔗 **Source:** [Why Good Enough Performance Is No Longer Enough](https://www.youtube.com/watch?v=1Vg7CycTdWU) > [!note]+ Document Notes > tags: #web-performance #lcp #user-experience #core-web-vitals #conversion-optimization > author: Georgie Petra > source: YouTube > reference: https://www.youtube.com/watch?v=1Vg7CycTdWU > date created: 2025-01-03 15:30 > > # 🏷️ Title: Why Good Enough Performance Is No Longer Enough >  ## 🧰 Action Inbox - [H1] Implement real user monitoring over synthetic testing for live websites — production sites need RUM data #web-performance - [H2] Benchmark LCP against industry averages and direct competitors — user expectations set by last visited site #competitive-analysis - [H3] Measure LCP impact on engagement metrics like page views per session and conversion rates — correlate performance to revenue #analytics ## 🧭 Table of Contents - [[#📝 Summary]] - [[#📖 Main Content]] - [[#🗒️ Notes]] - [[#📚 References]] - [[#🗣️ Paraphrase]] - [[#📰 Newsletter]] - [[#✅ Action Items]] - [[#🧪 Hypotheses]] - [[#📊 Methodology]] - [[#❓ Questions]] - [[#🔮 Future Research]] - [[#💡 Implications]] - [[#🧠 Learning Styles Assessment]] - [[#📊 Accessibility Assessment]] - [[#🌍 DEIB Assessment]] - [[#✅ Logical Validity]] - [[#🛡️ DARVO Analysis]] - [[#🎯 Conclusion]] ### 🔗 References - [UXifi](https://uxifi.com) - [Core Web Vitals](https://web.dev/vitals/) ## 📝 Summary - Performance expert Georgie Petra argues that Google's 2.5-second LCP threshold is insufficient for competitive advantage - User expectations vary by industry, device, and network conditions - generic thresholds miss business-specific optimization opportunities - Real user monitoring reveals direct correlations between LCP improvements and engagement/conversion metrics - Performance optimization should be treated as a business growth lever rather than just technical quality improvement - Modern web technologies often prioritize developer productivity over visitor performance, requiring conscious performance-first decisions ## 📖 Main Content Georgie Petra, co-founder of UXifi and former nitro pack contributor, presents a data-driven approach to web performance optimization that goes beyond Google's Core Web Vitals thresholds. The core argument challenges the industry standard 2.5-second LCP (Largest Contentful Paint) benchmark, demonstrating how user patience and expectations vary significantly across industries, audiences, and competitive landscapes. The presentation traces performance measurement evolution from 2018's fragmented 3-second loading standards to 2020's Core Web Vitals introduction, which standardized LCP as the primary perceived loading metric. However, Petra argues that as more websites achieve the 2.5-second threshold, competitive differentiation requires faster performance tailored to specific user contexts. The decision matrix framework starts with Google's 2.5-second baseline but progresses through industry averages, direct competitor benchmarks, and most importantly, user-specific LCP sensitivity analysis. This involves correlating real user monitoring data with engagement metrics like page views per session, add-to-cart rates, and checkout conversions to identify optimal performance targets for specific business outcomes. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/1Vg7CycTdWU?t=0) — [H1] Introduction and speaker background - Key idea: Georgie Petra has been in performance industry since 2013 with nitro pack and now co-founded UXifi - Evidence: "I'm involved in the performance industry since 2013 when we started the first version of nitro pack" #### ⏱️ [05:30](https://youtu.be/1Vg7CycTdWU?t=330) — [H2] Performance measurement history pre-2020 - Key idea: Before Core Web Vitals, there was no standardized way to measure performance for real users - Evidence: "In 2018 there was no core vitals there was no LCP and there was no anything that actually mattered for real understanding the real user" #### ⏱️ [08:45](https://youtu.be/1Vg7CycTdWU?t=525) — [H1] Core Web Vitals introduction changed everything - Key idea: 2020 launch of Core Web Vitals with 2.5-second LCP threshold standardized performance measurement - Evidence: "This completely changed the performance industry and everybody started to understand the significance" #### ⏱️ [12:15](https://youtu.be/1Vg7CycTdWU?t=735) — [H2] User patience varies across contexts - Key idea: User expectations differ by industry, device, and network conditions rather than universal thresholds - Evidence: "User patience is different across industries, across websites, different sometimes even within the same website across time" #### ⏱️ [15:20](https://youtu.be/1Vg7CycTdWU?t=920) — [H3] Real user monitoring vs synthetic testing - Key idea: Live websites need real user data, not lighthouse synthetic tests for optimization decisions - Evidence: "If you have a live website with real users you want to look at your real user performance" #### ⏱️ [18:30](https://youtu.be/1Vg7CycTdWU?t=1110) — [H1] Decision matrix for optimal LCP - Key idea: Start with 2.5-second baseline, then benchmark against industry and competitors - Evidence: "You start with a 2.5 seconds baseline... then go and find out the industry average LCP" #### ⏱️ [22:10](https://youtu.be/1Vg7CycTdWU?t=1330) — [H2] Users compare to last visited site - Key idea: Performance expectations set by competitor sites, not Google thresholds - Evidence: "Your users aren't comparing you to the Google thresholds. They're comparing it to the last site they visited" #### ⏱️ [25:45](https://youtu.be/1Vg7CycTdWU?t=1545) — [H1] LCP correlation with business metrics - Key idea: Real user data shows direct correlation between LCP improvements and engagement/conversions - Evidence: Customer data showing "page views per session are improving with the LCP" and "checkout rate upside" #### ⏱️ [30:20](https://youtu.be/1Vg7CycTdWU?t=1820) — [H3] Modern technologies prioritize developer productivity - Key idea: New web technologies optimize for build ease, not visitor performance - Evidence: "Modern technologies do not necessarily mean faster LCP... the motivation number one of every modern technology is to be easy to build" #### ⏱️ [33:15](https://youtu.be/1Vg7CycTdWU?t=1995) — [H2] Performance as business question not technical - Key idea: Performance optimization should be owned by business/marketing teams, not just engineering - Evidence: "We're changing the question from technical question to business question" #### ⏱️ [35:40](https://youtu.be/1Vg7CycTdWU?t=2140) — [H3] User-level optimization opportunities - Key idea: AI and user-level monitoring enable