AI in Digital Marketing — The 2026 Channel-by-Channel Playbook

Most AI-in-digital-marketing articles list tools by category — content, analytics, automation — and leave you to figure out which one solves the channel problem you actually have. Your problem isn't "I need a content tool." It's "my SEO is plateauing, my paid CPLs are climbing, my emails are getting flagged, and my CMO wants an AI roadmap by Friday." Those are channel problems. They need a channel answer.

This page is that answer. AI organized the way an operating marketing team experiences it — by the channel you sit inside. One section each for SEO (including AEO/GEO), paid media, email, content, social, personalization and CRO, analytics and attribution, customer engagement, and the cross-channel orchestration that ties it all together.

Named tools per channel. Named failure modes per channel. Real Rule27 client workflow disclosure per channel. Three pre-built stack blueprints (SMB, Growth, Enterprise) priced from $100 to $40,000+/month. No affiliate revenue earned on any tool named — same posture as our ai-marketing-tools review.

Most AI-in-digital-marketing articles are organized backwards. They list tools by category — "content creation," "analytics," "automation" — and leave you to figure out which one solves the problem you actually have. The problem you have isn't "I need a content creation tool." It's "my SEO is plateauing, my paid CPLs are climbing, my emails are getting flagged, and my CMO wants an AI roadmap by Friday." Those are channel problems. They need a channel-by-channel answer.

This page is that answer. We've organized AI in digital marketing the way an operating marketing team experiences it — by the channel you sit inside. One H2 per channel — SEO, paid media, email, content, social, personalization and CRO, analytics and attribution, customer engagement — with the specific AI tools that win in that channel, the specific failure modes that the affiliate articles don't mention, and the real workflow Rule27 runs on client engagements when we deploy AI into that channel.

A few orientation notes before we dig in. First, this page is long because the topic is broad. We tried writing a shorter version and it produced a worse map. If you only need one channel, use the table of contents and jump. Second, every tool named in the body is named again at the bottom in the schema so AI Overviews and ChatGPT can cite it cleanly. Third, we don't take affiliate revenue from any tool on this page — same posture as our ai-marketing-tools review. The point is to get you to the right answer for your channel, not to monetize the clickthrough.

How AI is reshaping digital marketing in 2026

Three shifts have rearranged the function since 2023, and any AI strategy that ignores them is solving last year's problem.

The first shift is that content creation has been commoditized. ChatGPT, Claude, Jasper, and a dozen wrappers around them produce competent first drafts faster than any human team can. The teams winning in 2026 didn't win because they have better AI — they won because they treated the commoditization as a feature, not a threat. Faster drafts mean more iteration, more testing, more channels covered. Slower drafts mean you're paying agency rates for work AI does for $20 a month.

The second shift is that search itself has split. Classic Google still drives the most traffic, but AI Overviews, ChatGPT Search, Perplexity, Claude, Gemini, and the answer-engine layer collectively handle roughly 30 percent of the high-intent queries Rule27 tracks across client accounts — up from under 5 percent in early 2024. That isn't a tweak inside the SEO channel; it's a new channel sitting between SEO and brand search. We treat AEO/GEO as a peer to SEO in the section below, not a sub-section, because that's how the work actually decomposes when you operate it.

The third shift is that attribution has gotten harder, then easier. Cookie deprecation, iOS privacy changes, and the AI-Overview-clickless query made classic attribution models progressively unreliable through 2024 and 2025. The 2026 fix isn't a better last-click model; it's AI-driven multi-touch attribution that incorporates signal from sources GA4 alone can't see — CallRail, intent data platforms, server-side conversion APIs, AI Overview citation logs. The teams that rebuilt their attribution stack in 2025 are now operating with more visibility than they had in 2022. The teams that didn't are flying blind and don't know it yet.

Keep those three shifts in mind as we go channel by channel. Every recommendation below is calibrated to the 2026 state of play, not 2023's.

AI in SEO (including AEO and GEO)

SEO is the channel where AI tooling matured first, which means it's also the channel with the most snake oil. There are four real categories of AI-SEO work in 2026: on-page optimization, content briefing and drafting, technical SEO automation, and the new AEO/GEO citation layer. Each needs its own tools and its own failure-mode awareness.

On-page optimization. Surfer SEO is the category leader at SMB-and-mid-market pricing ($89-$129/month). It scores your draft against the top-ranking SERP results and tells you which entities, questions, and structural patterns the winners use. For commercial-intent keywords in established categories, Surfer's score-to-rank correlation is real and measurable. NeuronWriter delivers roughly 85 percent of Surfer's value at half the price, which is the right call for freelance writers and price-sensitive SMB engagements. Clearscope is the enterprise alternative ($189-$399/month) — less aggressive optimization, more editorial-quality output, brand-safe scoring that survives enterprise legal review.

Where all three fail: low-competition long-tail keywords with thin SERPs. When the top-ranking pages are themselves weak, the optimization layer pushes you to match a weak baseline. Surfer in particular over-reports search volume on long-tail terms — we've seen "720 monthly searches" turn into zero impressions in Search Console after three months at position 3. Treat the volume number as directional, never authoritative.

Content briefs and drafts. Frase generates defensible content briefs faster than any human can — research questions, header structure, entity list, FAQ pulled from SERP analysis. Use Frase for the brief, then write the actual draft in ChatGPT ($20/month) or Claude ($20/month). The two-LLM stack is what senior content teams run in 2026; Claude wins on long-form editorial coherence, ChatGPT wins on speed and ecosystem features. MarketMuse ($399+/month) is the enterprise-grade topical authority modeler — right answer for $50M+ revenue businesses with 500+ pages of existing content, overkill below that scale.

Technical SEO automation. Screaming Frog added AI-driven content analysis in 2025 that flags thin content, hallucinated stats, and AI-generated patterns Google's helpful-content classifier penalizes. Sitebulb covers similar ground with better visualization. Ahrefs and Semrush both ship AI-powered technical audits inside their existing dashboards — useful for incumbents, not a reason to switch.

The AEO/GEO citation layer. This is the 2026 category that didn't meaningfully exist 18 months ago. AthenaHQ ($99-$299/month) tracks brand mentions across ChatGPT, Perplexity, Claude, Gemini, and the broader AI assistant surface. Profound is the enterprise-grade equivalent ($500-$3,000/month) with deeper prompt research and competitive intelligence. Semrush AI tracking and Ahrefs Brand Radar bundle citation tracking into existing subscriptions for incumbents. Buy AthenaHQ when you start running a GEO program; buy Profound when you cross enterprise scale. Don't subscribe to Semrush or Ahrefs just for the AI tracking layer — the category specialists are deeper.

Where AI SEO fails: confident hallucinations in regulated verticals (medical, legal, financial advice), generic outputs that miss vertical nuance, and a recurring tendency to optimize toward whatever's already ranking instead of what should rank. None of the tools above replace senior SEO judgment. They accelerate it.

AI in paid media (PPC and social ads)

Paid media is the channel where AI has the longest commercial deployment history and the highest ratio of vendor claim to operator skepticism. The platforms have been calling their bid management "AI" since 2018; what's actually new in 2026 is the creative-and-audience optimization layer running on top.

Search and shopping ads. Google Ads AI — Performance Max, Demand Gen, AI Max for Search — does real work that justifies the trust at the right account scale. Performance Max combines audience, creative, and placement optimization across the full Google surface with a single budget. Above $50K/month in Google spend with clean conversion tracking, Performance Max consistently produces better blended CPA than manually-structured campaigns. Below $20K/month, the algorithm doesn't have enough conversion volume to optimize against, and Performance Max underperforms classic Search and Shopping. The threshold matters.

Social ads. Meta Advantage+ covers the same ground inside Meta's ecosystem — audience expansion, creative variation, placement optimization. The lift is real for ecommerce and B2C lead generation. For B2B lead generation, Advantage+ is more uneven; the lookalike audiences trend lower-quality than custom-built lookalikes from a clean CRM seed list.

Cross-platform creative and bid automation. Smartly.io is the enterprise pick for cross-platform paid media automation — Meta, TikTok, Pinterest, Snapchat creative versioning, dynamic product ads, and bid management in one platform. Pricing is enterprise-only and opaque, typically $5,000+/month. Adzooma and Revealbot cover similar ground for mid-market accounts at SMB-friendly pricing ($99-$499/month). Albert.ai is the autonomous-execution pick at enterprise scale ($10,000+/month) — it doesn't suggest changes, it makes them.

Where AI paid media fails. B2B lookalike audiences in narrow verticals — the algorithm doesn't have enough comparable accounts to find genuine lookalikes, and the expanded audience drags CPL up faster than the volume justifies. Creative fatigue invisibility — Performance Max and Advantage+ both hide which creative is winning, which makes it hard to know when to refresh. Attribution distortion — AI-optimized campaigns often appear to outperform because they get credit for conversions that would have happened anyway through brand search. Always pair AI campaigns with incremental holdout tests, never trust the platform-reported ROAS without an external attribution check.

Rule27's blunt rule on AI in paid: trust the AI on optimization within a campaign once you've validated incrementality. Never trust the AI to choose the campaign strategy. Strategy is human work.

AI in email marketing

Email is the channel with the longest history of "AI" claims and the shortest list of claims that survived measurement. The four legitimate use cases in 2026: send-time optimization, subject-line and copy testing, lifecycle automation, and deliverability triage.

Send-time and frequency optimization. Seventh Sense is the category leader specifically for HubSpot and Marketo deployments. It learns each subscriber's engagement pattern and times sends to that individual instead of a single batch hour. For lists above 50K subscribers, the open-rate lift is consistent and measurable. Below 50K, the per-subscriber model doesn't have enough data to outperform a simple A/B test on send time, and the cost overhead isn't justified.

Subject-line and copy testing. Persado and Phrasee are the enterprise category leaders for emotion-targeted copy generation. Rule27's measured experience across three engagements in 2024-2025: the lift exists but didn't justify the six-figure annual contracts. The category itself is interesting; the price-to-value at enterprise scale is not. Anyword is the more accessible alternative ($49-$349/month) with predictive performance scoring built in. For most accounts, ChatGPT or Claude generating 10 subject-line variations and a simple email-platform A/B test produces equivalent or better results at one-twentieth the cost.

Lifecycle automation. HubSpot AI inside Marketing Hub Professional ($890/month) handles content generation, lead scoring, predictive list segmentation, and automated journey orchestration. Right answer for HubSpot incumbents; not a reason to switch CRMs. Marketo Engage (now Adobe) covers similar ground inside the Adobe Experience Cloud — strong for enterprise B2B that's already invested in the Adobe stack. Klaviyo AI is the ecommerce-specific equivalent — predictive analytics, product recommendations, and automated flow optimization tuned for DTC and ecommerce send patterns.

Deliverability triage. Email AI tools that promise "better deliverability" are mostly selling list hygiene with a confident voice. The genuine work is unglamorous — IP warmup, authentication (SPF/DKIM/DMARC), engagement segmentation, hard-bounce suppression. AI doesn't fix bad deliverability. Process does. Tools that claim otherwise are reselling 2018 list-hygiene services with an AI wrapper.

Where AI email fails. Mailchimp AI — perpetually underwhelming despite four years of promises. The Content Optimizer is mid-tier; the Customer Journey AI doesn't beat HubSpot or Klaviyo. If you're on Mailchimp for the email platform, fine; the AI layer isn't the reason to stay. Generic "AI email writer" tools that wrap GPT-4 — most are wrappers with worse UI than ChatGPT directly. Skip unless they integrate with your CRM in a way ChatGPT doesn't.

AI in content marketing

Content is the channel that the most articles cover and the channel that's hardest to do well, because the failure mode is invisible until the audience leaves. Three sub-channels: long-form, repurposing, and research.

Long-form drafts. Claude ($20/month Pro, $100-$200/month Max) is the editorial pick for any piece above 2,000 words. The output reads as having read books — paragraphs flow, transitions land, conclusions arrive without the "in conclusion" tic. The 1M-token context window lets you feed the entire site, brand guide, and competitive analysis in a single prompt, which dramatically improves on-brand consistency. ChatGPT ($20/month Plus) wins on speed and ecosystem; most senior content teams run both. Jasper ($69-$125/month and up) layers brand-voice training on top — earn your way in once you produce 50+ pieces a month and need cross-writer voice consistency.

Briefs and research. Frase ($15-$115/month) at the brief stage. Perplexity ($20/month Pro) for research that needs citations — substantially better than Google for any query where you want the answer plus the sources behind it. BuzzSumo for trend research and competitive content analysis. MarketMuse for topical authority modeling at enterprise scale.

Repurposing. Opus Clip for long-form video into short-form clips with auto-captions and B-roll. Descript for podcast and video editing with text-based timeline. Lately was the category leader pre-acquisition; quality declined post-2023 and most teams now replicate the value with custom ChatGPT or Claude GPTs at one-fortieth the cost.

Where AI content fails. Niche B2B accuracy — Jasper, ChatGPT, and Claude all hallucinate confident, plausible, completely wrong technical claims in narrow verticals. Brand-voice drift — the longer the piece, the more the AI defaults toward its trained-mean voice and away from your brand. Fact hallucination at scale — even with citations enabled, the citations themselves sometimes hallucinate (the URL exists but the claim isn't actually on the page). Rule27's internal rule, no exceptions: every AI-generated piece goes through a human editor before publication. Not because the tools aren't good — many are excellent — but because the failure modes are unpredictable enough that human review is cheap insurance against expensive mistakes.

AI in social media

Social is the channel where AI most often produces output that's worse than not posting, because AI-slop captions and AI-generated hero images both signal lazy operation to followers who can spot the pattern. Used well, AI is a force multiplier for social. Used badly, it's reputation damage in slow motion.

Posting, scheduling, and listening. Sprout Social ($249-$499/seat/month) and Hootsuite ($99-$249/month) both ship AI features on top of established scheduling platforms — Sprout's social listening AI and Hootsuite's OwlyWriter for caption generation. For brand-managed accounts at mid-market and above, both pay for themselves on the scheduling and listening side; the caption AI is mid-tier and best treated as a starting point. BuzzSumo and Brandwatch cover deeper social listening and sentiment analysis. Sprinklr AI is the enterprise pick — full customer experience management with native AI across listening, publishing, and care.

Captions and hashtags. Flick AI and Copy.ai for high-volume short-form variations. Useful for performance social teams running A/B tests across 50+ creative variants per campaign. Less useful for organic brand social where consistency of voice matters more than volume of variation.

Video. Runway for short-form generative video. Opus Clip for long-to-short repurposing. HeyGen for AI avatar video — useful for explainer and onboarding content, still uncanny-valley for brand campaigns. Descript for podcast and video editing.

Where AI social fails. Generic AI captions — followers can detect the pattern within three weeks, and engagement falls visibly. AI hero imagery on organic brand posts — same critique. Over-automated posting at the expense of brand voice — the agencies that run "AI social management" for $500/month tend to produce visible-to-the-audience slop within two months. Social is the channel where the human-in-the-loop requirement is most rigid.

AI in personalization and CRO

Personalization is the channel where AI's promise is largest and the SMB execution gap is widest. The enterprise tools deliver real lift; the SMB attempts to replicate them mostly don't.

Site personalization. Mutiny ($5,000-$15,000/month) is the B2B personalization category leader — different landing pages to different visitor segments based on firmographic and behavioral data. Case studies from Snowflake, Segment, and Notion show 30-70 percent conversion lift on personalized variants. Right answer for enterprise B2B with $100K+ deal sizes. Dynamic Yield (now Mastercard) covers ecommerce personalization at enterprise scale. Adobe Target is the Adobe Experience Cloud option.

A/B testing and optimization. VWO and Optimizely both layer AI on top of classic A/B testing — automatic variant generation, statistical-significance acceleration, multi-armed bandit allocation. Real productivity gains for teams running 5+ tests a month; overkill for teams running 1 test a quarter.

Email and SMS personalization. Klaviyo, Bloomreach, and Salesforce Marketing Cloud all ship predictive personalization tuned for retention and lifecycle. Adobe Experience Platform is the enterprise CDP-plus-personalization layer — consolidates data across systems and feeds it into AI-driven journey orchestration.

Where personalization fails. Overfit segments — too-narrow personalization rules produce variants nobody sees enough times to learn from. Novelty fatigue — personalization that worked at launch decays over 6-9 months as visitors normalize to the personalized experience. Regulated industries — healthcare, financial services, and legal verticals have personalization constraints that most AI tools don't model out of the box. Always pair AI personalization with explicit guardrails.

AI in marketing analytics and attribution

Attribution is the channel that pays for the rest of the stack. If you can't measure which AI investments earn back, you can't justify the budget for the next round. Three sub-channels: multi-touch attribution, predictive analytics, and AI-assisted reporting.

Multi-touch attribution. Cometly ($399-$799/month) is the mid-market category leader — cross-platform attribution across Meta, Google, TikTok, and LinkedIn with a unified revenue model. Strong pick for performance teams above $20K/month in paid spend. Below that threshold, attribution overhead exceeds the visibility lift. HubSpot AI handles attribution natively for HubSpot CRM incumbents. Triple Whale is the ecommerce-specific attribution pick (DTC and Shopify-heavy stacks).

Predictive analytics. Adobe Experience Platform and Salesforce Einstein are the enterprise predictive layers — customer lifetime value, churn prediction, conversion probability, next-best-action recommendations. Real value at $50M+ revenue with multi-million-row datasets. Below that scale, the cost-to-value math doesn't work and a Google Analytics 4 plus BigQuery setup with custom modeling delivers most of the same insight.

AI-assisted reporting. Looker Studio with Gemini integration (Google) lets non-analysts query dashboards conversationally. Power BI with Copilot covers the Microsoft equivalent. The category is genuinely useful — junior marketers can now answer ad-hoc data questions without an analyst's time — but the outputs need senior review, because the LLM will confidently produce charts that misinterpret the underlying query.

Where AI attribution fails. Cookie deprecation distortion — Safari and iOS privacy changes leave significant gaps that AI models smooth over without flagging the uncertainty. Over-confident causal claims — multi-touch attribution shows correlation; AI summaries often present it as causation, which is wrong and dangerous for budget decisions. Always require a holdout test before reallocating significant budget based on attribution-only signal.

AI in customer engagement and chat

Chat is the channel between marketing and sales. Done well, it qualifies inbound, books meetings, and removes tier-1 support load. Done badly, it's a worse version of a contact form.

B2B inbound. Drift ($2,500+/month) is the conversational marketing category leader. AI qualifies inbound visitors, routes to the right sales rep, and books meetings without manual triage. Strong fit for B2B teams with consistent inbound demand and a sales team that can convert conversational leads. Qualified is the Salesforce-native alternative — same use case, tighter integration with Salesforce CRM.

Support. Intercom Fin handles tier-1 support questions autonomously with a fine-tuned LLM. Resolution rates of 50-70 percent are typical, which removes meaningful load from human support teams. Per-resolution pricing — model it carefully; high-volume months produce four-figure bills. Ada and Zendesk AI cover the same category with different pricing models.

SMB chat. Tidio ($29-$499/month) covers 80 percent of Drift's use cases at one-tenth the cost. Lyro AI Agent handles qualification and booking for SMB ecommerce and service businesses. Right tool for the segment.

Where chat AI fails. Per-resolution pricing surprise — high-volume support months produce bills nobody modeled. Cultural mismatch on tone — chat AI defaults to neutral-helpful, which works for SaaS but jars on luxury brands, professional services, and any vertical where the brand voice is distinctive. Always train the AI on actual brand transcripts, not just product documentation.

How Rule27 actually uses AI on client engagements

We're a marketing agency. We don't sell tools; we use them on client work. Here's the production reality channel by channel.

SEO. Every client engagement runs Surfer or NeuronWriter for on-page optimization, Frase for briefs, ChatGPT plus Claude for drafts (always human-edited before publication), and AthenaHQ for AEO/GEO citation tracking once we're producing content that earns AI Overview citations. We publish citation logs for every client — which prompts trigger our citations, which trigger competitors', and which queries we're targeting next quarter.

Paid media. Performance Max for Google clients above the $20K/month spend threshold; classic structured campaigns below. Meta Advantage+ for ecommerce; custom-built lookalikes for B2B lead gen where Advantage+ underperforms. Always paired with incremental holdout tests, never trusting platform-reported ROAS without external validation. Cometly for cross-platform attribution at every paid engagement above $25K/month total spend.

Email. Seventh Sense for HubSpot and Marketo clients above 50K list size. ChatGPT for subject-line and copy variations (with platform A/B testing for selection, not vendor-AI ranking). Klaviyo AI for ecommerce flows. No Persado, no Phrasee — the measured lift hasn't justified the cost in our engagements.

Content. Frase for briefs, Claude for long-form drafts, ChatGPT for short-form and ideation, Jasper for clients with 50+ pieces/month and strict brand-voice requirements. Every piece edited by a human before publication.

