Frontend Optimization Handbook

Chapter 4. Building a Performance Culture in the Organization

“Ad augusta per angusta” (to high places by narrow roads)

Victor Hugo

This final chapter considers aspects of performance from a business management perspective. Although you may provide oversight, ultimately the details of monitoring and analysis will be performed by analysts and other IT professionals of one sort or another. Two areas remain where line-of-business managers can have a beneficial influence. One is tactical; the other strategic.

Tactically, if preliminary monitoring versus your key competitors indicates a significant performance deficit, or if you are in the happy position of being able to monetize performance (see Chapter 1) and such monetization indicates a major impact on digital revenue, then you may wish to invoke external assistance in terms of third-party performance-enhancement solutions. The strategic area is that of evolution of a performance culture within your organization. This is easier said than done, as any cultural change is a major undertaking. However, the real-world benefits for organizations adopting such a path are large. I will address this longer-term approach later in the chapter.

Note

Significant UK examples of an explicit performance orientation in both digital and broader cultural areas include:

• The Guardian and Financial Times in the media sector

• Laterooms in the travel sector

• Next in the retail sector

For the FT, reference “Performance management: five priorities for designing a system that aligns with IT article”. Each of these companies have experienced broad benefits,¹ some intangible, although The Guardian is currently website of the year, which is proudly displayed on their masthead, shown in Figure 4-1.

Figure 4-1. The fruits of performance orientation; two media-sector examples

Reverting briefly to tactical approaches: depending on circumstances (e.g., global distribution of customers), a number of performance interventions can be “bolted on” to an existing application infrastructure to enhance performance. They do work, although some are somewhat based on the principle that “in a hurricane, even a turkey can fly.” If the fundamentals of your application are poor, then such interventions (CDNs, etc.) can become very expensive. However, it is to be hoped that applying some of the principles outlined earlier will enable some optimization of the base situation. Differences obviously exist in functionality (for example, CDNs vary widely in certain characteristics such as regions served, points of presence per region, etc.) as well as price. That being the case, it is important to undertake a structured proof-of-concept trial prior to adoption.

I will now go on to consider building a performance culture in more detail.

Building a Performance Culture

End user monitoring and analysis are important and should pay dividends, but they are not ends in themselves. They must compete for resources with many contending priorities—OPEX control (e.g., headcount freezes), exotic marketing campaigns, and many others. All too often, they can lose out, despite the associated business risk from neglecting this area.

The most effective businesses manage to make digital performance part of their culture. I conclude with a few thoughts on this.

The Bricks-and-Mortar of Performance Culture: Some Practical Ideas

In the immortal words of Michael Brunton-Spall when opening the 2015 Scale Summit event in London: “Technology is easy; people are hard.”

Many people have observed three elements in achieving successful delivery of IT performance excellence: people, process, and tooling. The following are some thoughts on the people part of the triad that may assist others treading this path.

The Nature of the Problem

Every group has its culture, but in large organizations, culture often serves to counteract the requirements of a performance focus. Particular points that might resonate with readers are internal politics, siloed operations, and in particular, a conservative mindset whereby the business is resistant to any change in IT systems due to a perception that change equals risk. These features of most large organizations by virtue of their size are exacerbated in conditions of high growth and/or acquisition, with the imperative to subsume all outliers (acquired companies, small teams) into a common mothership mentality. This is often underpinned by a desire to impose the parent’s brand values, and to promote and impose their management across the empire. In this environment, small IT teams can feel trapped within a “don’t touch anything” mentality, with any change moderated through endless committees, and any concept of enterprise or initiative stifled.

Sound familiar? So what strategies are available for seeking change from within, rather than joining the mass exodus towards superficially greener pastures?

Requirements for Change

From the point of view of the IT teams responsible, to enable effective engagement in the delivery of application performance, the company culture has to engender four key factors, assuming that core skills around monitoring, unit, and performance load testing are available.

Assuming that it is supported by the wider culture, the bases of success are (what I term) the CEFO elements:

• Cooperation

• Empowerment

• Freedom of action (within agreed boundaries)

• Objectivity

These combine to provide the final component: visibility. Visibility is the tangible expression of performance-centricity throughout the organization. Such expression will typically have many forms but should be apparent to those outside the business (e.g., visitors or customers) through, for example, logo straplines or heads-up dashboards, as well as to employees and managers via balanced scorecards or incentive schemes.