personalized performance optimizations - Evidence: "User level optimization monitoring and AI on top of them" for finding optimization opportunities #### ⏱️ [38:25](https://youtu.be/1Vg7CycTdWU?t=2305) — [H2] Happy users drive business success - Key idea: Fast performance creates subconscious positive associations that increase return visits - Evidence: "She subconsciously remembers it just worked... this all piles up together and fundamentally results into higher engagement" ## 📚 References - [UXifi](https://uxifi.com) - [Core Web Vitals](https://web.dev/vitals/) - [Google PageSpeed Insights](https://pagespeed.web.dev/) - [CrUX Dashboard](https://developers.google.com/web/tools/chrome-user-experience-report) ## 🗣️ Paraphrase - Google's 2.5-second LCP threshold serves as a minimum baseline, but competitive advantage requires faster, user-specific performance targets based on industry benchmarks and real user behavior analysis - Performance optimization effectiveness should be measured through business metrics like engagement and conversions rather than just technical scores from synthetic testing tools - Modern web development frameworks prioritize developer experience over visitor performance, requiring conscious performance-first architectural decisions to achieve optimal user experience ## 📰 Newsletter **Performance Beyond "Good Enough": Why Your WordPress Site Needs Competitive Speed** Web performance expert Georgie Petra shares eye-opening insights about why Google's Core Web Vitals thresholds are just the starting point for serious performance optimization. As more sites achieve the 2.5-second LCP baseline, competitive advantage comes from understanding your specific users' patience levels and industry expectations. Key insight: Your visitors compare your site speed to competitors they just visited, not to Google's generic thresholds. Real user monitoring data reveals direct correlations between LCP improvements and business metrics like engagement and conversions. **Practical takeaways:** - Use real user monitoring for live sites, synthetic testing for staging - Benchmark against industry averages and direct competitors - Correlate performance metrics with conversion data to find your optimal speed target - Treat performance as a growth lever owned by business teams, not just engineering The WordPress ecosystem offers powerful optimization tools, but success comes from data-driven targeting rather than generic speed improvements. Resources: [Learn WordPress](https://learn.wordpress.org/), [Developer Docs](https://developer.wordpress.org/), [WP-CLI](https://wp-cli.org/) ## ✅ Action Items #### WordPress - Core Performance: Integrate industry-specific LCP benchmarks into performance recommendations - Documentation: Create guides for correlating Core Web Vitals with business metrics - Community: Develop case studies showing performance-to-conversion correlations across industries #### DevRel - Developer education: Workshop on real user monitoring vs synthetic testing decision matrix - Content strategy: Performance optimization guides focused on business outcomes rather than technical scores - Community engagement: Performance optimization challenges with business metric tracking #### Community - Contributors: Performance testing methodologies that account for user context and industry standards - Maintainers: Tools for correlating performance improvements with engagement metrics ## 🧪 Hypotheses - Sites optimizing below industry-average LCP will show measurably higher engagement than those meeting only Google thresholds - Real user monitoring data reveals performance-conversion correlations that synthetic testing cannot detect - Performance optimization ROI increases when treated as business growth initiative rather than technical maintenance - Industry-specific LCP targets provide better competitive advantage than generic 2.5-second thresholds ### 📊 Methodology - A/B testing: Compare conversion rates across different LCP performance tiers within same industry - Cohort analysis: Track user behavior patterns across performance improvement rollouts - Correlation studies: Analyze relationship between competitor performance benchmarks and market share ### ❓ Questions - How do performance expectations vary across different WordPress site categories (e-commerce, SaaS, content, etc.)? - What performance metrics beyond LCP most strongly correlate with WordPress-specific business outcomes? - How can WordPress hosting providers use industry benchmarks to differentiate their performance offerings? ### 🔮 Future Research - Industry-specific Core Web Vitals benchmarking database for WordPress sites - AI-powered performance optimization that considers user context and business metrics - Integration between WordPress analytics and real user monitoring for automatic performance-business correlations ### 💡 Implications - Practice: Performance optimization teams need access to business metrics, not just technical measurements - Policy: WordPress performance guidelines should account for industry context rather than universal thresholds - Education: Performance training should emphasize business impact measurement alongside technical optimization techniques - Funding: Performance improvement projects justify investment through conversion impact rather than just speed scores ### 🧠 Learning Styles Assessment Reading level estimate: 12th grade #### 👀 Visual learners - Present: Performance correlation graphs, decision matrix flowcharts - Improve: Interactive LCP calculator with industry comparisons, visual before/after case studies #### 🔊 Aural learners - Present: Video presentation with clear narration - Improve: Podcast-format discussion of performance optimization strategies #### 🗣️ Verbal learners - Present: Clear definitions of LCP, RUM, synthetic testing terminology - Improve: Performance optimization glossary, analogy-based explanations #### 🧩 Kinesthetic learners - Present: Step-by-step decision matrix implementation - Improve: Hands-on performance audit worksheets, interactive optimization checklist #### 🧠 Logical learners - Present: Data-driven decision framework, correlation analysis - Improve: ROI calculation templates, performance optimization decision trees #### 🧑‍🤝‍🧑 Social learners - Present: Q&A session, industry discussion context - Improve: Performance optimization community challenges, peer benchmarking groups #### 🧘‍♀️ Solitary learners - Present: Individual analysis framework, self-assessment tools - Improve: Personal performance audit templates, reflective optimization planning guides ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Clear headings structure, meaningful slide titles - Improve: Alt text for performance graphs and charts #### Contrast and readability - Present: High contrast presentation slides - Improve: Ensure data visualization colors meet WCAG standards #### Media and captions - Present: Video with clear audio - Improve: Detailed captions for technical terminology, transcript availability ### 🌍 DEIB Assessment #### Cultural breadth - Present: Performance considerations across different network conditions globally - Improve: Examples from diverse geographic markets, multilingual performance considerations #### Inclusive language - Present: Technical terms clearly defined - Improve: Plain language alternatives for complex performance concepts #### Neurodiversity support - Present: Structured presentation format - Improve: Visual summaries of key points, consistent navigation ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Strong logical progression from user expectations to business outcomes - Inductive strength: Good use of customer data examples to support broader claims - Causal reasoning: Clear connection between performance improvements and business metrics #### Fallacies - Appeal to authority: Relies heavily on Google's Core Web Vitals without questioning limitations - Hasty generalization: Limited customer examples presented as universal patterns #### Tone of voice - Overall tone: Authoritative but practical, balancing technical expertise with business focus ### 🛡️ DARVO Analysis - No concerning patterns of denial, attack, or victim/offender reversal detected - Presentation maintains professional, educational tone throughout ## 🎯 Conclusion The key insight from Petra's presentation is that "good enough" performance defined by universal thresholds like Google's 2.5-second LCP is insufficient for competitive advantage. Success requires understanding your specific users' patience levels through industry benchmarking, competitor analysis, and most importantly, correlating real user monitoring data with business outcomes like engagement and conversions. This shifts performance optimization from a technical maintenance task to a strategic business growth initiative that requires data-driven targeting rather than generic speed improvements. ## 🧱 Refactor Appendix ### 📚 Literature Split Pad Purpose: Stage CEE units that become Literature notes. Split level: H3 ### [H1] User performance expectations vary by industry context > [!quote] Why Good Enough Performance Is No Longer Enough by Georgie Petra via YouTube. Published 2024. Accessed 2025-01-03. > "User patience is different across industries, across websites, different sometimes even within the same website across time." Performance optimization requires industry-specific benchmarking rather than universal thresholds. [view highlight](https://youtu.be/1Vg7CycTdWU?t=735#H1) #xref #articleseed ID:rw_performance_context ### [H2] Real user monitoring reveals business correlations > [!quote] Why Good Enough Performance Is No Longer Enough > "You can look how the page views per session are improving with the LCP... you get more checkouts with the improved similar correlation." Customer data demonstrates direct relationship between LCP improvements and conversion metrics. [view highlight](https://youtu.be/1Vg7CycTdWU?t=1545#H2) #xref ID:rw_performance_business --- ### 🔖 Atomic Split Pad (Zettel Candidates) Purpose: Promote reusable claims into Atomic notes. Split level: H4 #### [Z1] Users compare site speed to competitors not thresholds - Applies when users research across multiple similar sites - Pointer to 22:10 timestamp competitive analysis section --- ## Full Transcript [00:04] Hey everyone, what's up? And we are back [00:07] uh after that amazing session from from [00:10] Shane. I think it was very insightful. [00:12] Uh and we're back for another session. [00:15] Uh but before we begin, I would like to [00:17] thank all our sponsors, all our all our [00:19] partners who have partnered up with us [00:21] to make this event possible. I would [00:23] also like to thank all our guest [00:25] speakers who have been with us up till [00:26] now and who will be with us in in the [00:28] next few sessions as well. I would like [00:30] you all to sort of show some love to our [00:31] speakers and the partners using the [00:33] emojis. Let us know in the comment [00:34] section how you're liking the event. If [00:36] you have any feedback for us, please let [00:38] us know because obviously we want to [00:40] become better. We want to do better [00:41] events for you guys. So make sure that [00:43] you drop your feedback in the comment [00:44] section for us as well. Now we have a [00:47] very interesting session for you all. Uh [00:49] it's called why good enough performance [00:51] is no longer enough and it will be [00:54] presented by Georgie Petra who I would [00:56] like to call up on stage. Now you might [00:58] have heard that name in the past [01:00] associated with different performance [01:02] tools uh with Nitropack maybe with other [01:06] performance tools as well. Uh but as [01:08] soon as Georgie is on stage I would like [01:11] himself to introduce himself to the [01:13] audience. Uh, so if I can have Georgie [01:16] on stage real quick. [01:22] Yeah, Georgie. Georgie is here. Georgie, [01:24] if you can really quickly introduce [01:27] yourself and let us know uh what your [01:30] background is and uh what do you do [01:31] right now? What do you do to keep [01:33] yourself busy uh and then after that you [01:35] can just start off with the session [01:36] without uh any delay. [01:39] >> Thank you so much. Thanks for uh the [01:42] invite and super happy to be here. Uh so [01:45] basically like I'm involved in the [01:47] performance industry since 2013 when we [01:51] started the first version of nitro pack. [01:53] It ultimately grew into like a popular [01:57] initially s then WordPress focused [01:59] solution and now serving a big chunk of [02:02] the of the active internet. Um and [02:06] currently I'm involved with a company [02:07] called UXifi which co-ounded together [02:10] with my partner MO and uh we're focused [02:12] on user user level optimization [02:16] monitoring and uh AI uh on top of them. [02:19] Yeah, [02:20] >> thank you for that intro Georgie. And uh [02:22] for everyone else, UXifi is also our [02:25] partner for this event. So thank you so [02:26] much Georgie and UXifi for for for being [02:29] a partner for the for this event. So [02:31] Georgie your session is about you know [02:33] uh performance is