mbfinotti/advertising-skills / skills/retargeting-funnel
retargeting-funnel
VerifiedDesign a multi-stage retargeting sequence from a site's own funnel data - recency windows and behavioural-depth tiers per stage, a message and offer ladder for each stage, mutually exclusive audiences with exclusion logic, and per-stage frequency caps. Use whenever the user mentions retargeting or remarketing, cart or form abandonment, how long a retargeting window should be, frequency caps, a retargeting audience that is too small, or ads still showing to people who already bought - even if they never say 'funnel'. Covers B2B and B2C, and produces a stage-by-stage plan rather than campaigns built inside an ad platform. Do NOT use to design cold prospecting audience tiers - use mbfinotti/advertising-skills@ad-audience-targeting instead.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 31.4 KB |
| evals/evals.json | config | 36.8 KB |
| references/platform-constraints.md | markdown | 13.5 KB |
| references/worked-sequences.md | markdown | 9.9 KB |
evals/evals.json
View on GitHub{
"skill_name": "retargeting-funnel",
"evals": [
{
"id": 1,
"prompt": "I run Emberline, a DTC scented-candle store. We get about 140k sessions a month and roughly 1,900 orders. I pulled the Google Analytics time-lag report like our consultant asked: 58% of purchases happen within 1 day of the first visit, 86% within 5 days, 96% within 12 days, and almost nothing converts after day 30. Here is the retargeting plan I sketched - five stages, each its own campaign: (1) all site visitors 0-30 days, (2) product page viewers 0-30 days, (3) cart abandoners 0-30 days, (4) checkout abandoners 0-60 days, (5) past buyers 0-180 days. Can you refine the windows and the messaging for each stage?",
"expected_output": "A redesigned sequence with the hot edge derived from the 80th converter percentile (about 5 days), fewer collapsed stages, windows assigned inversely to depth, mutual exclusion and converter suppression with an explicit written window, past buyers moved to a win-back tier, and a B2C message ladder with any incentive on the final stage only.",
"files": [],
"expectations": [
"Derives the hot-window edge from the converter time-lag percentiles, placing it at roughly 4-7 days (the ~80% point sits near 5 days), instead of keeping the proposed 30-day windows",
"States explicitly that stage boundaries come from Emberline's own lag distribution and that platform default retention windows are arbitrary relative to this specific business",
"Collapses the five-stage plan to fewer conversion stages (a hot/warm/win-back shape) because 80%+ of converters buy within 7 days",
"Gives the collapse reason: more stages than the conversion cycle supports produces empty or wasted stages",
"Rejects the 0-60-day window on checkout abandoners: deep-intent actions get the shortest windows because intent decays fast",
"Assigns windows inversely to behavioural depth: cart/checkout tiers get the shortest windows and shallow all-visitor tiers the longest",
"Claims stages deepest-first along the depth order cart or checkout start above product page viewer above any visitor",
"Moves past buyers out of the conversion ladder into a separate win-back or replenishment tier",
"Makes stages mutually exclusive - each stage excludes all deeper and fresher stages - and points out that in the draft one user sits in several stages at once",
"Names the self-competition mechanism: overlapping stages bid against each other in the same auction and inflate CPM with zero incremental reach",
"Excludes converters from every stage and writes an explicit converter-exclusion window as a number of days, not an open-ended exclusion",
"Proposes an auditable audience naming convention that distinguishes inclusion audiences from exclusion audiences (for example RTG_ and EXCL_ prefixes)",
"Plans at least 3 distinct creative concepts per stage",
"Gives each stage a distinct message job following the B2C slot order reminder, then social proof, then objection handling, then incentive, with any discount confined to the final stage"
]
},
{
"id": 2,
"prompt": "We're Corvid Systems - compliance software for mid-market banks, deals run 4 to 6 months with a security review in the middle. I'm setting up remarketing audiences this week and I just need the standard windows: what does everyone use for site visitors, feature page viewers, and demo requesters? We don't have a time-to-conversion report and honestly I don't have time to pull one before Friday.",