Given that CEFO conditions will not magically arise, it is necessary for performance to become a central tenet of the overall company culture. The difficulty, supported by a mass of literature on the subject, is that changing a culture, particularly in large organizations, is extremely difficult. However, it is not impossible given the right conditions. The following are ideas gleaned from successful attempts.

Building Blocks of a Performant Culture

Interaction with companies (such as those referenced above, and others ploughing the same furrow) reveal several core building blocks for a successful cultural change outcome. They are listed here, though others may apply to your circumstances, such as managing disparate international cultures and/or business unit autonomy.

Culture is for life—not just for Christmas.

Less flippantly, cultural transformation is a serious, core, long-term undertaking. It is not, and can never be, a tactical exercise.

Get senior buy-in.

It is probably no exaggeration to state that cultural change can never be effected from the ground up. Senior buy-in (preferably ownership) is essential. To achieve this, consider:

• Presenting the problem (for example, unperformant applications) in terms relevant to the audience. Senior management are not interested in server throughput, but they are likely to respond to loss of market share or revenue, and concern about share value.

• Visualize the issue. Tools such as WebPageTest’s side-by-side videos are great at driving home a competitive deficit.

• Support with objective data, especially from end-user testing (international markets, native mobile applications, etc.). Trending is particularly effective when data is available, and helps make the point that this problem is not going away.

Build awareness across the company.

Performance projects with five-letter acronyms are probably too simplistic, as are anodyne “ten values” statements. However, if the company goals can be succinctly stated and promoted by Level 1 executives, then half the battle is won. Within these, make statements referenceable rather than “motherhood and apple pie.”

Make performance an explicit, visible goal.

IT can crystallize these as specific KPIs, which are particularly effective if delivered across heads-up dashboards in key areas. Within dashboards, RAG (red-amber-green) outputs or speedometer dials are likely to have the most impact, especially if associated with revenue (e.g., orders per hour versus site response).

Take a structured, process-based approach to adoption.

A useful idea suggested by one of the companies alluded to in recent sections (borrowed from the late, often-lamented Steven Covey) was to require each subordinate department/level within the organization to state how they would achieve the published Level 1 goals. These statements should incorporate SMART (specific, measurable, actionable, realistic, timely) objectives. Within IT, product owners can be key to the solution by ensuring that nonfunctional requirements are core to the specification, tested (and budgeted for) as part of an integrated continuous-integration DevOps process.

Reach out to the whole company.

As an example, suggestion-box applications have a long history, stretching back at least 25 years to Archie Norman’s “Tell Archie”-led transformation of ASDA, although there are many others. Some form of incentive for the submission of particularly useful ideas is beneficial, although prizes do not necessarily have to be tangible. Advocacy can be a reward in itself by associating an individual or team with business success.

Take care when crossing the boundaries of the organization.

It is important to avoid partnership-based clashes (in the jargon, acculturation). If possible (and it is always possible at some level), it is important to undertake some form of cultural screening of partners. This is a field where (at last) there are a few companies who will add value. However, seek to require objective (as opposed to narrative-based) comparison. The latter is more relevant to the descriptions of far-flung indigenous populations in a different era. Even without such assistance, the performance-centric company rapidly becomes attuned to disparities in other organizations. Seek to undertake informal due diligence specifically around working culture. Signs and artifacts (even matters as superficially trivial as tea-room culture) within potential partner organizations can reveal a lot.

Recruit and structure with intention.

One of the features of company cultures (be they beneficial or toxic in the eye of the beholder) is that they tend to be self-perpetuating, in that those who like/can cope with (or derive some personal advantage from) them stay, and others leave. This is one of the reasons why change in this area is so difficult. However, any organization has a degree of staff churn—retirements, changes in circumstances, etc. This can and should be harnessed as a change agent. Thus, in addition to the cultural memes adopted (and referenced elsewhere in this chapter), it is beneficial to:

• Recruit core people willing to explicitly embrace performance, and who can reference such an orientation at interview.

• Consider an entrepreneurial digital-development role (separate from core IT). Such a role will need to be publicly supported by Tier 1 management if it is not to be suppressed by “the blob”—the forces of the ancien regime, particularly those of departments such as IT, which tend to be resistant to such change.

