The MarTech Napkin: How To Choose Your Marketing Stack In 2019

It’s hard to do a post about choosing the best marketing tools without referencing this graphic - the martech “supergraphic”.

Over 7000 logos, tools & painkillers. Whatever problem you have, whatever’s in vogue, whatever your HIPPOs want — there’s a tool for that.

But whilst they might solve individual pains, they don’t work together.

• Your team has to learn, setup, and maintain dozens of different tools
• Your budgets get spread across dozens of different tools
• Tools (and the tactics they serve) might have a limited shelf-life

But it also silos your customer data.

With the exception of pure content management systems and some project management tools, every tool in your martech stack will host or use some element of customer data.

Customer data is the lifeblood of these tools and teams. Tools are the bridge between customer data and your customer experience. Whilst individual tools might be able to solve individual pain points, they need to work with the complete context of every customer. This is the key to:

• Personalization
• Attribution
• Lead qualification
• Reporting
• Team alignment
• GDPR compliance

You can’t do any of this. All of this falls apart when customer data falls apart. This starts with how you choose marketing tools.

In the fourth part of The Complete Guide to Customer Data, we’ll show you our framework for “hiring” (not buying) tools for your sales, marketing, support & customer-facing teams so you can navigate the martech supergraphic to build a lean, productive martech stack.

Objectives:

• Understand our new framework for choosing martech
• Learn the importance of data usability & portability
• Outline the types of tools you need for your customer experience
• Review our checklists for choosing and evaluating tools

Resources:

• MarTech napkin
• 8x checklists

Lesson #1: Introducing the MarTech napkin

At Hull, we often get asked for tool recommendations. We see behind the scenes at a lot of scaling SaaS startups and have a broader view of what works (and what doesn’t). But the answer is always frustrating — “it depends”.

It depends on your:

• Company stage of maturity
• Team & talents
• Your other existing tools
• The scope of the funnel & customer experience you’re trying to create
• Your data model (B2B vs B2C; account-based vs. lead-based)

People want a straight and simple tool recommendation, but they come from different situations and bring different internal biases. Particularly as you hire new people for your team, they often arrive with a bias towards a particular toolset and ecosystem.

People need a straight and simple framework for choosing tools. Something to guard against common pitfalls (that lead to siloed data, bloated tech spend, high learning curves for team members, and so on) whilst still giving room for the internal biases and internal decision making.

Lists of tools are not helpful if you don’t know how to choose amongst them.

We also don’t agree “buying tools” is the right way to think. Most martech tools are monthly or annual subscription services. Instead of thinking of “buying” tools for a pain point, hire tools for jobs-to-be-done.

Every job-to-be-done can be defined from your customer data model, your customer experience, and how you plan to tie your data together. This is why the first three parts of The Complete Guide to Customer Data focused on the data you need. By focusing the toolset you need with the jobs-to-be-done framework, you can create a much leaner martech stack.

With this in mind, we created a back-of-the-napkin sketch to outline how to choose tools given your different biases.

Feeling you’ve seen this gag before? One of our investors Point9 Capital produce an annual SaaS Funding Napkin to show what it takes to raise money at each stage.

View the full size version.

Embed the MarTech napkin in your next blog post. Copy the code below

<p><img src="//images.ctfassets.net/nffxf5qv6v6n/1eWLODnKNeUUQuwQs6Iy4w/2434831e8c9c0c6b20447bc4b16432b8/martech-napkin-960.png"></p><p>Read more about <a href="https://www.hull.io/blog/choosing-your-marketing-stack/" target="_blank">choosing tools for your martech stack</a> and the full-size MarTech napkin in <em>The Complete Guide to Customer Data</em> on the Hull blog.</p>

Here’s how each of these jobs-to-be-done breaks down.

Lesson #2: Use your customer journey map to determine what tools you need

Tools are the bridge between your customer data and your customer experience. In Part 2: Customer Journey Mapping, Analytics & Attribution Modeling, you outlined the customer experiences and content you planned to create, the channels these experiences would occur in, and the types of tools you need.

Customer experiences are stages in your customer lifecycle along a specific customer journey path - for instance, your free trial. The content of your free trial experience may include in-app messaging and onboarding drip emails. This determines the types of tools you need.