Social. Sprout Social or Hootsuite for scheduling and listening (chosen by client preference and budget). ChatGPT for caption ideation, never for direct publication without editing. Runway and Opus Clip for video repurposing. Brandwatch for enterprise listening.

Personalization. Mutiny for enterprise B2B clients with the deal-size economics to justify it. VWO or Optimizely for mid-market A/B testing. Klaviyo native personalization for ecommerce.

Attribution. Cometly for mid-market multi-touch. HubSpot AI for HubSpot incumbents. GA4 + BigQuery + custom modeling for SMB clients below the Cometly threshold.

Chat. Drift for B2B inbound at mid-market and above. Intercom Fin for SaaS support. Tidio for SMB ecommerce.

The pattern across every channel is the same: AI accelerates execution; human judgment determines strategy; every output gets reviewed before it ships. That's the agency posture in 2026. It's not new; the discipline is the differentiator.

Three-tier AI digital marketing stack blueprint

The right combination of tools depends on revenue, team size, and channel mix. Three pre-built blueprints we use as starting points for client engagements.

SMB stack — $100 to $300/month total. ChatGPT Plus ($20). Claude Pro ($20). NeuronWriter Bronze ($23). Frase Solo ($15). Canva Pro ($15). Tidio Starter ($29). Zapier Starter ($20). Optional: Anyword Lite ($49). Total: $100-$200/month at the floor. What it does: content, SEO, basic chat, visual, automation. What it doesn't do: GEO tracking, attribution, personalization (all overkill at SMB scale).

Growth stack — $500 to $2,000/month total. All of SMB plus Surfer Essential ($89), Jasper Teams ($125), Midjourney ($30), Runway Standard ($15), AthenaHQ Starter ($99), Cometly Starter ($399 if paid spend justifies), Drift Premium ($2,500 if B2B inbound justifies). Total: $500/month at the floor; $3,500+/month with paid Drift and Cometly. What it does: cross-channel AI across SEO, paid, content, social, GEO tracking, attribution, conversational sales. What it doesn't do: enterprise personalization, autonomous campaign optimization.

Enterprise stack — $5,000 to $40,000+/month total. All of Growth plus HubSpot Enterprise ($3,600), Marketo Engage (custom), Clearscope Business ($399), MarketMuse Team ($399), Profound ($500-$3,000), Mutiny ($5,000-$15,000), Smartly.io (custom), Albert.ai ($10,000+), Adobe Experience Platform (custom). Total: $10,000-$40,000+/month depending on team size and ad spend. What it does: everything across every channel at enterprise quality. What it doesn't do: run cheap.

The most common stack-tier mistake is buying enterprise tools at sub-enterprise scale. HubSpot Enterprise at $50M revenue earns its cost; HubSpot Enterprise at $5M revenue is paying for capacity you can't use. Match the stack to the operation.

What AI in digital marketing won't do (and never will)

Three things to be clear about, because the vendor marketing rarely is.

AI won't choose your strategy. Every tool above optimizes execution within a strategy somebody else picked. Channel mix, brand positioning, market choice, segment targeting — that's senior marketing judgment. The teams that delegate strategy to the AI watch their differentiation erode in 18 months.

AI won't give you brand judgment. Voice, taste, story, narrative arc — these are still human-shaped. The AI can imitate brand voice once trained, but it can't decide what the brand should sound like in the first place. Brand work survives the AI shift; brand work gets harder because it has to be more deliberate.

AI won't be accountable for results. When the campaign underperforms, the AI doesn't explain why to the CMO. The marketer does. Accountability is a human function and always will be. The marketers who use AI as a multiplier compound results. The marketers who treat their job as the tasks AI now does get replaced — by the AI-fluent marketers who absorbed those tasks and moved up the stack.

If you've read this far, you're past the discovery stage and into the planning stage. Two next steps depending on where you are.

If you want a channel-by-channel audit of your current AI stack: Book a 30-minute audit via the form below. We'll review your stack against the channel framework on this page and tell you honestly where you're overpaying, where you have channel gaps, and where AI would produce the highest leverage on your specific operation.

If you want to see the underlying matrix: Download the 2026 AI Digital Marketing Stack Map PDF — the same channel × tool × tier spreadsheet we use to design client stacks. Print it, score your operation against it, and you'll see where to invest next quarter.