In Part 1 I laid out how this effort to get away from the confusion and cruft on your site has to start with

• empowering someone / team to create and maintain a coherent website,
  • or if that person already exists, like a CPO, to remind them this is part of the job, and also
• doing the hard work of formulating what quality looks like or how it shows up in metrics.

Next follows a set of steps that have worked in previous instances of doing this kind of renewal.

Recommit to your customers and how you can truly help them

So before anything, you need to have a vision of what you want to be for your customers. Jared Spool has written very clearly about what a vision is and what it must do and how to get there, and it starts with the question: What does the ideal experience for our users look like?

This means that you truly need to know what your customer’s life is like, and what their pain-points are. You have to do your structured research of who they are and what they want from you and what they dream you could do. Because you don’t want to just listen to a set of feature requests from your biggest clients and end with a vision of a faster horse.

This vision has to include:

• An overview of who you want to target. This can be personas, descriptions, actual customer profiles, but somehow you have to make these people come alive for your teams. They may think they know them already, but if your teams truly did, why is everything on your site so hard or unclear or old?
• Their pain-points, both as a context of using your service, and while using your service. What are the trying to achieve in the wider world?
• What you want to be for them in the future.
  • If you have ambition and capability, you could paint a whole dream scenario target of how your new efforts truly improved the lives of your customer. Sit down with customers and creative people from your company and imagine how the currently toughest problem for your customers would be solved as if by magic by your company.
  • At the very least, describe what you want your customers to feel and experience when they use your site, what kind of relationship you want to have with your customers on a brand level and a site level.
• A very rough outline of what kind of platform you want to be for your customers. No too detailed, because your teams need to be contribute their knowledge and experience to make this all better.

Making that vision

Normally you only need to make that vision in companies or divisions that have lost their way from the original founding culture leadership had in mind when they set the company up; companies that have a lot of legacy as a result of time of M&A or that have pivoted once or more. But you may also need it after having been in experimentation, learning or sheer survival mode for too long, AI-driven or not.

Whatever the reason, if you have to right a wrong course, it has to come from as high as possible. The C-suite might not have time to get involved to develop a vision of how the website works in the future, but the layer below will have to invest the time because in the end, getting back on track requires a culture change and the C-suite sets the culture, so they need to buy in to the point it looks like the vision comes from them.

The best version of this I have seen did involve an outside agency, but the cooperation was carefully designed for the outside agency to be a catalyst, not a supplier. The agency did not gather all input, go off for a few weeks, and return with a deck of advice; that model I have seen have zero results too often. What I have seen work is an agency working with the highest level stakeholders they all could have regular progress and work meetings, multiple times a week, discussing who to interview, what sources to consult, how to divide the work between agency and client, working together on gathering information, going over responses and results, constantly putting things in context, strongly benefiting from the outside POV the agency brought to question everything, and working towards a vision together with constant iterative feedback so that nothing about the final vision is a surprise to anyone. The storytelling about that result, however, was at the highest level like only a good digital agency can do.

You definitely want this done by the highest level management that has the time to do this, and keep the layers above involved. They should make the time for this because this is the future of the company, after all.

If only one part of the digital experience needs this refocus, it is OK for that part to take the lead, but they have to involve and consult with all the parts contributing to the digital experience, or resentment about being left behind or not joining up will build and infighting becomes a possibility, tanking the investment in the vision. I have seen this happen first hand, and the fact that the strife was petty AF somehow did not seem to deter grown adults from engaging in it.

Inventory

So now you know where to go. Now you have to look at what you got and make a first improvement. In practical terms: you have to inventory everything to know what to cut. And you do have to cut. Or merge. Or redesign. And all of the above.

You need to map out all the journeys a user can take through your site, together with:

• stats of how often they are taken,
• how users feel about the steps,
• and how they succeed or fail.

This is simple journey-mapping, but if you never did it or have not kept your maps up as changes were made to the site, whatever you have now is outdated and the totality of your site now only lives in a few people’s heads. And without them you simply cannot make good choices. You have to be able to put them all together so your team can see where the gaps and the issues are, and also so everyone is on the same page of what it is that that can be done and how on the site.

Having this mega-map will also help onboard new stakeholders and seniors much faster, so there is a secondary, monetary benefit to making and maintaining it. And if you really can’t swing it for your whole site, at least do it for the top 3 to 5 key journeys on your customers over their lifetime.