• Consider change agents such as ITIL, which can have positive benefits such as changing “operations” to “service,” but beware the elephant trap of bureaucracy. Who will have ownership of such introductions?

• Every level in the organization should have ownership. This enables organic development, rather than a perception of imposition from above.

Above all, note that the real values of an organization are expressed through who gets promoted and explicitly rewarded—not on what lip service is paid.

Further Reading

For anyone interested in reading further about the subject, I recommend the following resources (selected from the vast academic literature):

Cultures Consequences, 2nd ed, by Geert Hofstede (Sage, 2003)

An academic study of some of the dimensions of corporate culture, derived from Hofstede’s several decades of work at IBM. Old, and based on national cultural dimensions, but classic.

In Great Company, 2nd edition, by Q. Jones et al. (Human Synergistics, 2011)

A practical guide, based around real-world case studies.

“The big debate: performance management”, published in The Guardian

Discusses performance management more broadly within the UK public sector. Although not concerned with digital performance, it contains references to some useful sources of information for those wishing to adopt a more holistic approach to performance throughout an organization.

Conclusion: Everything Changes, Everything Stays the Same

In conclusion, the rate of change in the variety of end-user applications continues to increase. Such change is inevitably associated with performance challenges for two reasons. One is that applications developed for a particular usage and technology environment (e.g., PC browser/standard HTML) continue to be used while behaviors, technologies, and devices change (e.g., progressive applications delivered to mobile devices). The other is that developments in applications themselves render tools and techniques of understanding redundant (consider Single Page Applications and synthetic monitoring, discussed in Chapter 2, as examples). Some aspects of new technologies may be black-boxed as far as monitoring and analysis is concerned—at least, if without access to core systems and code (e.g., Internet of Things).

However, in the clear majority of cases, every meaningful aspect of an end-user transaction will continue to involve visible changes to the GUI (Graphical User Interface; in simple parlance, the user screen). As such, end-user performance can continue to be monitored. Network interaction will still be required, even if it’s on a store-and-forward rather than real-time basis. Together, these provide the basis of frontend analysis and optimization.

Provided that there is access to developers and source code, outputting timing metrics via the range of APIs becoming available provides a robust, production-ready richness to the analysis of visitor performance experience. These can prompt intervention and provide a deep understanding of production performance. The caveats already expressed—monitoring availability, competitors, and performance in low-traffic situations (including pre-production)—still apply.

When presented with new challenges to effective practice, the cascade shown in Figure 4-2 should prove its worth.

Figure 4-2. Dealing with new developments: a cascade flowmap

Final Thoughts

I will leave you with a few final thoughts about FEO recommendations. My primary objective in writing this book was to link two superficially disparate worlds—business (revenues, stock price, market share) and IT (bits and bytes, development code). In fact, as I hope I have conveyed, digital channels to market is an area where these two orbits intersect. This provides a great opportunity to share understanding and harness the resources of the business to a common goal. In seeking to achieve these benefits, consider the following imperatives:

• Link technical priorities to business drivers such as competitive revenue exposure, etc. Seek objective understanding of deficits and advantages in terms of performance (e.g., transaction speed) and customer behavior (e.g., basket size, page bounce rate, time-on-site).

• Live in the real world. What can be changed at economic cost, in realistic timescales? Turbocharging an existing application to be the fastest in the market to all users all the time is unlikely to be fiscally prudent, even if it is achievable. All the engagements in which I have been involved where this was an initial stated aim have settled for a process of progressive iterative improvement and have reaped the rewards of so doing.

• Beware major effort for marginal improvement. It is always worth standing back from a given intervention and considering the big picture—in particular, what is the projected lifespan of the application or element in question? Can this change be more readily baked in to the next-generation solution?

• Seek to deliver a combination of immediate prioritized interventions and ongoing governance/management of objectives, and set iterative goals for improvement (unless in crisis mode). These can be supported by a combination of process change and documentation. For example, what are the existing controls on adding third-party affiliate tags or huge animated GIF images (Father Christmas is a common offender here!)?

• Suggest triggers for ongoing intervention based on a combination of direct (synthetic monitoring) and indirect (web analytics, RUM) alert flags. In short, many technologies are available, and all will give you some kind of answer. Try to use the information in this book and elsewhere as a guide to determine elegant monitoring techniques that provide maximal useable data for minimal overhead.

¹ Author, personal communication