Barker & Bowl Pet Foods Ltd

• You have identified five goals for the next six to twelve months: growing revenue, increasing repeat business, improving data and reporting, adopting AI across the business, and automating manual work. That is a full plate, and the risk is treating them as five parallel projects rather than a sequence. Repeat business and better reporting are the fastest routes to sustainable revenue growth, so those two deserve priority over channel expansion or AI adoption until the foundations are solid.

• Your three self-identified priority areas, winning and keeping customers, marketing and sales channels, and measuring performance and reporting, map directly onto the scores where you rated yourselves highest for importance (80 out of 100 for this element). The gap between that importance score and your capability score (36) is the sharpest signal in the whole audit: you know what needs doing, but the tools and processes to do it are not yet in place.

• The biggest operational drag you described, too many manual processes and disconnected systems, is not just an IT problem. It is a marketing problem. When the data that tells you which customers are lapsing, which products are margin-positive, and which channels are actually producing sales lives in separate places and has to be assembled by hand, the decisions that follow are slower, later and less accurate than your competitors who have that picture automatically.

• Marketing is reasonably active but organised around stores rather than commercial objectives. Each store manager posting their own local social content without a central plan means the brand communicates in fragments. Head office runs email and your existing paid media on Google and Meta, but without a joined-up content calendar those channels pull in different directions rather than reinforcing each other.

• The AI activity you mentioned, using ChatGPT for email and product copy drafts, and a handful of Copilot licences at head office, is a start, but it is informal and undocumented, and the stores are entirely outside it. You noted that demand forecasting and reporting are the prizes you most want from AI. Getting to those requires clean, connected data first. The AI question and the data question are the same question. That said, formalising how AI tools are used across the team, who can use what, for what purpose, and with what guardrails, is work that falls squarely within Anicca's separate AI Adoption Roadmap rather than this one.

• Key-person dependency sits at a moderate level by your own assessment: one person owns the product feed (the data file that tells Google and Meta which products to advertise and at what price), one owns email, one owns buying. That is thin cover for three functions that directly affect revenue. If any one of those roles is disrupted, a core channel goes dark or buys badly until cover is arranged. Process documentation and cross-training close that risk, but so does moving those functions onto platforms that give visibility to more than one person.

• The audit process itself is the practical first move. A structured review across channels, content, systems and reporting produces a clear ranked list of where to put effort, with two or three immediate actions that do not require new budget or new headcount, followed by a phased plan that grows as results justify it. Armadello, Anicca's analytics and reporting product, would then give leadership a single view of channel performance, spend, sales and margin rather than the current situation where that picture has to be assembled manually from separate platform exports. That one change alone would shorten the time between a problem appearing and a decision being made about it.

What the business does and why it matters to position it clearly

• Your product range is genuinely broad: dry food across grain-free and classic lines, wet food, natural treats and chews, life-stage ranges for puppies and seniors, a cat food range, and monthly subscription bundles. That breadth is a commercial strength, but it also means you need a clear answer to "why Barker & Bowl rather than one of the specialists?" that works at every price point and every touchpoint, from a store conversation to a paid ad to a product page.

• You are priced at a slight premium against competitors including Lily's Kitchen, Burns Pet Nutrition, Butternut Box, tails.com and Pets at Home own-label. That premium only holds if the customer understands what justifies it, and right now the advice-led in-store experience is probably doing more of that work than your digital channels are. Bridging that gap is one of the clearest commercial opportunities in front of you.

• Your growth direction is UK market share, larger basket sizes and expanded EU shipping, with no new physical stores planned. That means the ecommerce side of the business needs to carry increasing weight, and the five-person ecommerce team is the engine of that. The question is whether their processes and tools are matched to that ambition.

Where the internal friction is actually costing you

• You named reporting and spreadsheets as one of the biggest time consumers in the business. That is almost certainly being felt across marketing, ecommerce and finance, where people are exporting data from different places and manually assembling a picture of what is happening. Armadello, Anicca's reporting product, would bring channel performance, spend, sales and return on spend into one place rather than having your team stitch it together by hand each week.