The second thing you want out of this exercise is agreement about what the basic concepts in this service are (also known as “the mental model you want the user to have of what you offer”):

• the objects and ideas you offer or sell to the user, that they need to interact with and keep track of to get to success,
• the actions a user can take, most often seen as steps or transitions in their journeys,
• the visual items they interact with on the screen, like profiles, settings, shopping baskets,
• the names you want them to use when interacting with chat prompts or visual controls like buttons and links to next steps in the journeys.

This is where you need to involve stakeholders and product teams for each area, just to surface the different viewpoints on all these steps and objects, why they exist, where they originate from, what they cover, and what differences are between similar concepts. Skipping this step to “hit the ground running, trial innovations” gets very expensive in the long run: confusion will be endemic in the project teams and users, and the simplifications after take longer than if you had just started out with a known, curated list of what is what.

Simplify

During your inventory you or your team already noticed duplicates, or worse, not quite duplicates: concepts or journeys that are almost the same but not quite.

Now that you know in what direction you want to go, you can actually, with purpose,

• group similar items together, and that can be several groupings: by journey, action, goal, page, etc.
• merge items (concepts, pages, journeys, visual constructs like icons and images, actions) that are either duplicates of each other or need only small changes. You want less of everything.
• cull items that confuse, distract, annoy, or users simply do not easily glide through,
• make peace with the fact that there’s always going to be a category in any categorization system of oddities and exceptions.

This is now a first practical step your teams can get to work on that will visibly improve the experience and have pay-off in faster future development.

Get your foundations right

Consistency

Go through all your content and fine-tune it to only use the concepts, journeys, branding, visual language, and names you settled on. Make sure you stick to what your vision needs to come to fruition; don’t just work to get everything right now, but work towards the future.

You have to document this. Online, internally, describing these guidelines so new creators in your company can be on-boarded quickly, and future output stays on course. The “new creators” here include the LLM AIs you will set up to review your current materials and help you review new output for not being on-brand and on-guideline. AIs absolutely shine at this work of comparing one structure to another.

In practice this means that your workflows now have to be retooled so that anything created for your website or app goes into central places for the AIs to help flag compliance. All imagery, all content, all video, as a step in getting published. This can be in directories, the CMS, repositories, but they all have to be known and monitored. Very often these workflows are about your designers and programmers giving their AIs prompts on what to make, and then having other agents check for compliance and validity when the work is added to repositories. It should also include a stage for a human to check the results. You end up with many teams working more or less together, but subject to a consistency step to align everything.

Design System as a marker of collaboration failure

These guidelines sound remarkably like an expansion or completion of a Design System, and they really are, serving the same function of speeding up production while maintaining ease of use. Except this consistency is not just at the visual level, but at the language and conceptual level, minimizing what cognitive objects like brands, languages, actions, journeys the user has to keep track of.

In theory, implementing and using a Design System already has given you the experience of how to maintain consistency and speed up production. But you have to be honest, did you really fully implement the Design System? Did you empower the DS team to enforce using the DS, including updating the old code? Or maybe make a junior FE engineer available to systematically transfer the codebase to the DS, even if it took 6 months? Or did you just hope the DS would be used by developers moving forward and that old pages would be updated to use the DS when they got touched (which ensured the oldest and most stable flows would never be updated because nobody ever touches them)? There’s a ton of DS managers burning out because they are now made responsible for DS outcomes they have no authority over, and now the DS is one of five sources to be synchronized for every brand update instead of the only one.

If you didn’t get the balance of responsibility and authority right for your Design System creation and management, how is this new push going to work then? How are you going to make it sustainable? How will you keep it going?

Show a pilot

Before I go deeper into the issues with collaboration that are the real problems holding back a tight, solid offering, a word of warning: it will be around now that your main stakeholders like your CEO or CTO will complain that they spent a lot of money on agencies and time and haven’t seen anything breakthrough on their website. This level of stakeholder, unless from a design background, doesn’t see the value in creating consistency and getting rid of cruft as a preparation for new, considered functionality. They just see the bill from buying a vision and doing first alignments and no new 10x lines going up anywhere yet.