not is not enough good [02:36] enough performance is no longer enough. [02:38] Can you tell us a bit about what that [02:39] session is and what will be discussed in [02:41] this session? [02:43] >> Right. Uh so [02:46] the core of the session is uh thinking [02:48] of per performance uh through the very [02:51] practical meaning of why performance [02:54] matters and then identifying what is the [02:58] good performance for you and why [03:01] actually it's much smarter if you have a [03:04] website which uh is has a business [03:07] function like it can be e-commerce can [03:08] be SAS can be even hosting like why when [03:11] we have this website you should look at [03:13] performance performance from a growth [03:16] perspective, from marketing perspective [03:17] rather than from technical perspective [03:19] and how to correlate that connected to [03:22] the business outcomes. And uh this [03:25] basically answers the question why good [03:27] is not good enough. uh because most of [03:29] the times what we're seeing with our [03:31] customers is like we're seeing how [03:34] actually they're striving to go some [03:37] crazy numbers of low lowc [03:41] because they're being out competed and [03:43] they see how their users are reacting on [03:46] this LCP. [03:48] >> Yeah, that sounds amazing. Georgie, and [03:51] I won't let our audience uh wait more. [03:54] So, the floor is yours, Georgie. You can [03:55] start off with your presentation. [03:57] All right. [04:00] Share my screen. [04:08] All right. So that's my deck. Um, okay. [04:12] Let's start with a little bit of history [04:14] and actually understanding why we are [04:17] here today of in 2026 2026 discussing [04:22] steel performance when actually it has [04:24] always been the topic uh one of the big [04:26] topics and web um so I will just briefly [04:30] rewind till 2018 uh for like the members [04:35] who were doing performance at that time [04:37] please share identify yourself and share [04:40] a funny story [04:43] I can just share that at 2018 there was [04:45] no co- vitals there was no LCP and uh [04:48] there was no anything that actually [04:51] mattered for real understanding the real [04:53] user that there was nothing out there [04:56] and uh in 2018 uh there was this um [05:01] saying propagated like basically there [05:03] was a common industry uh knowledge where [05:06] tools like Pingdom like Kissmetrics or [05:10] like a few case studies here and there [05:13] saying that if your website is 3 seconds [05:16] like or lower that's super fast website [05:19] basically 3 seconds was like the gold [05:21] standard where everyone was striving for [05:24] um that was at the time where basically [05:27] the user patience was kind of um at that [05:31] time was like the lowest it could have [05:34] ever been but still it was cons like [05:36] from the perspective of today was like a [05:39] very high. And um at that time also h [05:45] there was no a like a singular like [05:49] uniform way to measure performance and [05:52] uh like some people were doing uh page [05:55] load together with all the JavaScripts [05:57] and everything that is loaded together [05:59] page load time other people were doing [06:02] content loaded and so so it was very [06:04] very basically a dispersed effort across [06:07] different people and uh and that's why [06:10] also So generally the the understanding [06:13] of performance was not so high among [06:15] website owners. And then luckily in 2020 [06:19] we had the launch of core web vitals [06:23] which I think a year later they went [06:24] into the their page P2 [06:27] and uh that solved a couple of issues. [06:31] So the first one is like uh a new [06:33] standard appeared which was an AOCP and [06:36] there was like the metric of largest [06:38] paid. I'll have it on the next slide to [06:40] explain to the new folks. But basically, [06:42] we introduced an AOCP where [06:46] this AOCP was considered to be the most [06:51] reliable way to measure the perceived [06:54] loading for any website. And why [06:56] perceived loading is important? Because [06:59] it's designed to be for the users, [07:01] right? So when we optimize, we want to [07:03] optimize for the users. and perceived [07:05] loading was where everyone was striving [07:08] for and this uh in this initiative and [07:12] uh 2.5 seconds uh was chosen but another [07:16] interesting thing that happened was like [07:17] it was not an average it was not a [07:19] median it was P75 [07:22] which means that it basically measures [07:27] uh the 75% of the fastest user page [07:32] views uh And [07:35] the next what happened after this launch [07:38] was is just history like it completely [07:41] changed the performance industry and uh [07:43] and then basically everybody started to [07:46] understand the significance Google [07:49] highly propagated it through their [07:50] efforts perform and huge performance [07:53] efforts across across uh the industry uh [07:57] still they're still ongoing and um yeah [08:00] and u moving up to 2026 We still use [08:04] heavily these thresholds in the [08:06] standards which is um which is great. Uh [08:12] right [08:16] yeah that's right GT GT metrics I'm [08:18] seeing just some of the some of the [08:20] comments and the questions like uh yeah [08:22] you can run metrics directly. Uh the [08:24] thing is that um so basically that [08:27] timemetric was also popular and of [08:30] course this was like even one of the [08:34] best metrics tools because it was like [08:36] very very powerful and running like good [08:40] so-called synthetic tests with today's [08:43] terms. [08:44] uh and yeah uh even if you people who [08:47] remember from that time uh there was a [08:49] popular technique because I heard the [08:51] previous topic one of the popular [08:53] techniques was actually putting your [08:54] images in sprites right creating one big [08:58] image and then moving it with CSS. All [09:01] right. So going back to today, um today [09:04] we have a very interesting uh [09:07] opportunity and situation to actually [09:09] start thinking about performance from a [09:11] completely new perspective and this is [09:14] what we have been doing um in UXifi uh [09:17] recently like basically uh our whole [09:21] whole value propositions is around user [09:24] the user and focusing about how the user [09:27] reacts [09:28] uh on performance. performance and other [09:32] type of frictions because right [09:34] performance is by definition like the [09:37] way it's organized currently is like by [09:38] definition