A relationship that began with thirty UK and EU websites and became a fifty-plus site programme across Europe and Asia Pacific. This is what a decade of owning somebody else’s web estate actually looks like, including the parts that are not interesting.
Green Arrow Consultancy has run Energizer’s international digital estate since around 2014. The engagement started with more than thirty UK and EU websites, added sixteen Asia Pacific sites in 2017 to reach fifty-one or more, and covers development, digital strategy, multi-country campaigns, user testing, privacy compliance, accessibility and, more recently, production AI tools.
Energizer Holdings is a global consumer products business with a large family of brands. Batteries and portable power under Energizer, Varta, Rayovac and Eveready. Automotive care under Armor All, STP, AC Pro, Lexol, Eagle One, Driven, Refresh and California Scents. Each brand carries its own tone, its own product catalogue and, in most markets, its own website.
Around 2014 we took over more than thirty of those websites across the UK and the EU. Taking on an estate rather than a site is a particular kind of engagement. Estates accumulate: sites built by different agencies in different years on different platforms, with different content models, different analytics setups and different levels of maintenance. In takeovers of this kind some sites turn out to be in excellent shape, and others have not been touched since launch with no clear record of who approved the last change.
The first job on any takeover like this is not development. It is inventory. Which sites exist, which domains and subdomains resolve, what each one runs on, who requests changes to it, what third-party scripts it loads, where its forms post to, and which of them are actually still needed by a market team. That exercise produces the uncomfortable list, and the uncomfortable list is the real starting point.
Inventory, platform consolidation and the beginning of a single operating process across brands and markets.
Campaigns across France, Germany, Italy and the UK, and the point at which user behaviour studies and journey tracking entered the programme.
A promotion supporting a major film release, with social integration, running simultaneously across thirteen markets.
The managed total reaches 51 or more sites. Privacy work begins in earnest as GDPR approaches.
Across all clients, including Energizer family brands such as STP, Armor All and AC Pro.
Ten years later the shape of the work has changed several times. The constant is that somebody has to be responsible for the estate as a whole, and that responsibility does not fit neatly inside a brand team, a market team or a project.
Green Arrow Consulting has been a strategic partner in the development and maintenance of the Energizer International Digital platform, working with our regional offices in Asia, Middle East and Africa and Europe.
After a decade on this estate and others, the pattern is consistent enough to plan for on day one.
| The problem | What people assume | What it is really |
|---|---|---|
| Scale | Fifty sites is fifty times one site | Fifty sites is one site plus forty-nine exceptions, and the exceptions are where the work is |
| Consistency | A design system fixes it | A design system fixes the visual layer. Data layers, consent categories and form handling drift far faster than CSS does |
| Ownership | Every site has an owner | On any estate over about twenty sites, some do not. Finding the orphans is a standing task, not a one-off audit |
| Change requests | They come from one place | They come from brand teams, regional offices, legal, agencies and retail partners, often for the same page in the same week |
| Third-party scripts | Marketing adds a tag occasionally | Tags arrive without notice, load other tags, and are the single most common cause of both consent failures and performance regressions |
| Retirement | Old sites get switched off | Old sites stay up because nobody will sign the decision. Managed retirement, with redirects and archived content, has to be part of the process |
Observed across the estates this firm manages, which passed 200 websites in total by 2022.
Six strands. Only the first is what most people picture when they hear website agency.
New brand sites and market sites, plus migrations of inherited sites onto maintainable platforms. The test of a rebuild on an estate is not how it launches, it is whether the next twenty sites get cheaper.
One intake route for change requests, one release process, one inventory of record. The point of a shared process on a multi-region estate is that a market team can see where its request sits and when it will land.
Advising on platform choices, market rollout order, what to consolidate and what to retire. A recurring part of the job is arguing against building a new site when an existing one would do.
Introduced with the 2015 multi-country promotional work. Watching how people actually move through a product finder or a promotion entry flow, then changing the flow rather than arguing about it.
Structured comparison on the pages that matter, most often product finders, entry mechanics and purchase handoffs to retailers, run as a discipline rather than as a one-off experiment.