Suggest a pilot for the vision: a project of limited scope that illustrates one or more aspects of the vision, and a time-frame for it. Get input from the teams currently working on those pages or features about how long things take, what the workflow really is, and what is on their list to fix anyway. Then choose carefully with your Vision project team what will show off the direction the best in the time that you have. Maybe it is updating a section, a complex journey, a big piece of the navigation. If you can, do more than one piece.

You need to deliver or your career is internally over. Once you deliver and the internal forces start understanding what you are doing, make sure to credit their backing and wisdom. They will not understand from this pilot the full extent and ambition of the vision, but you have bought yourself time.

Fix your collaboration architecture

Why did your UX designers let it get this bad? Weren’t they in charge of the experience? Weren’t they running the Design System? How did they let everything get so fragmented?

Again, look at your product teams, look at the flow of features from ideas or requirements to screens or devices, and then answer: who has the power in that chain to say No? What mechanism was there to make sure new features fit what research said was what users struggle with? Ever since the lean product discovery team became a way of life inside companies with their squads (or domains or tribes) and their product triangles of product manager, design, and one or more devs, the designers did not get to say no anymore; it was either the product manager who made the final calls, or decision making was split between the three with a strong bias to having to deliver something, always, and telling Design that the obvious gaps would be fixed when iterating to version 2, after the MVP ‘won’ and would be committed to. Don’t blame design for something they had no final authority over.

(Spoiler: in 90% of all SaaS, the MVP ‘winning’ meant that it never got retired or enhanced but that the product team got put on another gap identified or idea from the CTO or CEO or other initiative that had to go live in a ridiculously short time-frame to keep up with the competition. The words “Version 2” are just by themselves a bitter joke inside UX.)

As I noted in Part 1, this wasn’t the biggest problem until all the Copilots made it so easy to throw things out there over and over, and since every team is incentivized to succeed by themselves, they tried to. The operating framework in which your teams deployed their autonomy is missing a few pieces, like:

• A reward framework that rewards more than Line Go Up. If you just have one main goal, like say x% conversion / sales / increase in usage so you can increase pricing, that one goal is what the teams will game. If you want to get away from that, you’ll have to look at your digital offering and ask yourself, what does quality mean in this system? What does it mean for the user? Do we care about lifetime value? What will get us the customers that are now asking Claude how to solve their problem the easiest and are being sent somewhere else?
  
  This is if you believe that investing in a streamlined, consistent, simpler experience gets more customers. It may be that simply lowering prices and keeping good inventory on an average experience (the ALDI / Lidl model) is all you need. But if you believe or can prove that a quality app or website is key, you will have to start rewarding your employees and contractors for delivering that quality, and discouraging anything else.
  
  If you want to express those goals as numerical targets, make a basket of them, and apply them all to everyone so all teams take on each-other’s concerns and make a joined-up experience. For example, let’s say besides the standard increased sales target, you are also setting a goal of lower customer support costs and fewer returns, to get that incentive of a better online customer experience going. But if only the customer support department is made to care about customer support costs, the website will never get that simpler login that could fix 70% of all support calls because the website sales team is too busy making a bigger Buy button to hit the sales target; the customer support department will just be told to outsource better. Joined-up outcomes only happen when holistic thinking is rewarded.
  
• A proper empowering of the glue people. Find your glue people in your organization, especially the ones being approached by multiple departments. No, it is not your Product Managers, they are supposed to be interfaces that talk to outside divisions as part of their mandates. Ask around: who has frequently contributed data, insights, ideas, details from an unexpected source outside of the team? Who in the team often gets called by other divisions for insights into how the digital side works (which means they are very well trusted)? Who recently made sure two different digital areas aligned?
  (If most of the people in question turn out to be women and/or minorities, you have a culture problem, BTW.)
  
  Once found, ask them why they are having to be glue people, because often that is not their actual job. What walls between silos are they trying to puncture? What information about business processes in other divisions, or how users are using the online services, or other insights from various areas, is not being shared but for them? Why do people reach out to them to pass on information between groups or find out how things really work inside digital teams?
  
  Find a way to formalize this information exchange, but not through meetings and reviews. In a large or fast-growing organization, details of how your product or service gets produced and delivered need to be as widely spread as the clarity on who your customer is and what their actual problems are.
  
