One Two Three Send

Author: onetwothreeadmin

  • Analytics event sampling: why your data vanishes above 500K sessions

    Analytics event sampling: why your data vanishes above 500K sessions

    If you’re running a content site that pulls 500,000+ sessions a month, you’ve probably noticed something strange: your GA4 reports don’t always match. Run the same custom report twice, and the numbers shift. Filter by a specific page or traffic source, and suddenly your totals don’t add up.

    You’re not imagining it. You’ve hit GA4’s sampling threshold.

    Sampling means Google is analysing a subset of your data—not all of it—then extrapolating the results. It’s faster for Google to process, but it quietly erodes the accuracy of every decision you make based on those numbers.

    When sampling kicks in (earlier than you think)

    Google’s documentation says GA4 standard properties support up to 10 million events per month before sampling applies. That sounds like plenty of headroom.

    But sampling doesn’t wait until you hit 10 million events. It starts much earlier—often around 500,000 sessions—when you build custom reports, apply filters, or request data outside the standard date range.

    Standard reports (the pre-built ones GA4 ships with) use aggregated data tables and rarely sample. But the moment you:

    • Add a secondary dimension
    • Apply a segment or filter
    • Build an exploration report
    • Request data older than 14 months

    …GA4 switches to on-the-fly querying, and sampling kicks in if your property is processing enough events.

    You’ll see a green badge in the top-right corner of your report that says “This report is based on X% of sessions.” If that number is below 100%, your data is sampled.

    What gets lost in sampled data

    Sampling doesn’t affect every metric equally. High-level numbers—total sessions, pageviews, users—tend to hold up reasonably well even at 20–30% sampling rates.

    But the more specific your question, the worse sampling performs. If you’re trying to:

    • Identify which blog post drove the most newsletter signups from organic search last quarter
    • Compare conversion rates across traffic sources for a specific product page
    • Measure how a site-speed optimisation affected bounce rates on mobile

    …sampling can distort the answer enough to reverse your conclusions.

    I’ve seen sampled reports show a 15% conversion rate on a landing page when the unsampled data (pulled via BigQuery export) showed 11%. That’s not a rounding error—that’s a different business decision.

    How to reduce or avoid sampling

    If you’re consistently hitting sampling in GA4, you have four options.

    Option one: Narrow your date range. Instead of analysing the last six months, break it into monthly or bi-weekly segments. Smaller queries are less likely to sample. You’ll need to stitch the data together manually, but at least it’s accurate.

    Option two: Use the Data API. GA4’s Reporting API gives you access to unsampled data if your query stays within certain limits. Tools like Google Sheets (via the GA4 add-on) or Looker Studio can pull data this way. It’s slower, and there are still daily quotas, but it bypasses the sampling you see in the GA4 interface.

    Option three: Export to BigQuery. GA4 offers a free BigQuery export (up to 1 million events per day). Once your data is in BigQuery, you can query it without sampling. The trade-off: you need to learn SQL, and you’re managing your own data warehouse. But if you’re making six-figure decisions based on this data, it’s worth it.

    Option four: Upgrade to GA4 360. The enterprise tier starts at $50,000/year (sometimes negotiable to $150,000 depending on scale) and raises sampling thresholds significantly. Unless you’re running a seven-figure media operation, this isn’t realistic.

    When sampled data is good enough

    Not every report needs to be unsampled. If you’re checking whether traffic went up or down this week, or whether your top-performing post is still your top-performing post, a 70% sample is fine.

    But if you’re deciding whether to double down on a traffic source, kill a product, or restructure your content strategy, don’t trust a sampled report. Pull the data via API or BigQuery, or narrow your query until the sampling badge disappears.

    The costliest analytics mistake isn’t picking the wrong tool—it’s trusting incomplete data and not knowing it.

    Using GA4 for attribution or conversion tracking? Subscribe to One Two Three Send and get one operator-focused article like this every day.

  • ChatGPT’s memory feature: what it remembers and when to reset it

    ChatGPT’s memory feature: what it remembers and when to reset it

    ChatGPT’s memory feature lets the model remember details across conversations—your business model, your audience, your tone preferences—so you don’t have to repeat context every time you open a new chat.

    In theory, it’s a time-saver. In practice, it can quietly corrupt every output if you’re not paying attention to what it’s storing.

    Here’s how the feature actually works, what gets saved, and when you should delete everything and start fresh.

    How ChatGPT memory works

    When memory is enabled (it’s on by default for Plus and Team users), ChatGPT stores snippets of information you share across sessions. It doesn’t save full transcripts—it extracts facts, preferences, and instructions it thinks will be useful later.

    For example, if you tell ChatGPT you run a weekly newsletter about WordPress hosting, it might remember that. The next time you ask it to write an email subject line, it’ll assume your audience cares about uptime and page speed without you saying so.

    You can view what’s stored by going to Settings → Personalization → Memory. You’ll see a list of bullet points—some you explicitly told it, others it inferred. You can delete individual memories or wipe everything at once.

    Memory is tied to your account, not a specific conversation. If you start a new chat, the model still has access to everything it saved before.