• Data entry and re-keying appearing on the same list as reporting is a warning sign. When information moves between systems manually, it slows decisions and introduces errors, and in an ecommerce and retail operation, those errors tend to show up as too much stock or not enough in the wrong places, or orders that take longer to process than they should.

• Customer service is a nine-person team handling what sounds like a meaningful volume of contacts, with returns and complaints also calling out as time-heavy. Without knowing the split between routine enquiries (order tracking, subscription changes, delivery queries) and more complex complaints, it is hard to say how much of that resource could be freed by better self-service options or smarter triage, but the fact that it features twice on your high-effort list suggests the current setup is not coping cleanly.

• Store staff turnover is the people pressure you named specifically. Training new staff on an advice-led sales approach takes time, and when turnover is ongoing across 20 stores, that onboarding cost compounds. A standardised onboarding pack, with product knowledge, common customer questions, and the brand's positioning explained in plain terms, would reduce the time to confidence for new starters without depending on a busy store manager to pass everything on informally.

• Your marketing team of eight, supported by an external agency, is the resource doing the work that connects all of this outward-facing activity. The internal process question for them is whether briefing, sign-off and content scheduling works to a clear rhythm or whether things circulate informally and create delays. You did not describe that as a named problem, but with a head office team that size, supporting both retail and ecommerce, it is worth checking whether the bottleneck is resource or process before assuming you need more of the former.

The problem is that the way you currently manage and develop customers is largely manual, reactive, and not connected to what you actually know about their purchase behaviour.

Who your best customers are and what keeps them

• Your subscription customers are your most valuable segment by a significant margin, yet they represent only 15% of revenue. The 3.5% monthly cancellation rate means you are losing roughly one in twenty-eight subscribers every month. That adds up quickly: at that rate, without new subscribers joining to replace them, you would turn over your entire subscriber base inside two and a half years. Before you invest further in acquisition, it is worth understanding why people cancel and whether a simple intervention, such as a pause option, a personalised check-in email, or a reminder of savings made, would bring that rate down.

• Your in-store walk-in trade accounts for 55% of revenue, which is a meaningful strength. But there is currently no formal way of converting those footfall customers into something more durable, whether that is a subscription, a loyalty account with a proper email address attached, or a referral. The store visit is the warmest possible moment to capture a customer relationship, and it is largely going unrecorded.

• You run paid media on Google and Meta, which is the right direction for reaching new customers online. The question is whether those campaigns are working in support of the subscription and loyalty journey or simply driving one-off orders that do not build the underlying customer base you want. That is worth examining before you increase spend on either channel.

Retention and cross-sell

• Your cross-sell opportunity is straightforward to describe: a dog-food subscriber who has never been shown the cat range or the treats line is leaving money on the table that you have already earned. The gap is not about product range, it is that your email communications are broadcast rather than triggered by what a customer has actually bought. A triggered sequence, where a dog-food subscriber gets a relevant cat-food message only if you also know they have a cat, or a treats recommendation based on what they already buy, would require a basic connection between your email platform and your order data. That is not a complex build, but it does need to be set up deliberately.

• Your loyalty scheme and regular email newsletter are both in place, which is further along than many businesses at this stage. The problem is that neither appears to be driving personalised communication based on purchase history. Running a loyalty scheme as a points accumulator without using it to segment and message customers differently means you are collecting data you are not acting on.

• You run remarketing, which will be recapturing some lapsed browsers and basket abandonments. Whether it is recapturing lapsed subscribers or one-off purchasers who have gone quiet is worth checking. Lapsed subscribers who cancelled three to six months ago are often the easiest group to re-engage, especially if the reason for cancelling was price or convenience rather than dissatisfaction.