• Decentralized programming. I’ve written this before, but I am repeating it again as part of the solution: if your product or service is sold or delivered over digital, every division of your company that contributes to those solutions needs to have their own teams to put their part on the website. No more jockeying by various business divisions about “when will they get time from the web team” to put their ideas on the website. If 70% of the customer support calls are about the login experience, why can’t the customer support team fix it themselves?
  
  But didn’t I just advocate for centralization and gatekeeping? Yes, to fix your current mess. After that you need to get to a model where the local autonomy for new ideas and improvements live in harmony with a centrally-supported drive towards clarity, simplicity, and consistency, all in the service of being customer-oriented.

Make a platform for the future

The vision you made tells you where you want to go. You have simplified and streamlined, cut down, removed, built up, written guidelines, created workflows that support re-using the guidelines and the Design Systems, and there’s a person or people trying to make sure everything works together smoothly, while you are also empowering more teams, closer to their stakeholders and their customers, to contribute to the system. What is the goal here after that?

The goal here is a place where there’s a framework of real customer-centricity. Whether it is a new culture that proudly refuses to release anything that does not hit a certain quality mark or does not make the user’s life remarkably better, whether it is a framework of OKRs that doesn’t allow for simple optimization gaming, whether it is a structure of rigorous testing and assurance that even experiments adhere to coding, security, and design standards, you have to create that sandbox.

You want to end up in a place where many can contribute and experiment, being enabled by central guidelines, frameworks, and workflows, supported by AI, that create the security that even the tiniest team trying to get something out for their small group of customers, can design and deliver it quickly and effectively, up to the cyber-security, coding, and design standards set by the company experts in that area. This means that central teams of experts create the sandbox everyone can play in, guided by the company goals and the supervision of the product executives. And yes, the sandbox can have corners for disruption and experiments.

But “let a 1000 flowers bloom” and then end up with 7 half-working competing or retired systems is a huge waste of resources, burns out workers who feel useless or in constant competition with each-other, and hostile to customers—but it sure makes the VPs responsible look like they are “innovating” while they happily get to ignore any details and get away with making no choices. It’s been done and now with AI speeds it is making companies release features that are empty, cloying, and code-wise grossly insecure. The only reason Google survived that is because their ads make so much money.

The reason any start-up eats incumbents’s lunch or opens a whole new category without incumbents is because they have a vision and a culture and they stick to it. Even if the vision comes after a pivot, they stick to their directions and put all their resources into that direction, with leadership that will curate the offering towards a goal. Any company can do that if they have the leadership with the guts to get stuck into details and take bets with enough direction so designers and developers can focus, even in traditional companies now turning digital. But you have to stop treating the digital arm as its own contained department you can look at from a distance and occasionally yell ideas at.

If you’re the CEO of a media conglomerate and you only find out from an external source that the search function on your main delivery channel, the website, is hard to use and hasn’t comprehensively searched your database for years now, you’re in the wrong job, because how your website fundamentals work aren’t minor details anymore you can just “leave to the IT boffins”. If you’re a CEO of a SaaS and your users are disappointed by the experience after purchasing a subscription because it is so much dowdier than the ads and funnel that got them to subscribe, you’ve lost sight of how you serve users and how your tech teams are unbalanced and half is in the dumps. (Both are actual examples I lived through.)

Your website and apps are not an afterthought to the amazing Big Deals you fly around the world to close. Linus Torvalds managed thousands of external contributions to the LINUX code-base but through his vision, tastes, and help from his closest colleagues still kept his product coherent, stable, and on-track for decades. Steve Jobs could quickly communicate in a single meeting to the home video editing software team the details of what level of usability he wanted so as to get them going and then maintained enough overview to demand hardware and software work together and be perceived as high quality across all product lines for years. For both these examples I can give you five of competitors, sometimes huge ones (Nokia), that lost because they could not get their act towards the user together and let all their middle-managers compete against each-other, ending up with disjointed crap nobody trusted.

Everything King Midas touched turned to gold, as he wished for, but people forget the part where the abundance starved him. Gen AI may be exactly what you asked for, too, with real features being delivered at the highest velocity, but the abundance is confusing users and turning them off, and you lose them and their money. That same AI, through a chat bot, is sending people to cleaner competitors, warning them about your dark patterns, and guiding them through your offering in ways that do not allow you to bind them. And on top of that, the token-economy is collapsing, so you really need to use your AI more targeted than pray-and-spray.

Time to clean up or slowly starve.