"expected_output": "A refusal to hand over generic windows: either the user exports lag data, or the plan proceeds on a named proxy (the CRM sales-cycle export, given the multi-month cycle) with every boundary flagged provisional, sized to the 4-6-month cycle, list-based at the long end, with a no-discount B2B ladder.",
"files": [],
"expectations": [
"Does not hand over generic industry-standard windows as the answer; states plainly it will not guess windows without time-lag data",
"Offers the two-way choice: export a time-to-conversion or sales-cycle report, or proceed on a named proxy with every boundary flagged as provisional",
"Names the proxy in the output rather than silently substituting it for real data",
"Ranks the CRM sales-cycle export above an analytics lag report for this account, because a 4-6-month cycle outruns what the pixel can see",
"Flags every proxy-derived boundary as provisional, to be re-derived once real lag data exists",
"Sizes windows to the multi-month cycle, on the order of 90/180/365 days, not 30-day defaults",
"Warns that a default-length window on a multi-month cycle silently drops most of the pipeline",
"Recommends list-based (CRM or first-party upload) audience membership rather than pixel-based at the long end, citing browser privacy limits on long cookie windows",
"Suggests CRM deal stage (evaluation, proposal, negotiation) as a depth signal, since the CRM knows more about depth than the pixel",
"Builds the B2B ladder with no discount rung; the incentive-equivalent rung is a lower-friction ask such as an assessment, audit, or tailored demo",
"States which mechanics stay identical to B2C (mutual exclusion, converter suppression, the after-exclusion size check) so the plan is not read as a softened B2C plan"
]
},
{
"id": 3,
"prompt": "Growth lead at Marrow & Sage here, DTC cookware. Cart abandonment sits at 74%, so we built a coupon ladder: 10% off the day after abandonment, 15% at day 7, 20% at day 14. The coupon ads convert great. One weird thing - since launching it three months ago, abandonment has climbed to 79%. My read is we need the discounts sooner and bigger to catch people before they cool off. Tighten up the ladder for me?",
"expected_output": "A rejection of sooner-and-bigger: the rising abandonment is diagnosed as discount-trained abandonment, the discount is pushed to the final stage only, earlier stages get reminder then social proof then objection messages, repeat abandoners are excluded from incentives, and the un-training timeline is stated.",
"files": [],
"expectations": [
"Does not recommend moving discounts earlier or making them bigger",
"Diagnoses the abandonment rise from 74% to 79% after coupon launch as trained abandonment: customers learned that abandoning triggers a coupon",
"Restricts the discount to the final stage of the sequence only",
"Makes the first stage a plain reminder or dynamic item ad with no incentive",
"Fills the middle stages with non-price messages in the B2C slot order: social proof, then objection handling (shipping, returns, guarantee)",
"Excludes repeat abandoners from every incentive-carrying stage",
"Names the discount's structural costs: it spends margin every time it runs and is the hardest offer to withdraw once the market has learned it",
"Applies the principle that re-showing the same offer harder is the weakest stage design - change the offer or the angle instead",
"States that un-training the coupon behaviour takes on the order of a quarter, while the ladder change itself takes minutes",
"Adds converter exclusion on every stage with an explicit written window",
"Makes the day-band stages mutually exclusive so one abandoner cannot sit in the 10%, 15%, and 20% pools at once",
"Derives or requests the brand's own time-to-conversion data to set the stage boundaries instead of keeping the arbitrary 1/7/14-day rungs"
]
},
{
"id": 4,
"prompt": "We're Quillhaven, sales-assisted procurement SaaS, ACV around $40k. Traffic is about 11k sessions a month - 65% blog, roughly 700 people a month hit pricing, 1.5% request demos. I want a five-tier LinkedIn retargeting build: pricing visitors 30d, feature-page visitors 30d, blog readers 90d, video viewers 30d, webinar attendees 180d, each in its own campaign with manual bids. Launch is next month - I'll create the audiences the day we go live. Also want to seed a LinkedIn lookalike from the pricing-page audience. Thoughts?",
"expected_output": "A size-checked redesign: tiers checked against LinkedIn's 300-member floor after exclusions, merges or a single warm pool presented as the correct outcome, audiences created immediately because LinkedIn collection is non-retroactive, the lookalike corrected to predictive audiences, automated bidding for thin pools, and a no-discount B2B ladder.",