To connect the data in your customer profiles with your experience, you need to be able to transform it into data you can use within tools. This fits within a hierarchy. In the example of creating the free trial experience we need:

1. Entities (like people and companies)
2. Associated data (like events & attributes)
3. Groups of entities (like segments, lists, views, or audiences)
4. Actions (like workflows and templates) in tools to create the experience

Entities have data associated with them based on an identifier (discussed in Part 3: Creating a “Single Source of Truth” for your customer data). In most tools, a profile will include a timeline of events and a set of user attributes.

To create experiences, you need to group people together to ‘enroll’ them in a workflow, campaign, sequence, or whatever method is involved in each tool for triggering the set of content and experiences to be delivered. This involves building segments, lists, views, or audiences (depending on the exact tool).

Profile and segment data is used to trigger workflows and populate templates in your tools to deliver that customer experience.

For every channel, your tool has to host this full hierarchy of data. The more complete this hierarchy, the better the customer experience you can provide.

Tools determine the customer journey with events and segments.

Segments control who you talk to. They group together alike entities (like every person in your free trial) for the same customer experience. Most tools enable you to group people or companies together into:

• Email lists
• CRM views
• Ad audiences

To group people together for the same customer experience, you need to track the actions that take a user into this segment — these are your Conversion Events you’ve outlined in your customer journey in Part 2.

For example, a person should enter the segment o free trial users after a Trial started event, and exit after a Trial canceled or Subscription started event (which takes them to another experience).

Conversion events (actions) are the “source of truth” for any person or companies position in the customer journey. Not all tools can support events, or integrate natively with every source of your event data (like your backend product database, or subscription billing software).

For tools that cannot ingest event data (particularly from other tools), you’ll need to derive an attribute like Lifecycle stage or In Free Trial = TRUE to determine their state in the customer journey and use that to create segments in each tool. State-based attributes derived from event data (like Trial started) have many advantages:

• Easier to store (vs. many events)
• Easier to sync between tools
• Quicker to summarize

However, you need to ensure state-based attributes are accurately updated any and every relevant event. Stateful attributes become less reliable as your customer data model and customer journey increases in complexity, but your tools can still all trigger actions in real-time. For instance, if a person has an Email unsubscribed event, all your tools should reflect this new state before they trigger the next email to send.

Events are not easily synced between multiple tools, however. The solution is to compute your segments centrally from your single source of truth, to control the customer experiences across all your tools — particularly where they are in your customer journey (their state).

Whenever an update or change is detected, this will reassign and regroup people’s experiences and sync consistently across all your tools. Instead of relying on every tool having to ingest event data and updating state, you just need one tool and a “global” state.

Checklist for segmentation features (segments, lists, views, audiences):

• Can you build segments with events?
• Can you build segments with events from all your data sources?
• If it cannot be built with events from all your sources, where can the attributes come from to determine state?
• Can you build segments with segments from other tools? (Centralized segmentation)

Tools fill-in-the-blanks in templates with your customer data

The goal is not just to determine WHO to talk to, but WHAT to say to them (more of that coming up in Part 6: Orchestrate as Personalization Strategy).

Tools often include templates to enable you to define content, and dynamically replace content based on your customer data. The simplest example of this is using a person’s first name in a template - we’ve all used Hi {contact.firstName} in emails before.

Outside of email, live chat, and other 1:1 channels where you have a known customer, you still need to be able to dynamically adjust templates. For instance, personalizing websites you need to be able to edit whole or part of an HTML element such as a <p>, <img>, <div>, or <span>.

Dynamic content insertion depends on fallback strategies (discussed in Part 3) to maximize the occurrence of personalized content being displayed. For example, company name could be sourced from a demo request form, sales rep in your CRM, data enrichment, billing tools. You need to prioritize these data sources for each person and company in your tools and sync your “truest” data source for each person’s attributes to your tools.

The best tools don’t limit dynamic content to word replacement. Though you may have all the context on a person, word replacement doesn’t feel natural — you wouldn’t talk like it in person.

Some templating engines enable more flexible logic to determine which variation of content they might see, and create a far more fluid, natural experience from all the context in your customer data. Liquid (a templating language by Shopify) is a good example of this.