Privacy notices, cookie policies, consent configuration, accessibility conformance and the ongoing checking that keeps them true after the next content release.
In 2015 the programme took on multi-country promotional work across France, Germany, Italy and the UK. This is where user behaviour studies and journey tracking entered the account, because a promotion that runs in four countries produces four different drop-off patterns and no useful way to argue about them without data.
In 2016 a campaign supporting a major film release ran across thirteen European countries with social integration. Thirteen markets is where a promotional mechanic stops being a creative exercise and becomes an operational one.
Promotion and prize law varies by market, and so does the wording you are obliged to show before capturing an entrant's details. Age gating differs. Consent requirements differ. Prize fulfilment and its tax treatment differ. Translation is the easy part, and it is still not easy, because a mechanic that fits a button in English may not fit in German.
Then there is timing. Campaigns tied to a film release cannot slip. Thirteen markets means thirteen approval chains, and any one of them can be the reason a launch date moves. The engineering response is to build the mechanic once, parameterise everything that varies by market, and give each market a way to go live independently rather than as one synchronised release that any single delay can hold hostage.
Journey tracking and user behaviour studies were part of this work from 2015 onwards, so the campaigns were measured rather than assumed. The resulting figures belong to Energizer, and we do not publish client campaign performance. What we can say is what changed in the way the work was run. From 2015 onwards there was recorded behavioural evidence about how people moved through these promotional flows, which meant design decisions on this account could be settled by observation rather than by argument.
Privacy work began around 2017, as the General Data Protection Regulation approached. It has never stopped, and it changed what this firm is.
The mechanical problem is straightforward to describe and slow to solve. Every website in an estate loads scripts. Analytics, tag managers, media pixels, chat widgets, video embeds, social buttons, consent tools themselves, and whatever a regional team added for a campaign three years ago. Under GDPR and the UK GDPR, most of those cannot fire before a user has given consent, and the organisation has to be able to say which ones exist, what they do, what they set, and on what lawful basis.
On fifty-plus sites across many markets, the first step is discovery: crawl everything, catalogue every cookie and script, map it to a purpose category, and work out which market rules apply to which properties. Only then does configuration start. A consent management platform is the visible part, but the difficult part is tag governance: making sure that the tags a consent platform is supposed to block are actually routed through the mechanism that can block them, rather than hard-coded into a template somebody forgot about.
A site can be correctly configured in March and non-compliant by June because a market team added a media pixel for a campaign. So the compliance strand is a monitoring discipline: periodic re-scans, checks after significant releases, and a route for regional teams to get a tag approved quickly enough that they do not simply add it themselves. We wrote more about the general method in cookie consent management and privacy consulting.
Privacy also changed how the rest of the work is done. Once you have inventoried every script on an estate, you know exactly how much of a page's weight is third-party, which makes performance work concrete. Once you have mapped every form, you know where personal data actually goes, which makes analytics governance possible. The privacy exercise produces the map everything else runs on.
Darren Tyler has been doing work for me on Energizer’s website compliance. Not only has he met our teams expectations but exceeded them.
What accessibility means when the unit of work is an estate rather than a website. The general approach is described under accessibility.
Five production systems came out of this relationship. Generalised versions run as public demonstrations under Neuro, on sample data.
A customer describes or selects a device and the system returns the correct cell. Sounds trivial. It is a large, messy compatibility dataset plus a natural language front end that has to refuse rather than guess when the match is uncertain.
Comparison across a catalogue with real specifications, and a replacement path when a product has been superseded. The value is in the data work, not the model.
A business customer maps a facility zone by zone and receives a specified parts list. This one came from the industrial side of the account and became the template for the multimodal work we now do elsewhere.
An assistant grounded in the corporate code of conduct that answers employee questions with the relevant section cited, and escalates rather than improvising when a question needs a human.
Turns recurring reporting into a drafted narrative with the underlying figures attached, so that analysts spend their time on the interpretation rather than the assembly.
Every one is grounded in the client’s own data, cites what it used, and is allowed to say it does not know. That is a direct consequence of a decade of privacy work in the same account.