"files": [],
"expectations": [
"Checks every proposed tier against LinkedIn's minimum matched-audience size (300 members), measured after exclusions are applied, not before",
"States that a tier clearing the floor before exclusions can fall under it once fresher and deeper tiers and converters are carved out",
"Does not ship the five-tier plan as-is on 11k sessions: recommends merging tiers, and presents a two-stage or single-pool outcome as a correct result, not a failure",
"Repairs under-floor tiers in the order: widen the window, then broaden the trigger, then merge adjacent stages, then collapse to one combined warm pool",
"Corrects the launch-day audience plan: LinkedIn audiences are non-retroactive, collection starts only at creation, so every audience should be created now, before launch",
"Corrects the lookalike request: LinkedIn lookalike audiences were discontinued (February 2024) and replaced by predictive audiences",
"Presents the single combined broad warm pool as the legitimate default below the floor, moving to stratified tiers only once each stage clears its floor with room to spare",
"Recommends automated bidding rather than the proposed manual bids, because small LinkedIn pools underdeliver on manual bidding",
"Uses a B2B ladder with no discount rung, ordered proof, then objection, then ROI content, then the direct ask",
"Writes mutual exclusion between surviving stages and excludes current customers and converters on every stage",
"Sets a small-pool creative discipline: 3-5 rotating creatives with refresh on the order of every 2-3 weeks",
"Considers CRM deal stage or uploaded account lists as the depth signal for the long-window end rather than web recency alone",
"Asks for or derives time-to-conversion data instead of accepting the arbitrary 30/90/180-day windows",
"Warns that an under-floor stage either serves nothing or serves at punitive CPMs while the learning phase never exits"
]
},
{
"id": 5,
"prompt": "Quick one - I run retargeting for Fernwell, home fitness gear, on Meta with the sales objective. Warm audience is about 3,200 people, one video ad running to it. What frequency cap should I set in Ads Manager so we don't burn the audience out? Last week frequency hit 9, CTR dropped from 1.8% to 1.1%, and CPM went from $14 to $19.",
"expected_output": "An answer that no cap field exists on Meta's sales objective: a cap-proxy is defined instead, the given CTR and CPM moves are read as breached decay thresholds, the single-creative small pool is named as the structural cause, 3+ rotating concepts are required, and replacements launch alongside rather than as edits.",
"files": [],
"expectations": [
"States that Meta exposes no impression-cap field on the sales/conversion objective (only on reach or awareness-type objectives) and therefore does not instruct the user to set a cap in Ads Manager for this campaign",
"Defines a cap-proxy instead: the frequency reading at which to intervene, plus decay signals",
"Names the decay thresholds: CTR falling 15-20% or more and CPM rising 10% or more, measured against a 7-day rolling baseline",
"Applies the thresholds to the given numbers: the CTR drop (1.8% to 1.1%, about 39%) and the CPM rise ($14 to $19, about 36%) both breach them, so the cap is already breached in effect",
"Places frequency 9 against the retargeting operating band (roughly 2-4 watch, 4-6 warning, above 6 act): the account is past the act threshold",
"Identifies the structural cause: a roughly 3,200-person pool with one creative is an uncapped frequency machine regardless of settings - audience size and creative rotation are the real cap",
"Requires at least 3 distinct creative concepts in rotation for the stage",
"Warns that editing the live ad resets Meta's learning phase while pausing does not, so replacement creative launches alongside, not as an edit",
"Lists rising negative feedback (hides, see-less-of-this) as a third breach signal to monitor",
"States that retargeting tolerates higher frequency than prospecting because the audience knows the brand, but tolerance is not immunity",
"Sets a review cadence for the frequency and decay readings rather than a set-and-forget answer",
"Treats published frequency bands as starting points to recalibrate against the account's own decay curve, not as evidence-based constants",
"Connects budget to the pool: spend beyond reachable pool times capped weekly impressions times CPM only buys frequency past the cap, so the stage's budget is capped at that ceiling rather than raised"
]
},
{
"id": 6,