negative metrics but there [09:41] are also other negative metrics which [09:43] are also reuser metrics and they are not [09:45] directly performance related but there [09:47] are still frictions. So moving to 2026 [09:50] uh we have uh still this uh general good [09:55] threshold by core web revitals of 2.5 [09:57] seconds but what changed in the last six [10:00] years. So one of the things that changed [10:03] is that [10:05] more and more websites [10:08] uh started to get under the 2.5 seconds [10:11] threshold which is great amazing amazing [10:16] but also something else changed and I [10:18] will share with you what changed in in [10:20] the next slide. Uh but let me tell you [10:22] for the folks who are just joining uh [10:24] what is LCP? LCP is currently the most [10:27] popular page world performance metric [10:30] which is in use by website owners, [10:34] agencies, tools like basically everyone [10:37] uses AOCP to measure how fast a page [10:41] loads and in very simple terms uh AOCP [10:46] directly calculates the time and shows [10:49] you the time of load of the largest [10:52] element on the page. This could be the [10:53] hero image, the headline or like [10:56] anything that is large enough to be [10:58] insignificant enough on the page that [11:00] can be considered an LCP. And this to [11:03] some extent has like in our in our own [11:05] studies it has about almost 80% of [11:10] direct correlation to perceive load. [11:14] Still there are 20 something% of LCP [11:17] which does not directly correlate to [11:18] perceive load. So think of like pages [11:21] where they have multiple elements like [11:23] think of a grid where every image is the [11:25] same and uh it's like a big grid very [11:28] often it will select the first element. [11:30] So LCP has its own problem but in 80% of [11:33] the times it's correct right to [11:36] understand the perceived load and if it [11:40] is correct to understand the perceived [11:41] load which means that we can measure the [11:45] user patience with it but now there is a [11:48] new challenge which is the following. [11:52] User patience is different across [11:54] industries, [11:56] AC across websites. [11:59] Uh it's uh [12:02] different [12:04] sometimes even within the same website [12:08] across time. So [12:12] depending of how you market the website [12:16] and who actually is entering your [12:18] website from what type of device, what [12:20] is their previous experience like [12:23] one of the big contributors of changing [12:26] of the user perception of how fast the [12:27] website is is actually how fast is also [12:30] your internet connection because if [12:32] you're user browsing for a very fast [12:34] internet connection, you tend to expect [12:36] every website to be fast, right? And uh [12:39] that sees uh everything piling up into [12:44] into different different uh uh cases [12:47] where fast website actually means [12:50] different for every website for every [12:53] audience. [12:55] But right still everyone of in every [12:58] industry in every website will still use [13:00] 2.5 right and um that's a a very uh good [13:05] start but I will share a bit more [13:09] what's the patience of your users is the [13:11] biggest question here uh and uh how to [13:14] find it out so in the next slide [13:17] basically I will walk you through a [13:18] decision matrix which we use um uh for [13:22] our uh in our product and uh for our [13:26] customers. Uh which we have proven to be [13:29] um [13:32] good enough uh to uh find uh [13:37] opportunities for increasing like [13:40] engagement, increasing revenue and uh [13:42] finding where exactly [13:45] performance performance improvements [13:48] matter. Uh so before starting we need to [13:52] measure [13:55] performance, right? Um and uh there is [13:59] very often this um case of uh especially [14:02] from the smaller sites and just joining [14:05] and u agents managing small sites like [14:09] very often a lighthouse uh which is like [14:12] you can take it in variations it can be [14:14] within it's free within Google dev tools [14:17] and um it can be multiple tools on the [14:20] internet that do the so-called uh [14:22] synthetic test which is basically [14:24] lighthouse which is amazing uh I mean [14:27] nothing bad about it. Uh but it has one [14:30] main use case. First is to be used on a [14:32] staging website and the second main use [14:36] case which probably secondary is to use [14:38] still you can use it on a production [14:40] website but only for the purpose of [14:43] finding different deviations from like [14:46] anomalies and changes. Now the big point [14:49] is like if you have a live website with [14:52] real users you want to look at your real [14:56] user performance right so it's basically [14:59] the same LCP but the way the way that is [15:03] being experienced by the real users and [15:07] the rest of my topic will focus on the [15:09] real user uh experience. So advice here [15:13] is you have a staging website go with [15:16] lighthouse. You have a live website go [15:18] with reuser monitoring tool. That's it. [15:21] That's the advice we go with. Okay. The [15:25] first thing with decision matrix like [15:26] how fast how fast I should go with my [15:30] LCP like what is a good LCP for me. [15:34] the very start you start with a 2.5 [15:36] seconds baseline right that's the good [15:39] enough performance of which Google [15:42] suggests if you're passing it you [15:43] basically moves you into the passing cor [15:46] vitals given that the other correct [15:47] vitals are fine as well this is the [15:49] floor like this is where you start with [15:51] if you're on more than 2.5 seconds [15:54] that's easy you have work to do I mean [15:56] it's easy to understand where you need [15:57] to be heading you have work to do get [15:59] under 2.5 seconds this is uh the first [16:04] requirement that you need to cover in [16:06] terms of LCP. [16:08] Now it gets more interesting. Of course, [16:10] this is the uh the like you can use many [16:13] tools but like using the uh crux data. [16:17] This is the Google uh page speed tool [16:19] where basically the first part of it is [16:21] the important part where it shows you [16:25] what is the LCP and LMP and COS of your [16:28] real users as measured from Google [16:31] Chrome data. [16:32] Uh right. [16:38] Okay. Um so how fast I should go with [16:42] LCP? Number two, go and find out the [16:46] industry average LCP. Like you're in [16:50] e-commerce, fine. You you're in the US [16:54] selling