    When memory improves your workflow

    Memory works best when your business model, audience, and output format are stable. If you’re always writing for the same newsletter, using the same voice, and solving the same kinds of problems, memory removes repetitive context-setting.

    Use cases where it helps:

    • Drafting content: You write every Tuesday for a niche audience. ChatGPT remembers the format, tone, and typical topics without a fresh brief.
    • Generating ideas: You ask for post ideas weekly. It recalls your editorial themes and avoids suggesting topics you’ve already covered.
    • Code or automation help: You’re building Zapier workflows or WordPress plugins. It remembers your stack, your naming conventions, and the APIs you use.

    If you’re working solo and your projects don’t shift much, memory reduces cognitive overhead. You get faster first drafts with less prompting.

    When memory pollutes your output

    Memory becomes a problem when context changes but the model doesn’t know it.

    Say you used ChatGPT to write emails for a SaaS product last month. This month, you’re drafting newsletter content for a coaching business. If memory is still active, it might assume your audience is technical, your goal is conversion, and your tone is formal—none of which apply anymore.

    You won’t always notice. The output will feel slightly off—too corporate, too detailed, too salesy—but you might not trace it back to stale memory.

    Other scenarios where memory breaks down:

    • Client work: You’re writing for multiple clients with different voices. Memory blurs the lines unless you manually reset between projects.
    • Experimentation: You’re testing a new content format or audience. Memory anchors responses to what worked before, even when you’re trying something different.
    • Shared accounts: If you’re on a Team plan and multiple people use the same login, memory mixes everyone’s preferences into a confusing mess.

    The worst part: ChatGPT doesn’t tell you when it’s relying on memory. There’s no citation, no flag. It just quietly applies old context to new requests.

    How to manage memory (and when to delete it)

    Check your memory every few weeks. Go to Settings → Personalization → Memory and scan the list. Delete anything that’s outdated, project-specific, or no longer relevant.

    If you switch projects or clients frequently, disable memory entirely. You’ll lose the convenience, but you’ll avoid contaminated outputs. You can toggle it off in the same settings menu.

    If you want memory for some tasks but not others, use Temporary Chat mode (the icon in the sidebar). Conversations in that mode don’t update memory and don’t reference what’s stored. It’s useful for one-off requests or experimenting with a new voice.

    One non-obvious tip: when you do want ChatGPT to remember something, tell it explicitly. Don’t assume it’ll pick up on subtle hints. Say, “Remember: my newsletter audience is non-technical founders, and I always write in second person.” That instruction will stick better than hoping the model infers it from a single example.

    Want more breakdowns like this? Subscribe to One Two Three Send and get one operator-focused article every day—no fluff, no listicles, just tools and tactics that work.

  • Affiliate link cloaking: compliance, tracking, and when to skip it

    Affiliate link cloaking: compliance, tracking, and when to skip it

    Affiliate link cloaking—redirecting yoursite.com/go/product to partnersite.com/?ref=yourID—sounds like good housekeeping. Cleaner URLs, consistent branding, easier tracking. But it introduces technical risk, compliance obligations, and platform dependency that many operators don’t plan for.

    Here’s what actually changes when you cloak links, and when you’re better off leaving them naked.

    What cloaking does (and doesn’t) solve

    Link cloaking replaces long, ugly affiliate URLs with short, branded redirects. Instead of sharing https://convertkit.com?lmref=abc123xyz, you serve yoursite.com/convertkit and 301-redirect behind the scenes.

    The benefit is consistency. Readers see your domain. You can swap the destination URL without editing published content. You centralize click tracking in one place—your own server logs or a plugin like Pretty Links or ThirstyAffiliates.

    What it doesn’t do: improve deliverability, hide the affiliate relationship from networks, or make the link more trustworthy. Email clients still see the final destination after following the redirect. Affiliate networks log the same click. Readers who hover still see the redirect if they check the status bar.

    And you’ve added a round-trip dependency: your server must respond before the affiliate network sees the click. If your host is slow, you’ve introduced latency. If your site goes down, the link breaks entirely.

    FTC disclosure and platform terms

    Cloaking doesn’t exempt you from disclosure rules. The FTC’s Endorsement Guides require clear, conspicuous notice when you earn from a recommendation. A short slug like /go/ or /recommends/ signals commercial intent, but it’s not a substitute for plain-language disclosure near the link.

    Some affiliate programs explicitly prohibit cloaking. Amazon Associates’ Operating Agreement bans link shortening or masking that obscures the final destination. Commission Junction and ShareASale allow it, but require that the redirect preserves tracking parameters and doesn’t mislead users.

    Check your agreement before you automate. Violating terms can mean forfeited commissions or account closure, and most networks won’t warn you—they’ll just stop paying.

    When cloaking makes sense

    If you publish across multiple platforms—blog, newsletter, podcast show notes—and want consistent analytics, cloaking centralizes your data. You can see aggregate clicks in one dashboard instead of stitching together reports from five affiliate portals.

    If you rotate offers or test different partners, cloaking lets you update the destination without republishing old content. A post from 2023 can point to a new tool in 2026 if the /email-tool slug stays the same.