"prompt": "Board meeting fallout at Lanternway Outfitters (outdoor apparel, $48k/month paid budget, currently 75/25 prospecting-heavy). Retargeting ROAS shows 11x on the platform versus 1.9x for prospecting, and blended revenue has been flat all year. The board wants at least 50/50 next quarter. Retargeting pool is about 8,000 reachable people, CPM runs around $16, and we hold frequency to roughly 6 per person per week. Draft the new split for me.",
"expected_output": "A refusal grounded in arithmetic: the retargeting ceiling computes to roughly $750-800 a week (about $3,000-3,500 a month), the 11x ROAS is flagged as the cannibalisation signature given flat blended revenue, and an incrementality test with a named design and decision rule is scheduled before any budget moves toward retargeting.",
"files": [],
"expectations": [
"Does not deliver the requested 50/50 split; states that retargeting spend is bounded by pool size times frequency ceiling, not by ambition or platform ROAS",
"Computes the ceiling from the given numbers: 8,000 people times 6 impressions per week times $16 CPM comes to roughly $750-800 per week, about $3,000-3,500 per month",
"Caps the retargeting budget at that computed ceiling and states that excess budget forced in only buys frequency past the cap",
"Flags the 11x platform ROAS as suspect on its face: retargeting targets exactly the people most likely to convert on their own",
"Reads the combination of high platform ROAS and flat blended revenue as the cannibalisation signature: paying to intercept conversions that were coming anyway",
"Cites the scale of overstatement from large-scale randomized experiments: observational or platform attribution overestimated measured lift by roughly 7x to 9.5x",
"Schedules an incrementality test with a named design before moving budget toward retargeting",
"Selects the design by availability: a platform ghost-ad style lift test where offered, otherwise a randomized audience holdout withholding 10-20% of each stage",
"Includes the blended-efficiency metric, total revenue divided by total marketing spend, as the account-level sanity metric that drops the attribution layer entirely",
"States the decision rule: if holdout-measured lift is indistinguishable from zero, the retargeting budget folds back into prospecting regardless of platform ROAS",
"Treats the common 70/30 practice and the 50/50 move as conditional heuristics (justified only when traffic is high and conversion is the bottleneck), overridden here by the pool-bound ceiling",
"Brings in the reach-over-frequency objection (growth comes from reaching light and non-buyers) as a ceiling argument on the retargeting share of budget"
]
},
{
"id": 7,
"prompt": "Two audience problems at Vessel & Vine - we sell specialty coffee gear and beans, and a typical customer reorders about every 2 months. First: customers keep getting our dynamic ads for the exact grinder they bought the week before, and we're getting angry emails about it. Second, and I don't get this one: our prospecting campaigns' reach has been shrinking quarter after quarter for a year even though site traffic is stable. Our agency did set up a past-purchasers exclusion on Meta ages ago. Can you sort out our audience setup?",
"expected_output": "A two-part diagnosis: the dynamic ads need purchase-event suppression plus catalogue exclusions (audience exclusion alone does not stop the catalogue), and the shrinking prospecting reach is an over-long converter exclusion accumulating buyers - aggravated by Meta's purchase-audience retention extension to 730 days - fixed with an explicit window sized to the roughly 2-month repurchase cycle, plus a win-back tier.",
"files": [],
"expectations": [
"Fixes the bought-grinder problem with purchase-event suppression on the dynamic ads, not with converter audience exclusion alone",
"Adds catalogue-level exclusions (including out-of-stock items) so the catalogue cannot serve items the user already bought or cannot buy",
"Distinguishes the two mechanisms: audience exclusion controls who sees ads, catalogue suppression controls which items are shown, which is why the existing past-purchasers exclusion did not stop the grinder ads",
"Audits that converter exclusion actually exists on every stage and every channel rather than trusting that the agency's old setup covers everything",
"Diagnoses the shrinking prospecting reach as an over-long converter exclusion: the buyer pool accumulates and silently suppresses prospecting quarter over quarter",
"Connects the shrinkage to platform retention changes: Meta extended purchase-event audience retention to 730 days, auto-applied to existing audiences, so a purchaser exclusion built on it now suppresses about two years of buyers",