Reviews, referrals and NPS

• Google Reviews and Trustpilot are live per location but not centrally managed. That is a missed opportunity in both directions: positive reviews build local search visibility and trust for new customers, and unmanaged negative reviews can quietly damage the perception of a store even when the underlying experience is good. Centralising review monitoring and putting a simple process in place for responding to both positive and negative reviews would improve this without requiring significant resource.

• Your NPS process is one-off surveys rather than something triggered at a consistent point in the customer journey, for example thirty days after a first subscription order. That means you are sampling mood rather than measuring experience at a defined moment. Shifting to a triggered NPS survey at a fixed point would give you something comparable over time and would let you identify unhappy subscribers before they cancel rather than after.

• You have no formal referral scheme despite clear word-of-mouth strength from your stores. Customers who are willing to recommend you unprompted will usually refer more, and more reliably, when there is a simple structure behind it. A basic referral offer does not need to be elaborate; it needs to be easy to share and easy to redeem, with the reward landing for both the referrer and the new customer.

Trade accounts

• Your trade accounts (groomers and kennels) represent only 5% of revenue but have the potential to be a consistent, higher-volume segment. It is worth knowing whether they are being managed as a distinct customer type with their own pricing, communication and ordering process, or whether they are simply being handled as large individual orders. A small amount of structure here, such as a trade account portal or a dedicated email sequence, could make this segment more self-sufficient and more sticky.

Bringing all of this into a single view, where you can see subscription health, cross-sell performance, review scores and referral activity in one place rather than across separate platforms, is where Armadello would add practical value. It would give you and your team one reliable picture of customer performance without having to pull reports from your email platform, your ecommerce backend and your review sites separately every time you want to understand what is happening.

Customer communications

• Delivery queries stacking up over the weekend is the most visible symptom of a structural gap, not a staffing one. Your current Zendesk and live-chat setup assumes someone is available to respond; it gives you no way to set expectations automatically or deflect straightforward "where is my order?" questions to a self-serve tracking link. A simple automated response triggered by order-status keywords, set up inside Zendesk, would acknowledge weekend queries immediately and give customers a resolution path without waiting until Monday.

• Subscription pause and swap requests taking multiple emails to resolve is a different problem. The friction is almost certainly in the handoff between your email inbox and whatever system actually holds the subscription record. If a customer sends a message, gets a reply asking for their order number, replies again, then waits for a manual change to be made, that is three or four exchanges for a task that should take one. Mapping that exact sequence and removing the unnecessary steps, whether through a self-serve subscription portal, a Zendesk macro that pulls the right information in one go, or a clearer confirmation flow, would cut both ticket volume and customer frustration at the same time.

• Your email list running on basic customer-versus-prospect segmentation means you are sending broadly the same message to someone who has been on subscription for eighteen months and someone who bought once three weeks ago. Those two people have completely different relationships with you and respond to completely different messages. Splitting by purchase recency, subscription status and pet type (where you hold that data) would let you send fewer emails to each person while making each one more relevant. That is a meaningful improvement you can make inside your existing email platform without rebuilding anything.

• The vet-reviewed nutrition guides and the annual "state of pet feeding" piece are genuinely useful assets, and the local press pick-up on the annual piece confirms there is an audience for authoritative content. The problem is quarterly-at-best publication with no consistent thread between pieces. A short, plainly-written email newsletter, monthly or every six weeks, that curates your existing expert content and adds a brief editorial point of view from your nutrition or vet contacts would give subscribers a reason to stay engaged between purchases. Your existing social media manager and marketing manager have the resource to produce this; it is more a planning and scheduling question than a capacity one.

• Store teams and the online customer service team giving different answers to the same feeding questions is a trust problem, not just an operational one. A customer who gets one answer in a Leicester store and a contradictory answer from live chat the following week will not know who to believe. A single, shared FAQ document, reviewed by whoever signs off your vet-reviewed content and distributed to both store staff and the Zendesk team, is the fix. It does not need to be elaborate; it needs to exist and be kept current.