At Hull, we use Customer.io (which uses Liquid templating) to personalize our email newsletter and share the templating we wrote in a gist in the P.S. (you can see them all here). Here’s an example for our newsletter promoting our Spotted: Email Personalization at scale post (quite meta..), showing how we substitute words, phrases, sentences, and paragraphs according to different customer data. It happened to come at a timely point for certain subscribers…

With highly dynamic content like this, it’s critical to be able to preview different versions for different individuals to check if combinations of logic still make sense and you don’t have any stray mark up.

Templating can also control the customer experience 2nd hand. Internal notifications (like Slack or email notifications) about your leads and customers can shape and personalize the experience given by your sales reps, support team, and account managers. Using templating in your internal notifications, you can surface context that might be missed or disregarded by your team, that shapes how they might reach and out and talk to them.

Checklist for templating features:

• Can you dynamically replace content using profile data?
• Is there a flexible templating language (beyond just word replacement)
• Is the templating language easily readable?
• Can you easily (and quickly) preview content for many individuals?

Tools trigger actions with “if this then that” workflows

Workflow tools let you design and define the conditions for trigger actions in your tools (like sending emails) with “if this then that” logic, time delays, if/else branching logic and more. They are usually visual, point-and-click tools which have the benefit of being easy to use.

Workflow tools bring a number of challenges, however. Like attributes in segments, they hold a state. As your customer journey becomes more complex, often you need more if/else logic to precisely define which experience a customer should receive. Workflows also tend to become longer to trigger more content and experiences.

Using workflows to determine state — where someone is in your customer journey and what experience they should receive — can become very difficult to manage. It is also less reliable at scale since each workflow will have multiple if/else decision trees to determine a state.

By comparison, both segments and templates are computed once given all known information. This makes them simpler to create, manage, and maintain. They determine the state in a single computation, not a series of computations.

Workflow tools are best used once the state is determined (centrally) to trigger a series of actions that make up an experience. It is much more important to have powerful and flexible segmentation and templating to create personalized, multi-channel customer experiences at scale than it is to have a complex workflow tool. Keep your workflows simple, and build the complexity into your segments and templates.

Checklist for workflow features:

• Is there a visual, point-and-click workflow builder?
• Can you trigger workflows using any event, attribute, or segment?
• Is there a powerful segmentation and templating engine as well? (Remember, control your complexity)

Data usability determines the quality of your customer experience

It’s not hard to find marketing tools in each category. Leaders of quadrants, five-star reviews, and so on. With today’s martech scene, it can be challenging to differentiate between tools.

Data usability (segments, templates & workflows) determines the quality of your customer experience, and can often be a significant differentiator between tools. For instance, email tools support different capabilities to build segments, ingest events and event properties, and host unlimited attributes. Advanced workflow tools are common place, but advanced templating tools are not.

Some popular tools hold many limitations in their data usability. For instance, Mailchimp has a limited number of attributes you can sync. Despite having a segmentation tool, merge tags and templating, and automated workflows, the lack of attributes limits the context you can bring on each email subscriber which limits your customer experience.

Lesson #3: Choose tools to host your customer profiles for each channel

Now you know how you want to use data within your tools, you need to understand how it flows.

Customer data flows in loops. Your content, messaging, and experience is based on data in your profiles. How a customer, lead, or website visitor (or anyone) reacts to that gets tracked back into a profile. This must always form a closed loop.

Profiling is to make sure you maintain the full context on every person and company that has ever interacted with your brand or product, and make that data available to create customer experiences in each channel.

For every channel you’re operating in, you need to choose a tool that can profile every person and company, as well as message there. This only works if you can track everything — not missing any of the context — so you can react and personalize your experience.

Checklist for choosing profile tools for each channel:

• Which channels do you need a tool (and profiling) for?
• Which tools are you evaluating that match the segmentation, templating & workflows criteria?
• Do these tools track all behavior in that channel and show it in a profile?
• Can you trigger and personalize your content for that channel?

Data portability determines your overall customer experience

Tools are rarely best-in-class at multiple channels. This means you need multiple tools, which means profiling the same people and companies across multiple tools. To build the complete context, you need to propagate changes and updates to all your other tools. You need a data integration strategy.

As discussed in Part 3, one of these tools which host your profiles needs to become your single source of truth for everything else. Rather than tying everything to everything else, make sure you resolve identities, cleanse your data, define fallback strategies, and compute your golden customer record in one place, then sync updates to everywhere else.

All your profile tools need to be able to maintain the full context from their channel, as well as every other channel you operate in. This is easier in its “rawest” form instead of a derived attribute.

• Profile people and companies (and correctly associate them together)
• Include the complete customer data model as attributes — the minimum viable context to identify a person or company
• Include all events to pinpoint each person and company in the customer journey (without derived, stateful data)
• Include segments, computed in your source of truth (vs. reproducing the same segments in each of your tools)

How you integrate profiles around your single source of truth will be discussed at length in Part 5: Customer data integration best practices

*One of the clearest markers data usability and portability in a tool is their API documentation. *

Even if you’re not technical and are not going to build something yourself, understanding a little about their data structure and how they expose indicates the usability and interoperability of the tool. What is publicly documented will determine the possibilities with integrations, data imports and exports, and the ability to single source of truth with your data here.

RESTful APIs that are both read APIs (send data from the tool) and write APIs (send data to the tool), cover all known data objects in the hierarchy (entities like people & companies, associated data like events & attributes, and segments), and are clearly documented.

Checklist for evaluating a tools portability

• Is there a documented API?
• Is there a read API to fetch data from the tool
• Is there a write API to create & update data in the tool (e.g. POST, PUT)
• Does the write API include DELETE to remove data? (Think ease of GDPR compliance)
• Does the API show all the data objects? Entities (people & companies), associated data (events & attributes), and segments (or equivalents)?

Lesson #4: Use your ideal customer profile & customer journey map to determine other tools you need

You need tools besides those to track profiles and send messages to build your customer data model and create your customer journey.

1. Data enrichment tools can identify your ideal customer profile
2. Tracking tools can profile people & companies in channels where you can’t proactively message 1:1
3. Publish segmented variations to broadcast channels (like ads and your website)

Just as with profiles, the data in these tools flows in loops.

Data enrichment tools

Data enrichment tools, services, and scraping can be used to complete your customer data model and identify your ideal customer profiles (as discussed in Part 1), target people & companies in campaigns, and to personalize messaging.

Just as with profiles, data enrichment flows in a loop. You request data around an identifier like an email, domain, or ip and additional attributes are sent back.

These attributes need to map across all your profiles in all your channels to maintain the same, shared context on each customer. Usually, this happens through your data enrichment providers direct integrations. These may include your key tools, but to maintain a shared context across all your tools you may need to “forward” this data onwards from your single source of truth.

Data enrichment may supplement data from other sources. For instance, Job title could come from data enrichment, or form fill, or your sales reps direct input - you need fallback strategies in your single source of truth to determine which data source to use for each person.

Checklist for data enrichment tools:

• What identifiers can you sync to be enriched?
• Which tools does your data enrichment provider directly enrich with?
• How can you map this data into your single source of truth, and out to all your tools?
• Where are you going to determine which version of the same data (like Job title) you’re going to use for each person and company?

Tracking tools

Tracking tools enable you to gather and track interest in your brand and product outside of your 1:1 messaging channels, on third-party platforms & websites, and Common examples of this might include tools for:

• Website analytics
• Landing page tools (form and pop-over capture)
• Forms & surveys
• Payments & billing
• Meetings & video calls
• Third-party lead sources (like G2Crowd & Facebook Lead Ads)

These are commonly one-way inbound data flows (to your single source of truth), often supported with webhooks. Analytics tracking tools come in this category too, where the content and experience is often more “static” than 1:1 personalized. For instance, web analytics closes the loop between the content being published on your website.

Since these tools don’t necessarily share the same profile as another tool, they don’t always have a stable, persistent identifier — this can lead to duplicate profiles in your end tools or data that is never merged together. For example, web sessions preceding a form submission, or matching one email address on a form or meeting with the email address of the same person elsewhere.

To help minimize this, you need to be able to pass through additional identifiers and context to aid identity resolution. For example, pairing email with an analytics tracking id means web sessions can be associated with other known data already, or the ip address which can be used to look-up the company name. These are often passed through with hidden fields in forms.

Your tracked data (including multiple identifiers) needs to flow into your single source of truth and resolve into the correct person and company-level profiles, for high-volume data like web analytics, low volume billing subscription events, and everything in between.

Checklist for tracking tools:

• What channels, platforms, websites, and actions are not tracked
• What identifiers can be captured?
• Can you capture multiple identifiers at once to associate data together? How?
• How can this data and multiple identifiers be tracked into your single source of truth?

Publishing audiences, notifications & reporting

Finally, if you can’t directly control the experience of an individual person or company, you can still create a generalized experience for similar people using audiences. For instance, personalized ads or a website are often limited because you aren’t able to identify the person who will see the content. However, by using similar attributes you can group similar people together to experience something more targeted.

These are one-way outbound data flows too — from your single source of truth. To close the loop, you need tracking and attribution through other tools. For example, with ads and web personalization, your website analytics can record clicks and the variations of the experience being viewed.

Internal notifications (like email or Slack) and reporting are one-way data flows too. They aggregate and summarize sets of data from your single source of truth, and publish outbound.

Checklist for outbound data flows

• What channels can you not send 1:1 messages to, but you can still personalize?
• Can you adjust templating based on segments, attributes, or events?
• What internal channels need notifications?

Lesson #5: Jobs-to-be-done change with company stage

The channels, tools, and data needs change as your company matures. Many of these requirements will seem totally overkill for earlier stage companies.

At Hull, we work mostly with SaaS companies. They share a common stack, common data problems, and a similar lifecycle:

1. Pre-Product/Market fit
2. Post-Product/Market fit
3. Scaling up

Here’s how the jobs-to-be-done change over time.

Best martech stack for pre-Product/market fit teams

Companies before product-market fit don’t need to prioritize acquisition, but they do need to understand their customer data.

Resources might be minimal. At least one of the founding team might be taking the lead on customer development, using any budget (what budget?), and managing your tools.

At this stage, you might have signups from one-off efforts like a ProductHunt launch, funding or launch press release, and manual outreach, but not a sustained funnel. However, profiling people and companies at this stage is incredibly important to identify your ideal customer profile.

You still need a place to capture your customer data including:

• Contact details
• An identifier (like email and domain)
• All known data about them
• Your qualitative survey data in one place

Your data needs to be in a state to easily upload to other tools as a CSV, have clean & valid identifiers, and all your other data to classify people and companies together — an embryonic CRM.

Your customer data needs to be easily comparable at this stage. This is about spotting nuanced trends amongst your data, which is much easier with data side-by-side.

Crucially, this data must actually exist as data — not just in your founder’s brains :)

With all these requirements, it’s hard to argue against the value of a spreadsheet tool like Google Sheets as your source of truth, and simple email contacts tool to manage your tools. Even with a simple sheets add-on to organise it all.

Best martech stack for post-Product/market fit teams

Pre-product market fit is about figuring out who to sell to. Post-product market fit is about figuring out how to sell to them, whilst refining your ICP. Acquisition is the priority, tweaked by reviewing the types of customers who churn.

At this stage, you’ll bring in your first dedicated sales and marketing managers who will manage both the early team and the tools. There’s a budget for the basics, like a simple website, email tool, and so on.

At this stage, everything feels like an untapped opportunity.

The mistake at this stage is to make up for the lack of budget and resources with a ton of different tools to address all the channels and opportunities vs. getting over the initial hump to make each channel work:

• Breaking onto the first page of Google for a relevant, reasonable-volume term
• Getting your first PPC ad to bring in more than it costs
• Establishing referrals and partner traffic which converts
• Building your email list and social following to a level where it drives noticeable traffic

A frankenstack of deceptively simple, free(ish) tools creates chaos for tracking, organizing and personalizing your customer experience. This can radically slow down your team’s progress. There are two key tools to establish for each functional role.

For sales, your reps need a reliable tool to work out of which enables them to be productive, close deals, and your customer data to be used (segments, templates & workflows). At this stage, advanced data integration and reporting features come secondary to your reps productivity - you don’t need heavy overhead before you can get going.

For marketing, you need a tool to gather your master list of leads and subscribers, as well as engage in a channel like an email. You need this to integrate to (or be the same as) your sales rep’s tool, and ensure both teams are working off the same data and context from the off.

Marketing should not hire multiple tools and engage multiple channels whilst resources (team, budget, time) are still stretched thin. How your team uses tools (and whether they can dedicate time and focus to use them properly) has dramatically more impact than adding additional tools.

By limiting yourself to fewer channels, your data flows will also be much simpler and you’ll have a more complete context on each of your leads and customers.

Here’s an example what this might look like:

At this stage, integrations between tools can be pairwise (everything connects to everything else), although this will become an exponential problem later on until you establish a single source of truth.

Best martech stack for scaling SaaS teams

When teams are ready to accelerate growth, the resources applied to go-to-market activities before much more freely open.

Management is often divided here - leadership for the managing of people often separates from the management of tools and data. Ownership of the martech stack moves to the operations layer. Some teams run operations within functional units (sales operations, marketing operations), but the fastest growing teams we see centralize their tools and data ops across their entire organization — more in Part 7: Customer Data Management, Operations & Governance.

To grow faster, teams explore and max out new channels. This involves hiring additional tools too, which makes pairwise (everything connecting to everything else) data integration significantly more challenging. The number of and types of events will dramatically increase, and the logic to understand the “state” of your leads, customers, and accounts becomes significantly harder without a single source of truth and golden customer record.

To understand how these different channels contribute to the revenue goal, challenges like multi-touch attribution become common - synthesizing event data from multiple different sources. As teams try to max out each channel, advanced segmentation, dynamic content, and personalization become common too. All these require significantly different setup to your data flows and operations.

At this stage, your early stage “source of truth” (your sales-rep friendly CRM, or simple marketing automation tool) can’t keep up with the full context of your customer journey and answer questions about attribution, reporting, personalizing messaging, or aligning your teams. Where there are data silos from tools, silos amongst team will follow.

The big mistake at this stage is individual functional teams take individual ownership of the whole company’s data problem, beyond just their own channel-specific needs:

• Sales “upgrades” to an enterprise CRM (expected of their new sales leadership)
• Marketing finds a ‘bigger’ marketing automation platform
• Engineering starts a data warehousing project
• Favors are called in to create more pairwise integrations around each teams choice tools

This will never address the problem of data silos. New tools for new channels will always be needed. The sheer number of tools that need to be integrated together becomes unmanageable for native, pairwise integrations.

Best practice sees teams acknowledge and address this issue early on. Each functional team needs the data in their tools, but the source of truth for all these key tools is a centralized, company level problem. This becomes an operations problem.

In Part 3, we discussed four common types of tools that act as the “single source of truth” for your customer data:

1. CRMs
2. Marketing automation platforms
3. Customer data platforms
4. Data warehouses

If you’re scaling up, you need to be deliberate in what and how you choose your single source of truth. You need to combine data from all your tools, tracking, and databases, then sync to your end tools in real-time.

In the same way as you choose best-in-class tools to max out each channel, you need to choose the best-in-class tools to act as your single source of truth and enable all your tools, teams, and data to work as one. CRMs, marketing automation platforms, and data warehouses are all compromised in some way - an incomplete view of data, lack of control over transforming data, being a one-way data flow, lacking real-time computation or more. However, customer data platforms are purpose-built for this problem.

Without a clear, deliberate strategy to build a truly unified profile, your tools will never be able to act on the full context of each customer. Some teams build workarounds by synthesizing and abstracting data into attributes that can be synced between tools, but this becomes unstructured data (tag soup) that is difficult to read, structure, and organize. You still need a tool to create and update attributes like this.

Find your spot in the martech napkin

Long lists of similar tools aren’t useful if you can’t differentiate between them. In this guide, we’ve explored the importance of data usability and portability to make your tools, teams, and data work together, and how this can be a key differentiator between otherwise similar tools.

You’ve reviewed several checklists to compare tools, and how data can flow (in loops) between them.

• Segmentation
• Templating
• Workflows
• Profiles for each tools
• Portability of data & API quality
• Enrichment flows
• Tracking (inbound)
• Outbound flows (like audiences, notifications & reporting)

With the napkin, checklists, and frameworks around data usability and portability, you can review other people's recommendations.

Next Up: Customer data integration best practices at any scale

With a framework for choosing (and pruning) your marketing tools, and understanding what data you need (from your ideal customer profile, customer journey map, and creating a single source of truth), you need to bring these two together.

In the fifth part of The Complete Guide to Customer Data, we’ll discuss the best practices of customer data integration for teams at any scale from manual, automated, and engineered techniques and how your data integration strategy should mature as your grows. In most companies, data integration lags as the company scales. This guide will lay out what you need at each stage and why.