"Sets an explicit written converter-exclusion window sized to the roughly 2-month repurchase cycle (on the order of 90-180 days), not an open-ended exclusion",
"Names both failure directions of the window: too short re-ads recent buyers, too long silently chokes prospecting reach",
"Adds a standing review note to re-audit exclusion windows whenever a platform extends its retention defaults",
"Moves past purchasers beyond the exclusion window into a win-back or replenishment tier (the 2-month reorder cycle is a cross-sell or refill opportunity), outside the conversion ladder",
"Orders the fixes by waste removed per unit of effort: the suppression and audience-definition edits (minutes per channel) come first, before any funnel redesign",
"Proposes an auditable naming convention separating inclusion and exclusion audiences so any stage reads as include X minus Y minus converters",
"Verifies that the purchase event itself fires correctly as a prerequisite, since both the dynamic-ad suppression and the exclusion depend on it"
]
},
{
"id": 8,
"prompt": "My CFO at Gladehouse Software (self-serve team-collaboration tool) doesn't buy our retargeting numbers. The platform dashboard says retargeting did 14x ROAS last quarter and drove 640 signups, but total new revenue barely moved versus the quarter before we launched the campaigns. She wants proof the ads work. I was going to build a before/after comparison and maybe a regression of weekly signups on weekly spend - will that convince a finance audience?",
"expected_output": "A measurement plan that rejects before/after and observational regression, cites the RCT evidence that observational methods overstate lift severalfold, reads the flat-revenue-high-ROAS pattern as cannibalisation, and schedules a randomized incrementality test with a named design, holdout share, duration, and a pre-agreed decision rule, plus a blended-efficiency metric.",
"files": [],
"expectations": [
"Rejects the before/after comparison and the spend regression as proof: observational methods overestimated experimentally measured lift by roughly 7x to 9.5x in large-scale randomized trials",
"Cites the research base by name - the Gordon and Zettelmeyer et al. RCT comparison and/or the Blake, Nosko and Tadelis eBay brand-search result - rather than a vague studies-show",
"States why retargeting is the highest-risk case for inflated attribution: it targets exactly the people most likely to convert on their own",
"Reads the given pattern, 14x platform ROAS beside flat total revenue, as the signature of paying to intercept conversions that were coming anyway",
"Proposes a randomized incrementality experiment with a named design as the proof the CFO gets",
"Chooses the design correctly: a platform ghost-ad style lift test where the platform offers one, otherwise an audience holdout withholding a random 10-20% of each stage",
"Positions the geo or matched-market design as the heavy option (roughly a quarter of coordination), not the default unless matched-market infrastructure already exists",
"Includes the blended-efficiency sanity metric: total revenue divided by total marketing spend, which drops the attribution layer entirely",
"Pre-registers the decision rule: any stage whose holdout-measured lift is indistinguishable from zero is cut or restructured, no matter its platform ROAS",
"Does not use platform-reported conversions or ROAS as the success metric of the new measurement plan",
"Tracks new-vs-returning conversion share among the per-stage metrics",
"Sets the expectation that the experiment needs weeks of elapsed time before it answers"
]
},
{
"id": 9,
"prompt": "Planning retargeting for Thornbury Analytics - B2B data-quality platform. Our CRM says median first-touch-to-close is 95 days, 80th percentile around 150. About 55% of our traffic is Safari/iOS (design-agency-heavy audience). My plan: one Meta website-visitor audience set to 365 days and one Google remarketing list at 365 days, and we run book-a-demo ads to everyone in both. Simple and wide - anything wrong with that?",
"expected_output": "A correction on three fronts: the Meta 365-day website audience silently truncates to the 180-day ceiling, pixel windows beyond about a week undercount the Safari-heavy majority so the long end moves to CRM-list audiences, and the single book-a-demo message is replaced by a staged B2B ladder with depth tiers, exclusions, and a scheduled incrementality check.",
"files": [],
"expectations": [