Internal and team communications

• The gap between how head office communicates and how individual stores operate is typical for a retail business that has grown faster than its internal systems. Each store handling its own counter queries in its own way means there is no consistent record of what has been said to whom, and no way to spot a question that is being asked repeatedly across multiple stores, which might signal a product issue, a labelling problem or a gap in your customer-facing content.

• Your approval process, where the ecommerce lead and marketing director sign off budget changes above roughly £2,000 and day-to-day optimisation sits below that threshold, is a workable structure. The risk is that it works only if both decision-makers are genuinely in the loop on what is being tested beneath that line. A short written summary, weekly or fortnightly, of what has been tested, what changed and why, shared upward by whoever is running day-to-day activity, keeps that visibility without adding a meeting.

• On working with an external partner: you have told us you want a named contact who understands both retail and ecommerce, a monthly performance call, and written summaries the ecommerce lead can share with the wider team. That is a clear and reasonable working model, and it is exactly how Anicca structures its client relationships. The written summaries matter particularly because they compress what happened in a month into something a non-specialist can act on, which is where Armadello, Anicca's reporting product, earns its place. Rather than the ecommerce lead stitching together exports from Zendesk, your email platform and Google Analytics before a call, Armadello would pull traffic, sales, spend and return on spend into one view so the monthly conversation starts from the same agreed numbers.

• One note flagged by your answers rather than this report's scope: several head-office staff are using ChatGPT for email and copy drafts, and Copilot is available through your Microsoft 365 licences, but none of it is joined up or documented, and stores are not involved at all. That inconsistency will affect the quality and tone of customer communications over time. Closing that gap properly, with a clear policy on which tools are sanctioned, for what, and by whom, is the territory of Anicca's separate AI Adoption Roadmap, not this one.

What you produce and how often

• Blog twice a month, email weekly through Klaviyo (your email marketing and automation platform), social four to five times a week, video roughly monthly, and product descriptions refreshed in seasonal batches. That is a creditable volume for a team of your size, and it shows the content habit is already formed. The problem is not frequency, it is coordination.
• You have no formal content audit on record, only a shared sense of what exists. That means nobody knows which blog posts are pulling in organic traffic, which product pages are thin, or which topics you have covered three times and which you have never touched. You are producing content without knowing what is working, which means effort is probably spread unevenly.
• Product descriptions arriving in seasonal batches is the right rhythm for a retail and ecommerce business, but batch production under time pressure is exactly when inconsistent tone creeps in. If different people write to different unwritten standards, the product pages will feel mismatched, and that affects whether a visitor trusts the product enough to buy it.

Brand and tone of voice

• You have a logo pack, brand colours and a loose tone-of-voice note, but no single guidelines document that anyone can open and follow. The result is what you described: stores, ecommerce and social each interpreting the brand slightly differently. A customer who finds you on Instagram and then visits a store, or lands on your website after seeing an ad, should not have to adjust to a different version of Barker and Bowl. Right now, some of them are.
• The fix here is a one-document brand and tone guide that covers voice, vocabulary, what you do not say, visual rules for each channel and a handful of worked examples showing on-brand versus off-brand copy. It does not need to be lengthy. It needs to be specific enough that a new social media manager or a seasonal store team member can pick it up and produce something consistent without second-guessing it.

Content planning and the pipeline

• A content calendar built around a small number of defined topic territories, perhaps four or five things you genuinely want Barker and Bowl to be known for, would give everyone working on content a framework to plan against rather than filling the week reactively. It also feeds your SEO directly, because search engines reward sites that build genuine depth on a topic rather than publishing scattered articles.
• You have a designer, a social media manager, a marketing manager and an external agency. That is enough resource to run a structured pipeline. The gap is editorial coordination, deciding in advance what goes out, in which format, for which audience, and making sure it is checked against the style guide before it is published. A simple shared calendar, even a spreadsheet, changes the way the team produces content because it removes the weekly scramble of deciding what to write about.
• Your very high comfort level with spokespersons, testimonials and case-study content is a genuine advantage that most businesses cannot match. Real pet owners talking about their animals and what they feed them is the kind of content that performs well on social and in email, and it is exactly the sort of thing that gives Barker and Bowl a personality that a faceless pet food brand cannot replicate. That asset is currently underused.