in the US, great. You're sell [16:56] you're selling uh like health products [16:59] fine great you are in this industry [17:02] you're competing with it go and find [17:05] this industry average LCP like there are [17:08] tools the real user monitoring tools who [17:10] offer this information already where [17:13] also have this information you go and [17:15] find like if 2.5 seconds is for every [17:19] website including the ones that are like [17:22] not really relevant to [17:26] go and basically understand what is the [17:28] your industry LCP and uh it's a game of [17:32] competition. So knowing where you stand [17:34] with and what is the expectation for [17:36] your industry is the first thing we need [17:39] to know after the 2.5 seconds. more and [17:42] more industries which are like uh more [17:47] competitive [17:49] are also very competitive on LCP. And if [17:52] you're in a very competitive industry, [17:54] the chances are that your LCP industry [17:57] average LCP is very much lower than 2.5 [18:02] seconds. Uh yeah, right. Um that's an [18:07] example of what a comparison to the [18:09] industry would look like. Um but it's [18:11] from our tool. You can get it from any [18:13] tool of course. Um and uh [18:18] that's on the industry. Now we move down [18:22] the funnel and now we look at specific [18:25] competitors. So specific competitors are [18:28] the ones like not in general all [18:31] competitors not necessarily the one that [18:34] you share some customers with although [18:36] it's still super relevant but the most [18:39] relevant group of competitors are the [18:42] one who you share uh traffic with right [18:46] so this will be the ones that people [18:48] will go and click like on search or on [18:51] like loom uh search they will find you [18:54] they will click a few other people and [18:56] they will start browsing with you. So [18:58] these are the ones who are sharing [19:00] traffic with and um they're the most [19:03] relevant uh competitors. Um why it's [19:07] important to look generally in [19:08] competitors and in industry is because [19:11] your users aren't comparing you to the [19:14] Google thresholds. They're comparing it [19:16] to the last site they visited and that's [19:19] usually if they're researching now they [19:22] are usually a site of a competitor or [19:24] another site that might be faster [19:26] enough. So you should be focused on that [19:30] side. That's a tip like one example how [19:34] a competitor comparison might look like [19:36] in your favorite uh run tool. Um and you [19:41] can compare how you stand against them, [19:44] right? And now we get to the interesting [19:46] part that's finding your user LCP [19:52] sensitivity. [19:54] So um [19:57] that's something that [20:00] is measurable but is not a super [20:06] straightforward [20:08] in with the current tooling. Right? Uh [20:11] why it's not super straightforward? [20:12] because it involves multiple data points [20:15] and uh it involves like uh platform [20:20] specific knowledge where most of the [20:22] tooling are platform independent and so [20:25] on like platform agnostic and uh [20:29] in these cases uh like there are already [20:33] some correlations available on the [20:36] market but in these cases like that's [20:38] the most important way to look at LCP if [20:41] you look at it from practical [20:42] standpoint, from the business standpoint [20:44] of what is a good LCP and how fast it [20:46] should get. And I'll share with you a [20:48] few examples from a customer of ours I [20:52] just pulled before the presentation [20:55] um and uh show you how [20:59] this uh user sensitivity on AOCP on your [21:03] website actually makes the needle and [21:05] how it completely switches the question [21:07] moves the needle. Yeah. Uh so first of [21:10] let's start with um engagement and [21:13] conversion. So [21:16] when you start uh looking at your own [21:20] real user data and how your users react [21:23] uh for you will start seeing graphs like [21:26] this like you can look how the page [21:29] views per session are are basically [21:33] uh improving with the LCP right that's [21:38] first engagement signal what should be [21:40] your LCP second engagement like you can [21:43] look at how the LCP by pre- conversion [21:45] for in this case by add to cardart how [21:47] basically you get like [21:50] auto to cards more to cards with the [21:52] better right and uh then you have see [21:56] how basically you get more checkouts [21:58] with the improved similar correlation [22:01] here like checkout rate upside that's [22:06] you see how checkout rate [22:10] is different at different ALCP levels [22:16] like we are seeing here a data of [22:19] a like I guess I think about 2 million [22:22] users and um that's a a very good [22:28] upsight of what actually is achievable [22:32] as a correlation of um different LCP [22:37] and um when you start monitoring these [22:40] type of situations you tend to [22:42] understand the potential of looking and [22:46] improving your LCP, right? Um and one [22:50] bonus benchmark which you can look and [22:53] we also following very closely with [22:55] different uh fractions and like [22:59] performance metrics is how [23:03] you are signaling [23:06] different engagement uh like for example [23:09] how you're signaling dwell time uh for [23:13] your purposes with LCP and here you see [23:17] the distribution ution of LCP again by [23:19] second and dwell time and the [23:21] opportunity size there. And here also [23:24] you see how [23:26] your competitive landscape look like in [23:30] terms of LCP and who you should uh be [23:33] outranking [23:35] and these are all suggested like [23:37] something I highly suggest for you to [23:40] follow if you're optimizing for the [23:42] bottom line your website and if you want [23:43] to use the LCP and other performance [23:46] metrics at as a growth uh levers [23:50] compared to just for technical go [23:54] like quality of life optimization and [23:56] ultimately you should end up with [23:58] something that is target LCP and uh [24:01] that's where your target is. [24:04] So basically that's the decision matrix [24:07] of how to think of LCP from the [24:10] practical standpoint of improving your [24:13] bottom line. Uh but uh [24:17] and and yeah, ultimately your goal is to [24:19] find the ideal SP. Uh once you find it, [24:23] it's very often much lower. Okay, [24:27] sometimes less lower, but it's [24:29] definitely lower than than 2.5 seconds, [24:31] the suggested ones. And that's why how [24:34] fast is fast enough [24:37] and why good performance is not good [24:40] enough nowadays is um it's a it's a it's [24:44] a real question uh for uh decision [24:47] making and for businesses [24:50] um right and I'll give you a brief also [24:53] matrix which is rather suggestions say [24:56] we're giving you also something we also [24:59] have as an optimization tool um So how [25:03] do I get to my ideal LCP? So very often [25:07] like mid-market brands and um upper SMBs [25:11] they would be they will have someone in [25:13] house who is in charge of performance [25:16] might not necessarily be a dedicated [25:18] role but it could be [25:20] like in in the web team or marketing [25:23] team who is one of the tasks he has is [25:27] to perform is to deal with the [25:29] performance. So usually this type of [25:32] person is the first go-to person to [25:34] actually go and try to uh talk and see [25:38] what we can do about performance once we [25:41] see the real user sensitivity and the [25:44] competitors and the industry [25:46] benchmarking. [25:48] Um this data is datadriven approach is [25:52] um is uh useful for performance engineer [25:56] or the person task to know where exactly [25:59] and how exactly to look at the most of [26:01] the like in UX5 for example you have a [26:04] drill down with 100 plus segments where [26:07] you can drill down and actually identify [26:10] the groups and uh have the AI agent on [26:12] top where you basically can find where [26:15] exactly is the issue uh with the SP P [26:18] and um [26:20] like that's the go-to strategy. This [26:22] engineer sometimes might use like [26:26] optimization tooling [26:28] and uh like for example like navigation [26:31] AI of um [26:33] nitrop our previous components like [26:36] amazing optimization tooling for [26:37] WordPress pro another optimization [26:40] tooling for WordPress. Qualflare have [26:42] bunch of uh rather generic type but [26:45] still very reliable [26:48] optimizations. Um and um yeah that's all [26:51] piles up together. Still it's it's a [26:54] good strategy to use someone who is [26:56] aware of what needs to be done like [27:01] rather than going in applying multiple [27:03] tools at once. It's really gives the [27:06] best results. [27:08] Um and uh yeah and uh nowadays do we [27:12] have a lot of AI agents who uh tend to [27:15] more and more understand the code. We [27:19] for example use a lot of AI in order to [27:22] propose some meaningful fixes and to [27:25] find hard to find situations where like [27:28] would be [27:30] require a performance engineer. some of [27:32] his tasks are actually taken by this AI [27:34] agent identifying uh weak points in [27:39] performance based on u based on the real [27:43] user data. Um and uh for those of you [27:48] who are just uh hearing the real user [27:51] data and situation and wondering why do [27:55] I need that when I just can run a test [27:58] and tell me the score. [28:00] Well, the difference is big. If you're [28:02] running a like a live website, it's a [28:05] huge difference. And the main difference [28:07] is that it's not only the website [28:10] itself, the moving piece of the puzzle. [28:13] There's a lot of moving pieces with [28:15] network conditions with uh VPN [28:18] restrictions like with um like device [28:21] power [28:23] internets on and on and off for the [28:25] users like going hopping h c h c h c h c [28:26] h c h c h c h c h c h c h c h c h c h c [28:26] h c h c h c h c h c h copy on and off [28:27] different locations and all this is [28:30] actually resulting in the real user [28:33] experience knowingly or knowingly by the [28:35] user and you can do more like uh for [28:39] example there's already certain like [28:42] very strong motions going in the user [28:44] level optimiz optimization phase where [28:46] we are involved and uh basically keeping [28:49] the user closer to you regarding on the [28:51] environment. Um and um yeah that makes a [28:55] big difference uh to actually optimize [28:58] for the real uh user. [29:02] Um and that basically focusing the real [29:05] user and focus on the business first [29:07] perspective of actually what does it [29:09] mean for my engagement for my user for [29:10] my website they're all aligned. We [29:12] ultimately with this focus we ultimately [29:16] achieve the same thing and it is making [29:18] your user happy right because if you're [29:22] performing if he's happy he she is happy [29:25] she will continue this journey the [29:27] button that is clicked immediately [29:30] responds with visually and uh then she [29:33] clicks she tends to click more and uh [29:36] then she tends to return at a higher [29:39] chance rate uh to your website because [29:42] she remembers that website was very like [29:44] subconsciously remembers it just worked. [29:47] Um, and this all piles up together and [29:50] it basically fundamentally uh results [29:53] into like uh higher engagement and more [29:56] conversions and it fundamentally change [29:58] the ownership of the performance task, [30:02] right? uh because like if the from [30:06] landing the performance helps the users [30:09] like UXY like the journey is how we call [30:12] it or basically improve the uh user [30:15] experience of the journey to and this [30:18] correlates directly to success more [30:20] successful journeys that's now a [30:22] business topic right that's now a [30:23] business question that's a management [30:25] question that's like how we are more [30:27] successful uh with that and that's might [30:30] be on the marketing that might be on [30:32] growth or any team that is tasked to [30:35] make this website work [30:38] when it works for the user it works for [30:39] the business as well. Um so [30:43] basically we're changing the question [30:45] from technical question to business [30:46] question and uh when correlating to [30:49] engagement conversions SEO right that's [30:52] all folks thanks so much if you have any [30:54] questions [30:56] let hope it was interesting [31:00] yeah [31:02] thank you so much Georgie this was this [31:04] was fantastic uh very insightful uh we [31:06] have some questions in the in the chat [31:08] as well that I would like you to answer [31:09] we have a few minutes uh and we can use [31:11] that to answer questions from the chat. [31:13] So, Veronica has a question and she asks [31:15] that how can we actually measure how LCP [31:18] influences [31:20] my revenue. So, I will pin that on the [31:22] screen so you can read it. It should be [31:24] on the screen now. [31:27] >> Can you see it? [31:28] >> No, I don't see it. [31:30] >> All right, let me just read it out. Uh, [31:32] so it says, "How can we actually measure [31:34] how LCP influences my revenue?" [31:39] >> All right. Um yeah so it's just [31:42] basically using the real user data um [31:45] and uh the user this user real user data [31:48] already has the answer right so we [31:50] already have the real user data you just [31:53] need to visualize it and that's you use [31:57] the re user monitoring tool that's the [31:59] easiest way just the user user [32:00] monitoring tool should support this uh [32:03] visualization [32:06] >> awesome I will now go to the Q&A [32:09] section and we have a few questions over [32:11] there as well. Let me scroll down uh to [32:14] see if we have any questions for your [32:17] uh for your session specifically. I saw [32:19] a few questions around LCP and I wanted [32:21] you to take those. Uh so Hamam has a [32:25] question. How can design uh systems no [32:29] uh so sorry Yashi has a question. How do [32:32] modern web technologies help developers [32:35] exceed good enough performance [32:36] standards? [32:39] Yeah. Um, so there is a [32:44] how to say modern technologies [32:48] do not necessarily mean faster LCP, [32:53] right? uh because the the motivation [32:56] number one of every modern technology is [32:58] to be easy to build which is slightly [33:01] different motivation than uh performance [33:04] for the visitor and uh it's it's been [33:09] like that for quite some time with all [33:10] the frameworks and now all with all the [33:12] vibe coding and things like this that is [33:14] going on and all the technologies that [33:15] you pile up together. So when you when [33:18] you pile them up together, you [33:20] definitely optimize for your [33:21] productivity to build this uh this um [33:25] website. Uh but performance is rather an [33:30] optional add-on to most to these [33:33] technologies rather than an something [33:36] that goes with them. So we've seen some [33:39] um uh modern like for example teams that [33:44] is like we have one or two teams which [33:46] they have put some constraints on their [33:49] buildability on their technology but [33:51] they have really produced a performance [33:54] first team. uh we've seen that and uh [33:59] but generally the modern technologies in [34:02] general then they I don't know have [34:04] performance as first priority in reality [34:07] although their my marketing might be [34:08] saying something different it's often [34:11] not the case like performance usually [34:13] goes in does not go together with easy [34:16] to build right [34:19] >> all right uh makes a lot of sense [34:20] Georgie we'll take the last question for [34:23] your session uh and then we will just [34:25] wrap it up the Question is Georgie. [34:27] Could the pressure to exceed good enough [34:29] performance lead to unsustainable [34:31] expectations for workers and [34:33] organizations and how should this be [34:35] managed? [34:39] Um [34:40] yeah so the first part of uh the [34:43] question is um [34:46] it might really be a bit tough uh but [34:51] it's usually like you need to understand [34:53] why we're doing it and what do you need. [34:55] So from there it's it it simplifies a [34:58] lot of a lot a lot of situations. So if [35:01] the teams who are building it they [35:03] consciously know that they need to [35:06] strive for that they willing to take [35:08] tough decisions like tough decisions [35:10] like we don't want that big video here [35:12] we'll do something else right we don't [35:14] want that big JavaScript which is here [35:16] we do something else and they in this [35:19] way when we when they think technology [35:22] first like from performance first [35:24] perspective they're able to to to take [35:27] uh the right decisions from the [35:30] beginning of what is possible and what [35:32] is not possible because too many of the [35:34] nice things that they want to do are [35:37] actually not beneficial for performance. [35:40] >> Yeah. [35:42] >> All right. Thank you so much Georgie. [35:44] This was it from this session. Uh folks [35:47] I would like to thank Georgie for being [35:49] here and uh presenting very insightful [35:52] presentation and a demo. I'm sure a lot [35:54] of people uh will take a lot from uh [35:57] from Georgie in this session around LCP [35:59] and and you know why it matters so much [36:02] to your website and to your client's [36:03] website based on what area of work you [36:06] are in. So thank you so much Georgie. [36:08] This was it from from this session. As [36:10] for you folks don't go anywhere because [36:12] right after this we have a rapid fire [36:16] quiz. So, if you missed out on uh on the [36:18] last activity, this might be your chance [36:21] of redemption. Uh and you might end up [36:24] winning some amazing gift cards. So, [36:26] don't go anywhere.  📄 **Full text:** ![[Full Document Contents/Articles/Why Good Enough Performance Is No Longer Enough.md|Why Good Enough Performance Is No Longer Enough by YouTube]] ## Highlights - Georgie your session is about you know uh performance is not is not enough good enough performance is no longer enough. Can you tell us a bit about what that session is and what will be discussed in this session? >> Right. Uh so the core of the session is uh thinking of per performance uh through the very practical meaning of why performance matters and then identifying what is the good performance for you and why ([View Highlight](https://read.readwise.io/read/01kpv0xmeerr3p0k51d9gkb3cq)) - Tags: 👻 ai highlighted - good enough uh to uh find uh opportunities for increasing like engagement, increasing revenue and uh finding where exactly performance performance improvements matter. Uh so before starting we need to measure performance, right? Um and uh there is very often this um case of uh especially ([View Highlight](https://read.readwise.io/read/01kpv0xmfr9z73s08k7p0n3x3s)) - Tags: 👻 ai highlighted - different motivation than uh performance for the visitor and uh it's it's been like that for quite some time with all the frameworks and now all with all the vibe coding and things like this that is going on and all the technologies that you pile up together. So when you when you pile them up together, you definitely optimize for your productivity to build this uh this um website. Uh but performance is rather an ([View Highlight](https://read.readwise.io/read/01kpv0xmkt9d2g7s5v1c03mpw8)) - Tags: 👻 ai highlighted