"Flags that Meta standard website custom audiences cap at 180 days, so the requested 365-day Meta audience silently becomes a 180-day one - the ceiling truncates, it does not error",
"Distinguishes Google's remarketing-list ceiling (540 days, so 365 is possible there) and/or instructs verifying each platform's current retention ceiling against official documentation before building",
"Flags the Safari/iOS constraint: first-party cookies are capped around 7 days on Safari, so pixel-based windows beyond about a week undercount the 55% Apple-heavy traffic",
"Moves long-window membership to CRM-list or first-party uploaded audiences, which persist until removed, instead of the pixel",
"Derives the horizon from the CRM lag data given: the 80th percentile near 150 days sets an active-evaluation window on that order, rather than 365 days chosen for width or a 30-day default",
"Rejects book-a-demo-to-everyone as the single message: the direct ask converts intent already present and buys no new belief, so it cannot be the whole sequence",
"Builds the B2B ladder in the order proof, objection handling, ROI content, then the direct ask, with no discount rung",
"Splits the single audience into behavioural-depth tiers claimed deepest first (pricing or trial signals above feature pages above all visitors), or sequences by CRM deal stage",
"Offers the ranked privacy mitigations: enhanced or first-party conversion matching as the cheap first move, CRM-list audiences for long windows, server-side event delivery as the heaviest",
"Excludes current customers on every stage via a list-based exclusion, and makes the surviving stages mutually exclusive",
"Schedules an incrementality check (for example a 10-20% holdout) as part of the delivered plan, not as an afterthought",
"Warns that the plan needs the after-exclusion size check against each platform's delivery floor before shipping"
]
}
],
"trigger_queries": [
{ "query": "How long should my retargeting window be for cart abandoners?", "should_trigger": true },
{ "query": "Design a remarketing sequence for our Shopify store", "should_trigger": true },
{ "query": "People who already bought from us keep seeing our ads - how do I stop that?", "should_trigger": true },
{ "query": "What frequency cap should I use on my warm audiences?", "should_trigger": true },
{ "query": "My retargeting audience is too small to deliver on Meta - what do I do?", "should_trigger": true },
{ "query": "Set up a win-back ad campaign for lapsed customers", "should_trigger": true },
{ "query": "How do I structure retargeting stages for a 6-month B2B sales cycle?", "should_trigger": true },
{ "query": "Cart abandonment is 78% and I want to chase abandoners with ads - what's the plan?", "should_trigger": true },
{ "query": "Should cart abandoners and product viewers be in separate ad sets?", "should_trigger": true },
{ "query": "How do I keep my retargeting audiences from overlapping each other?", "should_trigger": true },
{ "query": "What's the right lookback window for website visitors on Google Ads remarketing?", "should_trigger": true },
{ "query": "Build me a message ladder for warm audiences - what do we show them second and third?", "should_trigger": true },
{ "query": "Our remarketing ads show the same offer over and over and users are annoyed", "should_trigger": true },
{ "query": "whats a good exclusion setup so buyers stop getting our ads", "should_trigger": true },
{ "query": "People who started the demo form but never booked - how do we get them back with ads?", "should_trigger": true },
{ "query": "How many days after someone views pricing should we keep showing them ads?", "should_trigger": true },
{ "query": "We retarget everyone who visited in the last 30 days with 10% off - good idea?", "should_trigger": true },
{ "query": "How do I sequence ads so people see social proof after the first reminder?", "should_trigger": true },
{ "query": "Is 30 days the right membership duration for my website custom audience?", "should_trigger": true },
{ "query": "Plan remarketing for our SaaS free-trial dropouts", "should_trigger": true },
{ "query": "My warm audience frequency is at 8 per week - is that too much?", "should_trigger": true },
{ "query": "How should retargeting differ between our B2B product line and our consumer app?", "should_trigger": true },
{ "query": "Customers who purchased last week still show up in our dynamic product ads - seeing the exact item they bought", "should_trigger": true },
{ "query": "How do I build an exclusion audience of purchasers?", "should_trigger": true },
{ "query": "Abandoned checkout ads - at what point do we offer the discount?", "should_trigger": true },