AI-assisted production and disclosure

• You are already using ChatGPT for first drafts of emails and product copy. That is a sensible use of the tool, but without a style guide behind it, AI-drafted copy will default to generic rather than on-brand. The style guide is the document that tells both your team and any AI tool what Barker and Bowl actually sounds like.
• There is also a disclosure point worth taking seriously now rather than later. The EU AI Act's transparency requirements under Article 50 require AI-generated content to be clearly disclosed to readers. If any of your content reaches EU customers, that obligation applies directly from August 2026. If your audience is UK-only, the direction of travel is the same and early-mover preparation costs you nothing. Keeping a simple log of which content was AI-drafted and which was written entirely by a person means disclosure, when it is expected of you, is a formality. Starting that habit now, before the volume of AI-assisted content grows further, is considerably easier than reconstructing it later.
• The broader question of how AI tools are used across your team, the fact that head-office staff have Microsoft Copilot through their existing Microsoft 365 licences while stores have nothing, and that none of it is documented or joined up, is a real gap. But solving it sits squarely in the territory of Anicca's AI Adoption Roadmap rather than this one. Name the gap, plan to close it, and treat this roadmap's job as making sure your content production is structured well enough that AI assistance, when it is properly governed, can genuinely speed it up.

Paid search and paid social

• Google Ads is spending £18,000 a month and generating a return on ad spend of 3.6, with a cost per sale sitting comfortably within your £20-25 target. That is your strongest-performing paid channel by some distance, and the numbers suggest it is not yet being pushed to its ceiling. The question worth asking is whether the current budget is the right one, or simply the familiar one.

• Meta Ads tells a different story. At £9,000 a month and a return on ad spend of 2.4, it is held to a softer awareness target rather than a cost-per-sale discipline. That is a legitimate choice for some businesses, but it means you are spending £108,000 a year on a channel where you have deliberately made it hard to hold performance to account. If Meta genuinely drives awareness that converts later through another channel, your attribution needs to show that. If it does not, the budget case for it weakens considerably.

• The blended cost of winning a customer online at £22 is a reasonable starting point as a benchmark, but it is only useful if you can break it down by channel. Right now, Google and Meta are pooled into one number, which means a channel outperforming its share props up one that is underperforming, and you cannot see which is which.

Organic and direct traffic

• Organic search and direct together account for 42% of your GA4 sessions, making them your largest combined traffic source. That matters because neither of those channels carries a cost per click. Protecting and growing them is worth more per visit than your paid channels, yet you told us your blog click volumes on informational queries have dropped noticeably as AI Overviews answer those questions directly. You have started rewriting guides to earn citations in AI-generated answers, which is the right instinct, but you have no way to measure whether it is working. That needs to change before you commit more writing time to it.

• You are already producing content with a five-person in-house team. The constraint is not resource, it is knowing which content to prioritise. Queries where AI Overviews are active, queries where your product pages rank but do not convert, and queries where a competitor outranks you on terms you sell directly are three different problems requiring three different responses. Treating them as one "write more content" task means effort is spread too thin.

Attribution and channel mix

• You told us that the split between store footfall and online is honestly guesswork. Forty percent of your new business comes through the physical stores, and that traffic influences online behaviour and repeat purchases in ways you cannot currently see. Without a way to connect in-store visits to subsequent online orders, or vice versa, your channel investment decisions are missing the largest single part of your acquisition picture.

• Your site converted at 2.1% over the last twelve months against a sector benchmark of 2.2%. That is close enough that the conversion rate itself is not the primary concern, but even a modest improvement across 620,000 annual sessions would add meaningful sales volume without increasing spend on any channel. Fixing what causes visitors to leave without buying is often quicker and cheaper than buying more traffic.

• The practical first step here is getting every paid and organic channel onto the same measurement framework, one that uses the same definition of a sale, the same attribution window, and the same view of margin, not platform-reported figures that each count differently. Armadello, Anicca's reporting product, is built to do exactly that: pulling spend, traffic, sales and return on spend into one view so you are comparing channels on equal terms rather than reconciling separate platform exports. Once you have that, the conversation about whether to increase Google Ads, tighten Meta's targets, or protect organic through structured content becomes a numbers conversation rather than a judgment call.

What is connected and what is not

• Shopify sits at the centre of your ecommerce operation and handles product listings, orders and basic customer records. It is a capable platform and the foundation is sound, but its value depends on what talks to it. Right now, some connections are in place and some are not, which means data is being re-entered by hand somewhere in the business, or simply not shared at all.
• The most pressing evidence of a connection problem is your subscription delivery slippage during peak weeks. When fulfilment timing goes wrong repeatedly and generates a recurring wave of customer service tickets, the root cause is almost always that your order management, stock levels and delivery scheduling are not sharing information in real time. Someone is either acting on data that is already out of date, or no system is flagging the pressure until it is too late to absorb it.
• Your in-store fulfilment is reliable, which tells you the problem is specific to the online subscription side rather than a general operations weakness. That is useful, because it means you are not starting from scratch. You are fixing a defined part of the picture, not rebuilding everything.
• If your email or customer relationship management platform is not receiving order and subscription data automatically from Shopify, your team cannot contact affected customers ahead of a delay. They can only react after the ticket arrives. Closing that connection, so that a flagged order triggers a customer message without anyone having to notice it first, would reduce complaints and the time spent answering them.
• The broader question is whether your stock and demand data flows anywhere useful. If Shopify knows what is selling and how fast, but that information is not reaching whoever makes purchasing or production decisions, you will keep cycling through periods of too much stock or not enough, particularly around seasonal peaks.

AI tools and the reporting picture

• You are already using ChatGPT and Microsoft Copilot, and several of the tools you pay for include AI features you may not yet be using deliberately. None of this is joined up at the moment, and that is fine as a starting point, but it does mean you are getting individual task-level benefits rather than anything that touches the bigger prizes you named, demand forecasting and reporting. That is territory Anicca's separate AI Adoption Roadmap is built to address, and it would be worth picking that up alongside or after this work rather than trying to solve it here.
• On the reporting side, the immediate problem is visibility. If your paid media performance on Google and Meta, your Shopify sales data, and any subscription metrics are all sitting in separate places, nobody has a single picture of what is actually working. Armadello would bring that together into one view: traffic, orders, spend, return on spend, and margin in one place, rather than someone manually pulling exports from three platforms and trying to reconcile them in a spreadsheet.

The practical first move

• Map every system you currently use and identify which ones already have a connection available and which do not. You do not need to connect everything at once. The pair to prioritise first is whichever link would directly address the subscription delivery slippage, most likely the connection between your order or subscription management and either your fulfilment process or your customer communications platform.
• Once that connection is in place and the recurring ticket problem reduces, the next step is extending the data flow outward so that stock levels, sales velocity and customer behaviour inform decisions rather than being discovered after the fact.

That time has a real cost, the numbers arrive late in the week, and there is no guarantee that the version the ecommerce lead trusts matches what the finance manager is working from on the monthly pack.

• Your five weekly KPIs, revenue, profit and margin, conversion rate, stock position, and repeat purchase rate, are the right things to track. The problem is not what you are measuring but how you are assembling those numbers. Each one currently lives in a different place and is joined together by hand.

• Store numbers arriving by email from each manager means the retail estate is entirely outside your digital reporting view. By the time those figures reach anyone who can act on them, the trading week is already over. There is no live picture that combines online and in-store performance into a single position.

• Nobody owns data quality explicitly. That matters more than it might appear. When figures from Shopify, GA4 and the ad accounts are stitched together manually each week, small discrepancies, attribution differences, time-zone mismatches, a filter someone applied and forgot, add up quietly. Over time, different people in the business start trusting different numbers, and decisions get made on whichever version is nearest to hand.