    If you’re building a resource hub or comparison page, branded slugs make the link structure easier to maintain. /bluehost, /siteground, and /bigscoots are more memorable than tracking IDs.

    When to leave links uncloaked

    In email, cloaking adds a redirect that some inbox providers flag. Mailchimp, ConvertKit, and MailerLite preserve click tracking without requiring your own redirect layer. Adding a second hop can increase latency or trigger spam filters if your domain reputation is newer than the affiliate network’s.

    If you’re solo and publishing sporadically, the maintenance overhead isn’t worth it. Cloaking plugins need updates. Redirects need testing. If a slug breaks and you don’t notice for two months, you’ve lost clicks and trust.

    If the affiliate program forbids it, don’t bother trying to hide the relationship. Amazon’s SiteStripe links work fine as-is. Trying to mask them risks more than you gain.

    And if you’re linking to a brand readers already recognize—Stripe, Shopify, Adobe—there’s no branding advantage. stripe.com is more trustworthy than yoursite.com/stripe to someone who’s never heard of you.

    Implementation trade-offs

    WordPress plugins like Pretty Links and ThirstyAffiliates handle the redirect and log clicks in your database. Easy to set up, but you’re querying your own database on every click. High-traffic affiliates can strain shared hosting if you’re not caching redirects.

    Link management tools like Rebrandly or Short.io offload the redirect to their infrastructure. Faster, more reliable, but you’re paying monthly and trusting a third party with your click data. If they change pricing or shut down, you’re migrating hundreds of links.

    Server-level redirects—defined in .htaccess or Nginx config—are fastest and least fragile, but require manual editing every time you add a link. Fine if you have six evergreen affiliate relationships. Unworkable if you’re testing new offers weekly.

    Pick the method that matches your volume and technical comfort. If you’re managing fewer than 20 affiliate links, a plugin is fine. If you’re running a deals site with 200+ partners, invest in dedicated infrastructure or a SaaS redirect service.

    Got a question about affiliate operations, tracking, or compliance? Reply to this email—operator stories make the best future articles.

  • WordPress staging environments: what they protect and what they miss

    WordPress staging environments: what they protect and what they miss

    Most WordPress hosting panels now ship with a one-click staging environment. You spin up a copy of your live site, test a plugin update or theme tweak, and push changes once you’re confident nothing breaks.

    It’s a smart workflow. But staging isn’t a perfect safety net, and treating it like one leads to false confidence and real downtime.

    Here’s what staging environments actually protect you from—and the failure modes they quietly ignore.

    What staging catches reliably

    Staging environments excel at isolating code-level conflicts. If you’re updating a plugin that hasn’t been touched in eighteen months, a staging site will surface PHP errors, broken shortcodes, and layout shifts before they hit your readers.

    You’ll also catch visual regressions. A theme update that changes your header structure or removes a custom CSS class will show up immediately. Same with a page builder update that reformats your landing pages.

    And staging is useful for workflow rehearsal. If you’re migrating from one form plugin to another, or restructuring your permalink settings, you can walk through the entire process without risking your live site’s SEO or user experience.

    Most managed WordPress hosts—BigScoots included—let you clone your production database and files in under a minute. The staging site runs on the same server stack, so you’re testing against a nearly identical environment.

    What staging misses entirely

    Staging environments don’t replicate third-party API behaviour. If your site connects to a payment processor, email service, or analytics platform, those integrations either won’t work in staging (because the API keys are sandboxed) or they’ll work differently (because staging traffic doesn’t match production load).

    You also can’t test caching behaviour accurately. Most staging environments disable caching plugins or CDN layers by default. That means a change that works perfectly in staging might still break when it hits your live site’s edge cache or object cache.

    And staging won’t catch performance problems under real traffic. A database query that runs in 200ms on a staging site with ten test posts might balloon to two seconds on a production site with ten thousand posts and a dozen concurrent users.

    Finally, staging environments don’t protect you from deployment errors. If your host’s push-to-live function skips a file, overwrites a manual edit, or fails to flush the cache, you won’t know until it’s live.

    When to test in staging—and when to skip it

    Use staging for major version updates: WordPress core, your theme, or any plugin that touches your site’s critical path. These are high-risk changes that can break layouts, disable forms, or trigger fatal errors.

    Also use staging for structural changes: switching themes, adding a new page builder, or enabling a plugin that injects code into every page. These changes touch too many files to trust without rehearsal.

    But skip staging for low-risk content edits. If you’re tweaking a blog post, updating a menu link, or uploading a new image, you’re adding friction without reducing risk. Make the change live, check it in an incognito window, and move on.

    And don’t rely on staging for plugin settings changes. Most plugins store settings in the database, and pushing changes from staging to production will either overwrite your live settings or skip them entirely. Test those directly in production—ideally during low-traffic hours.

    The non-obvious tip: test the push itself

    Most hosting platforms offer a “push to live” button that syncs your staging database and files back to production. But that push process isn’t guaranteed to be lossless.

    Before you push a major update, export your live database and store a backup locally. Then push from staging and immediately check three things: your homepage loads, your contact form works, and your most recent blog post displays correctly.

    If any of those fail, you’ll know within seconds—not hours later when a reader emails you.

    And if your host doesn’t offer staging environments, don’t build one manually. The risk of misconfiguring file permissions, breaking symlinks, or syncing the wrong database table outweighs the benefit. Either upgrade to a host that includes staging as a managed feature, or test updates during off-peak hours and keep a recent backup within reach.

    One Two Three Send covers WordPress hosting, email infrastructure, and every other tool solo operators rely on. Subscribe to get one operator-focused article every day—no fluff, no affiliate spam, just the details that matter.

  • Substack Notes vs. LinkedIn posts: which content strategy sticks

    Substack Notes vs. LinkedIn posts: which content strategy sticks

    If you’re running a content business in 2026, you’ve probably been told to post everywhere. But Substack Notes and LinkedIn represent two fundamentally different distribution strategies—and choosing the wrong one wastes time you don’t have.

    Both promise organic reach. Both claim to connect you with your audience. But the mechanics, the audience behavior, and the outcomes differ enough that treating them as interchangeable is a mistake.

    Audience intent: browsing vs. networking

    LinkedIn users open the app to see what’s happening in their professional network. They’re looking for career updates, industry commentary, and light business education. The platform rewards polish and positioning. A well-timed post about a lesson learned or a contrarian industry take can reach tens of thousands of impressions if it hits the algorithm right.

    Substack Notes users are readers first. They’re browsing updates from writers they already follow or discovering new ones through restacks. The feed skews literary, opinionated, and less corporate. A Note performs when it sounds like a person talking to other people—voice matters more than credentials.

    This difference shapes what works. LinkedIn favors declarative statements, clear takeaways, and content that signals expertise. Notes favor texture, specificity, and the kind of observational writing that makes someone want to read more of your work.

    Distribution mechanics: algorithm vs. restack

    LinkedIn’s algorithm optimizes for engagement velocity. If your post gets comments and shares in the first hour, it gets pushed to a wider audience. That means timing matters. Posting at 8 a.m. Eastern on a Tuesday will outperform the same post at 9 p.m. on a Saturday.

    The algorithm also favors native content. Text posts outperform link posts. If you’re driving traffic to your newsletter, you’ll get better reach by posting the insight directly on LinkedIn and mentioning the newsletter in a comment, rather than leading with a link.

    Substack Notes works differently. Distribution is driven by restacks—essentially retweets—and by how many of your subscribers have the Substack app installed. If your list is small or your readers don’t use the app, your Notes won’t travel far. But if your audience is active on Substack, a single restack from a popular writer can send your Note to thousands of new readers.

    Notes also lack an algorithmic feed in the traditional sense. They’re chronological within the subset of people you follow and discover. That makes timing less critical but makes your existing network more important.

    Conversion behavior: who subscribes?

    LinkedIn traffic tends to bounce. A viral post can send thousands of profile views, but converting those views into newsletter subscribers requires a very clear call-to-action and a compelling reason to leave the platform. Most LinkedIn users treat the platform as a feed, not a gateway.

    Substack Notes traffic converts better because the action you’re asking for—subscribe to this writer—is native to the platform. If someone likes your Note, subscribing is one tap. The friction is lower, and the context is already literary.

    That said, LinkedIn’s audience is larger and less saturated. A well-executed content strategy there can build authority and inbound opportunities that don’t require newsletter conversion—consulting leads, partnership inquiries, speaking invitations.

    Which platform to prioritize

    If your business model depends on growing a subscriber base quickly and you’re already writing regularly, prioritize Notes. The conversion path is shorter, and the audience is primed to subscribe. Spend 10 minutes a day sharing observations, restacking writers you admire, and engaging with your audience there.

    If your business model depends on authority and inbound opportunities—if you’re positioning yourself as an expert, building a personal brand, or selling services—LinkedIn is the better long-term play. Post two to three times a week, optimize for the algorithm, and treat the platform as top-of-funnel awareness, not direct conversion.

    Most indie operators don’t have time for both. Pick the one that aligns with how you make money, and ignore the other until you’ve exhausted the first.

    Want more comparisons like this? Subscribe to One Two Three Send and get operator-focused breakdowns of the tools and tactics that actually move the needle.

  • Stop paying for recurring subscriptions you forgot about

    Stop paying for recurring subscriptions you forgot about

    Most solo operators running online businesses are paying for between three and seven software subscriptions they either forgot about or no longer need. The average waste sits between $80 and $200 per month — enough to cover a year of decent hosting or a professional email service.

    This isn’t about extreme frugality. It’s about operational hygiene. Every unused subscription is a small leak in your business, and those leaks compound over time. Here’s how to audit your stack, cut the bloat, and build a system that prevents it from creeping back.

    Pull your transaction history from every payment method

    Start with your credit cards, PayPal, and any business accounts you use for software purchases. Download the last six months of transactions and filter for anything that repeats monthly or annually.

    Look for:

    • Charges under $20 — these are easy to miss and often auto-renew without warning.
    • Annual renewals you forgot about. A $200 charge in May for a tool you stopped using in February hurts twice.
    • Services billed through aggregators like Paddle or FastSpring, which obscure the vendor name on your statement.

    If you use a tool like Stripe for your own revenue, check your outgoing subscriptions too. Some operators set up recurring payments for white-label services or API access and forget they’re still active.

    Match every charge to a current use case

    Open a spreadsheet. List every recurring charge, the amount, the billing cycle, and — most importantly — the last time you actually used it.

    For each subscription, ask:

    • Have I logged in during the past 30 days?
    • Does this tool solve a problem I still have?
    • Could I replace this with a free alternative or a tool I’m already paying for?

    Common offenders include:

    • Design tools you used once for a logo refresh (Canva Pro, Adobe Creative Cloud).
    • SEO or analytics platforms you check twice a year (Ahrefs, Semrush).
    • Social media schedulers for platforms you stopped posting to (Buffer, Publer).
    • AI assistants you signed up for during a launch, then replaced with something else.

    If you haven’t touched it in 60 days and can’t articulate a specific upcoming use case, kill it.

    Downgrade before you cancel

    Some tools offer free tiers that cover 80% of what you need. Before you cancel outright, check if a downgrade makes sense.

    Examples:

    • Most email platforms (MailerLite, Brevo, Beehiiv) have generous free plans for lists under 1,000 subscribers.
    • Analytics tools like Plausible and Fathom offer lower-tier plans if you’re tracking fewer than 10,000 monthly pageviews.
    • Hosting providers often let you move to a cheaper plan if your traffic dropped or you consolidated sites.

    Downgrading keeps your account active, preserves your data, and gives you a fallback if you need to scale back up. Canceling outright sometimes means losing historical data or having to re-integrate from scratch later.

    Set a calendar reminder to repeat this every quarter

    Subscription bloat isn’t a one-time problem. New tools creep in during launches, experiments, or when you’re troubleshooting something urgent. Three months later, you’ve forgotten why you signed up.

    Block 30 minutes every quarter to repeat this audit. Use the same spreadsheet. Update your current charges, check usage, and cut anything that’s drifted out of your workflow.

    If you’re using a tool like Notion or Airtable to manage your business operations, add a “Software Stack” table with columns for cost, renewal date, and last-used date. Set up an automation (via Zapier or Make) to flag anything that hasn’t been marked “used” in 45 days.

    One rule to prevent future bloat

    Before you sign up for any new paid tool, add a note in your calendar for 60 days out: “Still using [Tool Name]?”

    If the answer is no, cancel before the second billing cycle hits. Most SaaS tools hook you with a generous trial or a strong first-month use case, then fade into the background as your workflow shifts. The 60-day check catches that drift before it costs you six months of fees.

    Running lean doesn’t mean running cheap. It means every dollar you spend has a job. If a tool isn’t doing that job, cut it and redirect the budget to something that moves your business forward.

    Want more operator-focused breakdowns like this? Subscribe to One Two Three Send and get one article like this in your inbox every week — no fluff, no filler, just the tools and tactics that matter for solo operators running content businesses.

    Heads up — some links in this article are affiliate links. If you sign up through them, we may earn a small commission at no extra cost to you. We only recommend tools we use ourselves.

  • ConvertKit’s subscriber score: how it ranks engagement and why it’s wrong

    ConvertKit quietly calculates an engagement score for every person on your list. It’s a single number—0 to 100—that’s supposed to tell you who cares and who doesn’t. The platform uses it to sort subscribers in reports, flag cold contacts, and guide re-engagement decisions.

    Most operators never look at it. The ones who do often misread what it measures—and make pruning or segmentation calls based on incomplete signals.

    Here’s what the score actually tracks, when it’s useful, and where it leads you astray.

    What drives the score

    ConvertKit’s engagement score weighs three behaviors:

    • Opens: How often a subscriber opens your emails in the past 90 days.
    • Clicks: How often they click links inside those emails.
    • Recency: How recently they’ve done either.

    Opens carry the most weight. A subscriber who opens every email but never clicks will score higher than someone who clicks occasionally but skips half your sends. Recency acts as a multiplier—someone who opened yesterday gets a bump over someone who opened 80 days ago, even if their long-term open rate is identical.

    The score doesn’t consider:

    • Whether they bought something
    • Whether they replied to an email
    • Whether they visited your site via a link (unless they also clicked in the email)
    • How long they’ve been subscribed

    It’s a deliverability proxy, not a business metric. ConvertKit designed it to help you identify contacts who hurt your sender reputation—not contacts who drive revenue.

    When the score matters

    The score is useful in two narrow scenarios.

    Pre-pruning cold contacts. If you’re preparing to scrub your list, sort by engagement score and review everyone below 20. These are the people who haven’t opened or clicked in months. Removing them improves your open rate and keeps inbox providers from flagging your domain. Just don’t auto-delete based on score alone—check signup date and source first. A subscriber who joined two weeks ago and hasn’t engaged yet isn’t cold; they’re new.

    Segmenting for re-engagement campaigns. Run a win-back sequence to subscribers scoring 10–30. They’re not dead, but they’re fading. A subject line refresh, a content pivot, or a simple “still interested?” email can pull them back. Anyone below 10 is harder to recover and may not be worth the send cost.

    Where the score misleads

    The engagement score breaks down when you treat it as a proxy for value.

    High scorers aren’t always your best subscribers. Someone who opens every email but never buys, never replies, and never shares your work scores higher than someone who buys twice a year but only opens when they need something. ConvertKit can’t see purchase behavior unless you tag it manually—and even then, it doesn’t factor into the score.

    Low scorers aren’t always dead weight. Plenty of valuable subscribers skim subject lines in their inbox and only open when a topic hits. They might visit your site directly, bookmark your archive, or consume your content via RSS. Their engagement score tanks, but they’re active in ways the platform can’t measure.

    The 90-day window hides seasonality. If you run a tax-prep newsletter, subscribers who engage in February and March will score poorly in June—even though they’re likely to come back next year. A hard cutoff at 90 days doesn’t account for cyclical engagement.

    What to use instead

    If you want to identify your most valuable subscribers, layer in context the score doesn’t capture:

    • Tag purchases and replies. Create segments for buyers and people who’ve replied to a broadcast. These are your highest-intent contacts, regardless of open rate.
    • Track link clicks by type. ConvertKit lets you filter by clicked link. Someone who clicks affiliate links or product pages is more valuable than someone who clicks every “read more” button.
    • Monitor unsubscribe timing. If low-engagement subscribers stick around for months without unsubscribing, they’re choosing to stay. That’s signal, even if they’re not opening.

    The engagement score is a starting point, not a verdict. Use it to spot patterns, but don’t let it override what you know about how your audience actually behaves.

    Want more tool breakdowns like this? Subscribe to One Two Three Send for weekly deep-dives on the features that shape how you run your online business—no fluff, just the mechanics that matter.

  • Semrush Position Tracking: what it watches and what it misses

    Semrush’s Position Tracking tool is one of the most widely used rank monitors in the SEO world. You feed it a list of keywords, connect your domain, and it checks where you rank every day. Simple premise. But the data it returns isn’t as complete—or as current—as most operators assume.

    If you’re basing content decisions, client reports, or traffic forecasts on Position Tracking alone, you’re working with a partial picture. Here’s what the tool actually does, where it breaks down, and how to fill the gaps.

    What Position Tracking monitors

    Semrush checks your rankings for a predefined list of keywords. You choose the keywords, set a target domain or subdomain, pick a country and device type (desktop or mobile), and the tool runs a daily check. Results appear as a line graph with position changes, estimated traffic, and visibility scores.

    It’s useful for tracking a curated set of high-priority terms—your ten core money keywords, your brand terms, the handful of informational queries that drive most of your traffic. If you know exactly what you want to rank for, Position Tracking gives you a clean dashboard.

    The tool also flags SERP features: if your keyword triggers a featured snippet, People Also Ask box, or local pack, Semrush notes it. You can filter by feature type and see which queries offer those opportunities.

    Where it goes blind

    Position Tracking only watches the keywords you tell it to watch. It won’t surface new queries you’re ranking for, seasonal spikes in tangential terms, or long-tail variations that suddenly start converting. If you don’t add a keyword manually, Semrush ignores it.

    That’s the opposite of how Google Search Console works. GSC shows you actual queries people used to find your site, even if you’ve never thought to track them. Position Tracking shows you only the keywords you already knew to care about.

    The second blind spot: Semrush checks rankings once per day, usually in the early morning UTC. If Google runs a volatile update, shuffles results midday, or personalises rankings based on user context, you won’t see it. The tool reports one snapshot per 24 hours. For stable, evergreen content, that’s fine. For news, trending topics, or anything tied to real-time search behaviour, it lags.

    Third issue: Semrush pulls rankings from a standardised environment—no personalisation, no location refinement beyond country-level targeting, no search history. Real users see results shaped by dozens of signals. Position Tracking gives you the cleanest possible view, which is also the least representative.

    When to rely on it (and when not to)

    Position Tracking works best when you have a short, stable list of target keywords and you want to monitor competitive movement or the impact of on-page changes. If you optimise a product page and want to see whether it moves from position 8 to position 4 over the next two weeks, this tool will catch it.

    It’s also useful for client reporting when you need a consistent, branded dashboard. The visibility score and traffic estimates give non-technical stakeholders something to latch onto, even if the numbers are modelled rather than measured.

    But don’t use Position Tracking as your primary traffic diagnostic. If organic sessions drop 20% in Search Console, Position Tracking might show no movement at all—because the traffic came from keywords you weren’t monitoring, or because the drop happened outside your tracked keyword set.

    And don’t assume the estimated traffic figure is accurate. Semrush models it based on CTR curves and search volume data, both of which are approximations. Actual clicks depend on SERP layout, brand recognition, title appeal, and a dozen other factors the tool can’t see.

    How to fill the gaps

    Run Position Tracking alongside Google Search Console, not instead of it. Use GSC’s Search Results report to identify which queries are actually driving impressions and clicks, then add high-performers to your Position Tracking list. That way, you’re monitoring the terms that matter, not just the ones you guessed would matter six months ago.

    If you’re tracking a large keyword set—say, 500+ terms—set up automated exports or use Semrush’s API to flag significant changes. Manually scanning a long list every day is a waste of time. Build a filter or script that surfaces keywords that moved five positions or more in the last week.

    For volatile niches—crypto, trending news, seasonal products—check rankings manually in an incognito window or use a tool like Accuranker that pings results multiple times per day. Semrush’s daily snapshot won’t catch intraday swings.

    One non-obvious tip: use Position Tracking’s competitor comparison feature to monitor domains you’re directly competing with for the same keyword set. Add up to five competitors, and Semrush will show you their rankings alongside yours. If a competitor jumps ten positions overnight, you’ll know to investigate their page—they either updated content, built links, or benefited from an algorithm shift. That signal is often more valuable than your own ranking data.

    If you found this useful, subscribe to One Two Three Send for weekly breakdowns of the tools and tactics that actually move the needle. We cover the full stack—SEO, email, AI, hosting, monetisation—without the fluff.

    Position Tracking is a solid tool. Just don’t mistake its clean, curated view for the messy, comprehensive reality of how search traffic actually lands on your site.

  • Transactional email vs. marketing email: when to use which

    Transactional email vs. marketing email: when to use which

    Most solo operators start with one email service provider and route everything through it: welcome emails, password resets, weekly newsletters, product updates, receipts. It’s simple, it works, and for a while there’s no reason to change.

    Then something breaks. A welcome email arrives four hours late. A password reset never shows up. Or worse: your newsletter gets a spam complaint, and suddenly all your emails—including order confirmations—land in the promotions tab or get delayed.

    The root issue is conflating two fundamentally different types of email: transactional and marketing. They serve different purposes, have different legal rules, and need different infrastructure. Mixing them creates risk you don’t see until it costs you money.

    What makes an email transactional

    Transactional emails are triggered by a user action and contain information the recipient explicitly requested or needs to complete that action. Examples:

    • Password resets and login links
    • Order confirmations and receipts
    • Account notifications (payment failed, subscription renewed)
    • Download links after a purchase
    • Two-factor authentication codes

    These emails are expected. The user did something, and your system is responding. CAN-SPAM and GDPR treat them differently because they’re not commercial messages—they’re functional infrastructure.

    Marketing emails are everything else: newsletters, product announcements, promotional offers, content roundups. They require explicit consent in most jurisdictions, must include an unsubscribe link, and are subject to stricter anti-spam rules.

    The line blurs with hybrid emails—like a receipt that also suggests related products—but if the primary purpose is commercial, it’s marketing.

    Why reputation matters more than you think

    Email service providers (Gmail, Outlook, Yahoo) track sender reputation at the domain and IP level. If you send both transactional and marketing email from the same domain, a single spam complaint on your newsletter can damage deliverability for your password resets.

    This is why companies like Stripe and Shopify send transactional email from dedicated domains (receipts come from @stripe.com, but newsletters come from subdomains or separate services). They’re isolating reputation risk.

    For solo operators, the practical version of this is: use a dedicated transactional ESP for critical emails. Route your password resets, purchase confirmations, and login links through a service built for speed and reliability. Send your newsletter through a platform optimized for bulk sends, engagement tracking, and unsubscribe management.

    Postmark is the gold standard here—transactional-only, no marketing allowed, designed for sub-second delivery. Pricing starts at $15/month for 10,000 emails, and because these are triggered sends (not bulk), most operators stay under 1,000/month. If you’re on WordPress and using a membership plugin or WooCommerce, you’re already generating transactional email. Route it through the right pipe.

    When to split your setup

    You don’t need two ESPs on day one. If you’re pre-revenue or sending fewer than 100 emails a month total, the complexity isn’t worth it. But you do need to split when:

    • You’re processing payments or running a membership site (receipts and login emails must arrive instantly)
    • Your newsletter list is growing past 500 subscribers (spam complaints become statistically inevitable)
    • You’ve had a deliverability issue with transactional email (password resets delayed, order confirmations in spam)
    • You’re sending time-sensitive notifications (webinar reminders, expiring cart links)

    The cost of a delayed or missing transactional email—lost sale, frustrated customer, support ticket—is higher than the $10–15/month for a dedicated service.

    How to route it correctly

    If you’re on WordPress, install a transactional plugin (WP Mail SMTP, Postmark’s official plugin, or Brevo‘s SMTP add-on) and configure it to handle system emails. Your membership plugin, WooCommerce, and form notifications should route through this.

    Your newsletter platform (Beehiiv, MailerLite, ConvertKit) handles everything else: weekly sends, product launches, content updates. These platforms are built for engagement tracking, A/B testing, and list segmentation—features you don’t need (and don’t want) in a password reset.

    If you’re not on WordPress, check your app’s email settings. Most SaaS tools let you configure SMTP credentials. Point transactional sends to your transactional ESP, and keep marketing sends in your newsletter tool.

    One non-obvious tip: set up separate subdomains. Send transactional email from mail.yourdomain.com and newsletters from news.yourdomain.com. This isolates reputation at the DNS level and makes it easier to debug deliverability issues later.

    If you’re routing everything through one service today and haven’t had a problem yet, you’re not wrong—you’re just early. But when you hit the threshold where mixing email types starts costing you conversions, you’ll know exactly what to fix.

    Got a question about your email setup? Reply to this email—I read every message and often turn answers into future pieces.

    Heads up — some links in this article are affiliate links. If you sign up through them, we may earn a small commission at no extra cost to you. We only recommend tools we use ourselves.

  • Claude’s prompt caching: what it saves and when to turn it on

    Claude’s prompt caching: what it saves and when to turn it on

    Claude introduced prompt caching in late 2024, and most solo operators still don’t use it — even when they’re burning through API credits on repetitive tasks.

    The feature lets you cache large chunks of context (style guides, product catalogs, documentation) so Claude doesn’t re-read them on every request. When it works, it cuts costs by 90% and speeds up responses. When it doesn’t, you pay a caching penalty for no benefit.

    Here’s how to know which side you’re on.

    How prompt caching actually works

    Every time you send a prompt to Claude, the API charges you for input tokens (what you send) and output tokens (what Claude generates). With caching enabled, Claude stores the first part of your prompt — the part that doesn’t change between requests — and reuses it for up to five minutes.

    Cached input tokens cost 90% less than regular input tokens. But there’s a catch: the cached section must be at least 1,024 tokens, and it has to appear at the start of your prompt. If your repeated context is buried mid-prompt, caching won’t trigger.

    Most operators structure their prompts backwards. They put the variable part (the user question, the draft to edit, the product name) first, then append the static instructions. That ordering breaks caching.

    To make caching work, flip it: static context first, variable input last.

    When caching saves real money

    Caching pays off when you’re running the same large prompt dozens or hundreds of times per day. Three scenarios where it matters:

    Batch content editing. You’re rewriting 50 product descriptions using the same brand voice guide (3,000 tokens). Without caching, you pay full price for that guide on every request. With caching, you pay once, then 10% for the next 49.

    Structured data extraction. You’re parsing invoices, receipts, or support tickets into JSON using the same schema definition (2,000 tokens). Each parse job reuses the schema. Cache it.

    Context-heavy chat interfaces. You’re building a support bot that references your entire help center (10,000 tokens) on every question. Cache the help center, send only the user’s question as new input.

    If you’re running fewer than 10 requests per day with the same context, caching won’t move the needle. The setup overhead isn’t worth it.

    How to structure prompts for caching

    Here’s the wrong way (no caching):

    User question: [variable input]
    Instructions: [3,000-token style guide]

    Here’s the right way (caching triggers):

    Instructions: [3,000-token style guide]
    User question: [variable input]

    In the API request, you mark the instructions block as cache_control: {"type": "ephemeral"}. Claude caches everything up to that marker. On the next request, if the cached section is identical, you pay the reduced rate.

    One non-obvious detail: the cache expires after five minutes of inactivity. If your workflow runs requests in bursts with long gaps, you’ll pay the caching write cost repeatedly without ever hitting the cache. Caching works best for sustained, high-frequency use — not sporadic jobs.

    When caching costs more than it saves

    Caching isn’t free. The first time Claude writes to the cache, you pay a 25% premium on those tokens. If you send a 5,000-token prompt once and never reuse it, you’ve paid extra for nothing.

    You also lose caching benefits if you tweak the cached section between requests. Changing even one word in your style guide invalidates the cache and triggers a new write. If you’re still iterating on your prompt structure, wait until it’s stable before enabling caching.

    And if your repeated context is small (under 1,024 tokens), caching won’t activate at all. The feature is designed for large, static blocks — not short instructions.

    What this means for your workflow

    Most solo operators should ignore caching until they hit a clear threshold: same large prompt, 20+ times per day, stable structure. Below that, the cost savings are negligible and the cognitive overhead of restructuring prompts isn’t worth it.

    But if you’re running batch jobs, building repeatable AI workflows, or prototyping a product that calls Claude hundreds of times, caching can cut your API bill in half. Just don’t bolt it onto your existing prompts without restructuring them first.

    Using Claude for high-volume workflows? Subscribe to One Two Three Send for more breakdowns of AI features that actually matter to solo operators.

    Heads up — some links in this article are affiliate links. If you sign up through them, we may earn a small commission at no extra cost to you. We only recommend tools we use ourselves.

Powered by
►
Necessary cookies enable essential site features like secure log-ins and consent preference adjustments. They do not store personal data.
None
►
Functional cookies support features like content sharing on social media, collecting feedback, and enabling third-party tools.
None
►
Analytical cookies track visitor interactions, providing insights on metrics like visitor count, bounce rate, and traffic sources.
None
►
Advertisement cookies deliver personalized ads based on your previous visits and analyze the effectiveness of ad campaigns.
None
Powered by