{ "query": "What behavioural tiers should my remarketing use - pricing viewers vs blog readers?", "should_trigger": true },
{ "query": "Give me a stage-by-stage plan to re-engage site visitors who didn't convert", "should_trigger": true },
{ "query": "Our retargeting ad set never exits the learning phase and delivery is stuck", "should_trigger": true },
{ "query": "How long do people stay in my remarketing audience after they visit?", "should_trigger": true },
{ "query": "We sell enterprise software - should our retargeting audiences come from the pixel or from Salesforce lists?", "should_trigger": true },
{ "query": "Someone visited pricing twice this week - what ad should they get vs a first-time blog reader?", "should_trigger": true },
{ "query": "How long should we suppress existing customers from seeing our ads after they buy?", "should_trigger": true },
{ "query": "Map out which ads people should see in week 1 vs week 3 after visiting our store", "should_trigger": true },
{ "query": "Do retargeting windows need to match our sales cycle length?", "should_trigger": true },
{ "query": "How aggressive can ad frequency be for people who already know our brand?", "should_trigger": true },
{ "query": "I want a remarketing funnel: visitors, engagers, cart, checkout - sanity check it?", "should_trigger": true },
{ "query": "Best practice for excluding converters on TikTok retargeting?", "should_trigger": true },
{ "query": "Our retargeting audience on LinkedIn is 200 people and won't serve", "should_trigger": true },
{ "query": "Set up abandonment ads for our booking flow - people bail at the payment step", "should_trigger": true },
{ "query": "After what point is a site visitor cold again and not worth retargeting?", "should_trigger": true },
{ "query": "Should we show discounts to cart abandoners immediately or wait?", "should_trigger": true },
{ "query": "Structure our warm-audience campaigns so each person only ever sits in one", "should_trigger": true },
{ "query": "quick q - remarketing membership duration on meta vs google, same or different?", "should_trigger": true },
{ "query": "Trial users who never activated - can we run an ad sequence to bring them back?", "should_trigger": true },
{ "query": "How many creatives do I need per retargeting stage?", "should_trigger": true },
{ "query": "Design retargeting for a 90-day B2B evaluation with a buying committee involved", "should_trigger": true },
{ "query": "People are complaining our ads follow them everywhere - fix our retargeting", "should_trigger": true },
{ "query": "What events should define my deepest remarketing tier - pricing views or trial starts?", "should_trigger": true },
{ "query": "Is it worth splitting cart abandoners from checkout abandoners into different audiences?", "should_trigger": true },
{ "query": "How do I win back churned subscribers with paid ads?", "should_trigger": true },
{ "query": "Our remarketing keeps showing ads for out-of-stock products", "should_trigger": true },
{ "query": "Which visitors deserve the longest ad follow-up window - blog readers or almost-buyers?", "should_trigger": true },
{ "query": "I have funnel drop-off data from analytics - turn it into a retargeting plan", "should_trigger": true },
{ "query": "What frequency caps should each retargeting stage get?", "should_trigger": true },
{ "query": "Do we need separate campaigns for people who watched the demo video vs visited the site?", "should_trigger": true },
{ "query": "Build cold audience tiers for our new fitness app - interests, lookalikes, broad", "should_trigger": false },
{ "query": "How do I map our ICP onto Meta ad audiences for prospecting?", "should_trigger": false },
{ "query": "My interest-stack prospecting campaigns overlap each other - fix my cold audience structure", "should_trigger": false },
{ "query": "Which customers should seed our lookalike audience?", "should_trigger": false },
{ "query": "Our lookalike audience is too small - the seed list only matched 400 people", "should_trigger": false },
{ "query": "What match rate is normal for a customer list upload?", "should_trigger": false },
{ "query": "My Meta pixel isn't firing on the purchase page", "should_trigger": false },
{ "query": "Conversions stopped showing up in Google Ads yesterday - is the tag broken?", "should_trigger": false },
{ "query": "Pre-launch check: is our conversion tracking set up correctly?", "should_trigger": false },
{ "query": "Meta says 300 conversions and our CRM shows 120 - why don't the numbers match?", "should_trigger": false },