Five things that transfer to any organisation running more websites than it has people.
The most valuable artefact we produce on any estate is a current, accurate list of what exists and who owns it. Every other piece of work, privacy, accessibility, performance, retirement, consolidation, depends on it, and almost no organisation has one when we arrive.
Regional teams will accept a shared consent implementation, a shared data layer and shared components. They will route around a mandated global page. Standardise the things that carry risk and let markets own the things that carry nuance.
Both regress with every content release. Budget them as a recurring cost with a monitoring cadence, or accept that you will pay for the same remediation exercise every three years.
They cause consent failures, performance regressions and privacy exposure, and they arrive without warning. A tag approval route that is faster than adding a tag yourself is worth more than a policy.
After ten years you know which market team will call at the last minute, which integration breaks under load, and which requests are really about something else. That accumulated knowledge is most of what an incumbent is actually being paid for, and it is the reason we prefer operational retainers to projects.
Darren’s practical mind makes him really effective at cutting through complexity to find the most suitable solution.
More on how we scope and run estate work in website management.
The relationship began around 2014 with more than thirty UK and EU websites. In 2017 sixteen Asia Pacific websites were added, taking the managed total to fifty-one or more. The firm's wider portfolio across all clients passed two hundred managed websites by 2022, and includes brands in the Energizer family such as STP, Armor All and AC Pro. The exact live count moves constantly, because sites are launched, merged and retired as brands and markets change.
Energizer Holdings owns a large family of brands, and the ones we are permitted to name in connection with our work include Varta, Rayovac, Eveready, Armor All, STP, AC Pro, Lexol, Eagle One, Driven, Refresh and California Scents. Each brand carries its own tone, its own product data and frequently its own regional variants, which is the reason a fifty site estate is not simply fifty copies of one site.
Content changes and product data updates across brands and languages, platform and plugin patching, consent and tag configuration per region, accessibility fixes, broken integration triage, campaign landing pages, redirect management when a market restructures, and a running inventory of what exists. The last one is more valuable than it sounds. On estates this size, the single most common finding is a live website nobody in the organisation currently owns.
With regional offices directly. As David, Senior Brand Manager at Energizer Holdings, has put it, the work involves the regional offices in Asia, the Middle East and Africa, and Europe. A central digital platform that regional teams cannot get changes into is a platform they will route around, so the operating model has to serve both the centre and the markets.
Privacy work started in earnest around 2017. The General Data Protection Regulation turned a marketing question into a legal one and made every embedded script, analytics tag, form and social widget something that had to be inventoried, justified and gated behind consent. On a fifty site multi-region estate that is a substantial discovery exercise before it is a configuration exercise. It is also the work that turned this firm into a privacy consultancy.
Several reached production: an AI battery finder that matches a device to the correct cell, a product comparison and replacement finder, an industrial facility power advisor for business customers, a code of conduct assistant, and an analytics reporting assistant. Generalised versions of these are running as public demonstrations on this site under Neuro, on sample data rather than client data.
Shared components and a shared release process, not shared content. Consistency that matters is structural: the same consent implementation, the same analytics data layer, the same accessibility patterns in navigation and forms, the same product data shape. Consistency that does not matter is editorial. Markets know their own audiences and forcing a single global page on them produces sites the local teams quietly abandon.
Substantially, and not for the reasons people expect. The creative translates. The difficulty is prize and promotion law, which differs by country, plus data capture rules, consent wording, age gating, tax treatment of prizes, and the fact that a campaign running in thirteen countries has thirteen approval chains. The engineering is the small part.
Yes, and the method transfers down more cleanly than it transfers up. An organisation with twelve websites has the same problems as one with fifty: no inventory, inconsistent consent, unowned subdomains, analytics nobody trusts. It simply has fewer of them, so the fix takes months rather than years. See website management for how we scope it.
Written and reviewed by the Green Arrow Consultancy team, led by Darren Tyler, founder and chief executive.
Most organisations we speak to are out by a margin that surprises them. A discovery gives you the real inventory, the third-party scripts on each site, and a prioritised list of what to fix, consolidate or retire.