• Your site conversion rate over the last twelve months sat at 2.1%, just below the sector benchmark of 2.2%. That gap is close enough that you should not read too much into it in isolation, but you also cannot diagnose what is driving it without clean, consistent data that connects traffic source to session behaviour to completed sale. Right now that connection requires manual work every time someone wants to look at it.

• Your Google Ads spend runs at £18,000 per month and your Meta Ads spend at £9,000 per month. Assessing whether those budgets are correctly split, or whether either channel is pulling its weight against a margin target rather than a platform-reported return, requires combining ad account data with your actual order and margin data. That is exactly what manual reporting makes difficult and slow.

• Armadello, Anicca's reporting product, is built to solve this specific problem. It connects your channel data, spend, traffic, conversion and commercial performance into one live view, so the Monday trading report becomes a dashboard that is already built when the week starts, not a task that takes the morning. It would also give leadership one agreed set of numbers rather than two separate packs compiled by two different people.

• Assigning explicit ownership of data quality, one named person who checks that the right data is flowing into the right places and flags when something looks wrong, is a process change that costs nothing but makes every number you look at more trustworthy. Without it, errors accumulate and nobody is accountable for finding them.

• The AI tools your team already use, ChatGPT for copy and Copilot through Microsoft 365, have a direct relevance to reporting: demand forecasting and automated reporting summaries are the prizes your team has already identified. Getting the underlying data into a clean, connected state is the prerequisite for either of those. That is a separate conversation sitting within Anicca's AI Adoption Roadmap, not within this one, but the two are not independent. Better data infrastructure now makes that AI work viable when you are ready to tackle it.

Without a clear view of margin by channel, you are making budget decisions on top-line signals rather than commercial ones.

What the margin picture actually looks like

• Your treat ranges carry the strongest gross margin at 55-60%, but if those products are disproportionately acquired through paid social (where your return on ad spend is running at 2.4, against 3.6 on paid search), the net return after media cost is narrower than the product margin implies. Until you can see ad spend by product category against the margin those products carry, you cannot confidently say which sales are worth chasing hardest.

• Subscription bundles are discounted down to around 40% gross margin. That is defensible if the lifetime value of a subscriber is genuinely higher than a one-off buyer, but that calculation requires knowing your actual cost of winning a subscriber through each acquisition route, not just the platform-reported figures. At the moment, that number does not appear to exist in a reliable form.

• Dry food sits at 38-45% gross margin and wet food at 30-35%. These are your lower-margin categories. If either is being pushed heavily through paid channels without a margin filter applied to the campaign settings, spend will optimise towards sales volume rather than profitable sales.

• Damaged-in-transit claims on subscription boxes are running at around 3%. That is not a catastrophic rate, but on a subscription line that already carries a discounted margin, each damaged box absorbs a disproportionate share of the profit. It is worth quantifying the total monthly cost of those claims (replacement product, re-shipment, customer service time) and deciding whether better packaging or a change of carrier is cheaper than absorbing the ongoing loss.

Software and platform costs

• At £6,500 a month across all software categories, you are spending around £78,000 a year on tools. That is not unreasonable for a business of your scale, but it is only sensible if you know what each tool is doing for you commercially. The honest question is whether any of those subscriptions are retained out of habit rather than active use.

• The mix of reporting you described as "a mix of the above" suggests that no single system currently brings channel spend, order data, margin and returns together in one view. That means any margin analysis you do is built by hand from separate exports, which is slow, prone to error, and unlikely to happen frequently enough to catch a problem early.

• Armadello would bring your channel spend, sales volume, return on ad spend and margin data into one place, so you can see, on a weekly basis, which channels are generating profitable sales and which are generating volume at a cost that the margin does not support. That is the difference between acting on a margin problem in the same month it appears and discovering it at year-end.

The marketing budget in commercial terms

• A monthly marketing budget of up to £35,000 represents a meaningful proportion of revenue at your scale. Allocating that by channel performance data alone, without margin attached, means you will tend to reward channels that look efficient on a cost-per-sale basis even when the products those channels sell are the lower-margin ones.

• Your paid search return on ad spend of 3.6 is the stronger of your two paid channels. That does not automatically mean it deserves more budget, but it does mean it deserves closer analysis: which product categories is it driving, at what average order value, and against what margin? Those three numbers together tell you whether to increase the paid search budget or hold it.

• The 14% of sessions coming through paid social is generating a return on ad spend of 2.4. That is not necessarily a channel to cut, particularly if it is doing acquisition work that feeds your subscription base, but it needs to be evaluated against what a subscriber genuinely returns over time, not just what the first order costs to win.

• Where your AI adoption gap touches cost directly, the fact that demand forecasting has not yet started means you are almost certainly carrying too much stock in some lines and running short in others without a systematic way to see it. That is a stock and cash-flow problem as much as an operations one. The AI Adoption Roadmap is the right place to address that specifically, but the financial consequence of not addressing it belongs in this conversation.

Given how frequently platforms update their data sub-processors, storage arrangements and retention policies, that is a real exposure, not a theoretical one.

Marketing consent and cookies

• Your use of Klaviyo for email marketing means your marketing consent and data collection practices fall under both data protection law (GDPR) and the marketing-consent rules (PECR). Most businesses that have grown their email list over several years without a formal consent audit find a proportion of contacts whose opt-in basis is unclear or out of date. That is the first thing worth checking.

• Cookie consent on your Shopify site needs to accurately reflect every tool and tracking tag you are actually running. If you have added pixels, analytics scripts or third-party tools since you last reviewed the cookie banner, your declared practices and your actual practices are likely to be out of step. The Advertising Standards Authority (ASA) is also increasingly attentive to product and health claims in pet food marketing, so your ad copy on Google and Meta and any on-site claims about nutrition or ingredient benefits are worth reviewing against current guidance before a complaint finds them first.

• Extended producer responsibility on packaging is a legitimate emerging concern for a business in your category, and changes to pet food labelling claims are an area where regulatory direction can shift without much warning. Neither of these sits inside Anicca's remit, but both are worth tracking through your sector body and building into your internal planning calendar, so that when guidance firms up, you are not rewriting packaging or updating product pages under pressure.

• The EU AI Act's transparency requirements under Article 50 require clear disclosure when a customer is dealing with AI-generated content or AI interaction, from August 2026, applying directly to EU-facing businesses and signalling the direction of travel for UK practice as well. You are UK-only at present, so the direct legal obligation does not yet apply, but if that changes, or if UK regulatory expectations move in the same direction, having no disclosure practice in place will be a problem to fix in a hurry. The sensible move is to fold a simple AI-content disclosure approach into the same review you should already be running on consent and cookies, rather than treating it as a separate exercise later.

AI tool use

• Your marketing team is using ChatGPT to draft emails and product copy, and a couple of head-office staff are using Microsoft Copilot through your existing Microsoft 365 licences. None of this is documented, nothing is joined up, and there is no usage policy. That is a governance gap with real implications for data handling, content accuracy and brand consistency. Closing it properly, including an AI usage policy, guidance on what data staff should and should not put into these tools, and a decision about which tools to sanction, is the territory of Anicca's separate AI Adoption Roadmap. This report is not the place to design that policy, but it is the right place to say clearly: this needs to be addressed, and the AI Adoption Roadmap is the product built to do it.

The practical first step

• Run a structured review of your consent practices, cookie banner, and all six of your current DPAs. Name one person internally who owns compliance as an ongoing responsibility rather than a one-off project. The review does not need to be a lengthy exercise, but it needs to be deliberate and documented so you can show your working if a regulator or a concerned customer ever asks. Where Armadello would help is in making visible exactly which tools are collecting data from your site and your marketing channels, giving whoever owns this a single place to see the picture rather than logging into six platforms to piece it together.