{ "query": "Platform-reported revenue is double what Shopify reports - is that normal?", "should_trigger": false },
{ "query": "Our prospecting ads' CTR has been sliding for three weeks - is the creative worn out?", "should_trigger": false },
{ "query": "When should I refresh creative on a cold traffic campaign?", "should_trigger": false },
{ "query": "Write me 10 headline variants for our spring sale ads", "should_trigger": false },
{ "query": "Our ads all sound the same - I need fresh copy angles", "should_trigger": false },
{ "query": "Draft a creative brief our designer can execute for the Q4 campaign", "should_trigger": false },
{ "query": "Script a 30-second UGC ad for our skincare serum", "should_trigger": false },
{ "query": "Rank these five video hooks and tell me which deserve test budget", "should_trigger": false },
{ "query": "Design an A/B test for two ad concepts - budget and sample size", "should_trigger": false },
{ "query": "How should I split $80k a month across Google, Meta, and TikTok?", "should_trigger": false },
{ "query": "What share of our ad budget should go to prospecting vs retargeting?", "should_trigger": false },
{ "query": "Our campaign is profitable - how fast can we scale the budget?", "should_trigger": false },
{ "query": "Are we on pace to spend our $30k monthly ad budget or under-pacing?", "should_trigger": false },
{ "query": "What's the maximum CAC we can afford before ads stop making sense?", "should_trigger": false },
{ "query": "Is 3.2x ROAS good for a DTC apparel brand?", "should_trigger": false },
{ "query": "Our whole ad account's CPA doubled since March - figure out why", "should_trigger": false },
{ "query": "Audit our ad account - something's off but I don't know what", "should_trigger": false },
{ "query": "We have 40 campaigns all stuck in learning limited - should we merge them?", "should_trigger": false },
{ "query": "Should we advertise on TikTok or LinkedIn for our B2B tool?", "should_trigger": false },
{ "query": "Mine our search terms report and build negative keyword lists", "should_trigger": false },
{ "query": "Ads get clicks but the landing page doesn't convert - audit the page", "should_trigger": false },
{ "query": "Carousel or single image for our awareness campaign?", "should_trigger": false },
{ "query": "Map the buying committee for our ERP deal and how to reach each role", "should_trigger": false },
{ "query": "We want to boost our CEO's LinkedIn posts as paid ads", "should_trigger": false },
{ "query": "How do we get our product recommended inside ChatGPT's sponsored answers?", "should_trigger": false },
{ "query": "How do I break into paid media as a career?", "should_trigger": false },
{ "query": "Interview questions for hiring a performance marketer", "should_trigger": false },
{ "query": "Which PPC newsletters and podcasts should I follow to stay current?", "should_trigger": false },
{ "query": "Starting a brand-new paid ads project - where do I even begin?", "should_trigger": false },
{ "query": "Target CPA or maximize conversions for our lead gen campaigns?", "should_trigger": false },
{ "query": "Set up an abandoned cart email flow in Klaviyo", "should_trigger": false },
{ "query": "Build a welcome email drip sequence for new signups", "should_trigger": false },
{ "query": "Design our SaaS onboarding funnel to reduce drop-off after signup", "should_trigger": false },
{ "query": "Improve our checkout flow UX to reduce cart abandonment", "should_trigger": false },
{ "query": "Map our full marketing funnel from awareness to purchase for the board deck", "should_trigger": false },
{ "query": "Create a sales funnel for my online course launch", "should_trigger": false },
{ "query": "How do I retarget my cold email list with a better follow-up sequence?", "should_trigger": false },
{ "query": "Analyze the signup funnel drop-off in Amplitude for our product team", "should_trigger": false },
{ "query": "Push-notification win-back campaign for dormant app users", "should_trigger": false },
{ "query": "SMS flow for customers who abandoned checkout", "should_trigger": false },
{ "query": "How long should our email nurture sequence run for MQLs?", "should_trigger": false },
{ "query": "What reach and frequency goals make sense for a TV-style brand awareness campaign?", "should_trigger": false },
{ "query": "Which keywords should we bid on for our brand search campaign?", "should_trigger": false },
{ "query": "Segment our customer base for the loyalty program tiers", "should_trigger": false },
{ "query": "Retention strategy to reduce churn for our subscription box", "should_trigger": false }
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |